By John McHugh, Equity Empire Small Cap Research

Most investors haven’t heard of Boost Run…yet.

But they will soon.

Boost Run is the Nvidia-certified GPU cloud quietly powering the frontier labs that are building tomorrow’s most important AI applications.

Yesterday after the market closed, the company filed an 8-K that stopped me in my tracks.

On May 21, 2026 - - just ten days after Boost Run began trading on the NASDAQ under the ticker - - the company entered into a material service agreement with Thinking Machines Lab Inc. for high-performance managed GPU compute and cloud infrastructure services.

It’s a meaningful deal for Boost Run.

Two Order Forms executed that concurrently cover an initial 36-month term with a combined total contract value of approximately $471.7 million.

The Orders call for the deployment of 5,000 Nvidia B300 GPUs across Boost Run’s data center facilities, along with related shared network storage and CPU node services.

It wasn’t just the size of the deal, but also the contract structure that caught my attention.

Thinking Machines can only exit for cause if Boost Run materially breaches its obligations.

There’s no walk-away clause. There’s no usage-based escape hatch.

This is $471.7 million in fully committed, locked revenue.

Now let me explain why the customer matters just as much as the contract.

The Nvidia Connection

Thinking Machines Lab was founded by Mira Murati - - the former Chief Technology Officer of OpenAI and the tech visionary who oversaw the development of ChatGPT, Sora, and the model releases that turned OpenAI into a household name.

Since striking out on her own, Murati has built Thinking Machines into one of the best-capitalized AI research newcomers on the planet, raising more than $2 billion at a valuation above $12 billion.

Now here’s where it connects directly back to the Nvidia thesis I’ve been building since I first wrote about Nebius in December 2024.

Nvidia has a pattern:

When the chipmaker makes a strategic investment in a company inside its ecosystem, that company’s shareholders tend to get very rich.

We saw it play out in real time with Nebius.

Nvidia invested, Nebius earned Exemplar certification, the hyperscaler contracts followed, and the stock ran from $31 to nearly $230.

The whole thing happened faster than most investors could react.

Earlier this year, Nvidia made another significant strategic investment.

This time they invested in Thinking Machines Lab, striking a multiyear partnership under which the startup will deploy at least one gigawatt of Nvidia’s next-generation Vera Rubin systems.

The logic is the same as it was with Nebius:

Nvidia deepens relationships with the labs most likely to adopt and showcase its newest hardware, tying the fastest-growing AI research operations directly into its ecosystem.

And now Thinking Machines - - Nvidia’s latest chosen lab - - just signed a $471.7 million non-cancelable contract to run its most demanding workloads on Boost Run’s infrastructure.

Nvidia picked a winner. And the winner picked Boost Run.

This is the Nvidia flywheel in motion.

Thinking Machines is co-developing optimized training and inference pipelines tuned tightly to Nvidia’s software and networking stack.

When a lab like that goes shopping for a high performance computing provider, it doesn’t pick randomly. It picks the infrastructure operator that Nvidia has certified and validated against its own reference standards.

Out of the entire global AI infrastructure industry, only five providers hold Nvidia Exemplar Cloud designation on Blackwell architecture:

CoreWeave , Nebius, Oracle , Microsoft Azure … and Boost Run.

Thinking Machines chose Boost Run.

This isn’t a coincidence. This is how Nvidia builds its ecosystem.

The chipmaker doesn’t just sell hardware. It creates concentric rings of validated, trusted partners - - and the operators inside that circle win customers that operators outside it simply cannot.

The 5,000 Nvidia B300 GPUs being deployed under this contract represent some of the most in-demand compute infrastructure on the planet.

The fact that Boost Run can fulfill an order at that scale, for a customer at that caliber, tells you something important about where this company sits in the hierarchy.

Now consider what this does to the numbers.

The Revenue Ramp Has Already Started. The Market Just Hasn’t Priced It In.

Boost Run entered its Nasdaq listing with $940 million in contracted revenue and guided to at least $375 million in annualized recurring revenue by year-end.

The Thinking Machines agreement adds $471.7 million in fully committed, non-cancelable contract value on top of that foundation.

That’s not upside speculation - - it’s signed paperwork with no exit ramp for the customer.

For context: when I wrote about Nebius in December 2024, the company had $170-190 million in annualized revenue and a $31 stock price.

Investors who acted on that report watched Nebius shares soar 642% to $230.

You can read my original Nebius report here:

Boost Run is starting from a larger contracted base, with a customer roster that now includes one of the most watched frontier AI labs in the world, backed directly by Nvidia.

The original signal I identified was straightforward:

Nvidia marks its preferred operators, and the market eventually prices that in.

With Nebius, Nvidia’s direct investment was the signal.

With Boost Run, the signal is compounding.

Nvidia certified Boost Run as an Exemplar provider, then invested in a frontier lab, and that frontier lab just committed $471.7 million in non-cancelable revenue to Boost Run’s infrastructure.

Much like Nebius in late 2024, the Boost Run growth story today is just getting started.

The window I described on Day One is still open.

But with a contract of this size, from a customer of this profile, now in the public filings - - it won’t stay open long.

Equity Empire Subscribers Who Bought Boost Run on My Initial Buy Alert are Already Up More Than 100%. The Next Alert Goes Out Soon.

My paid subscribers got the Boost Run buy alert before the SPAC merger closed. And many of them are already up more than 100%.

Want access to my next alert before the crowd rushes in?

Join Equity Empire Today ↗

John McHugh co-founded Equity Empire with Colin Tedards to help small investors achieve financial freedom. He’s a former Wall Street analyst and hedge fund manager with more than 30 years of experience investing in small cap stocks. This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

To join the fast growing Equity Empire subscriber community, visit:

myequityempire.com/elite-yearly-subscription/

This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

By John McHugh, Equity Empire Small Cap Research

Seven weeks ago, I told you SpaceX was going public.

Most people thought it was noise.

Wall Street was guessing at dates.

The S-1 was a rumor. The timeline was soft.

I said act now anyway.

I even shared 5 stocks which were ideally positioned to soar ahead of the IPO launch date.

Read my original report here:

Subscribers who acted on the 5 names I listed in the report above are already sitting on gains of 40% to 54% in just seven weeks.

And subscribers to our paid service at Equity Empire are up nearly 100% from my early buy alert on Spire Global.

They didn’t wait for the headlines.

They got ahead of them.

And here’s the thing.

The real headline still hasn’t hit.

On May 20, 2026, SpaceX publicly filed its S-1.

Not a whisper. Not a rumor.

A formal filing with the SEC.

The roadshow starts June 8.

The Nasdaq debut follows on June 12 under the ticker SPCX.

The financial media is talking about a SpaceX valuation approaching $1.75 trillion - - with some analysts already whispering north of $2 trillion.

The valuation hasn’t been determined yet.

But one thing is for sure.

This will be the single largest IPO in the history of capital markets.

Not the largest tech IPO.

The largest of any public financing. Ever.

IPO date: June 12

Nasdaq ticker: SPCX

Target valuation: $1.75T+

Days remaining: 17 days until June 12 listing

Now think about what happens the moment SPCX begins trading.

Every institutional investor on the planet - - pension funds, sovereign wealth funds, endowments, family offices - - will need a view on stocks in the space economy.

Most of them don’t have one today.

Their analysts will scramble to build coverage universes with detailed financial models overnight.

Many fund managers will deploy capital into every publicly traded name in the sector, looking for exposure they can actually buy.

But here’s the problem:

Most of them will be too late to buy the smaller names cheaply.

The five stocks I highlighted in April have already had big moves.

But the institutional wave that reprices this entire sector hasn’t arrived yet. That happens on June 12 - - or just before it.

The biggest gains in a sector re-rating always go to the investors who were already in position when the catalyst hit.

We saw it with cloud software stocks after Amazon AWS.

We saw it with AI infrastructure stocks after Nvidia .

And now we’re about to see it with the emerging space economy.

The window between now and June 12 is your final opportunity to get ahead of the flood of institutional money that is about to discover these companies for the first time.

Here’s where my five original proxy plays stand today:

1. Planet Labs - - The Daily Imaging Machine

Planet Labs is proving it can convert raw satellite data into highly profitable software. The company just crushed earnings expectations for the fourth consecutive quarter, hitting a vital financial milestone: adjusted EBITDA breakeven.

Financial Snapshot

• Q4 FY26 Revenue: $86.8M (+41% YoY)

• Total Backlog: ~$900M (Remaining performance obligations exceed $852M)

• FY27 Guidance: $415M–$440M (~39% growth at the midpoint)

• Next Catalyst: Q1 earnings on June 4 (Estimated $90M revenue)

Defense and intelligence demand grew by more than 20% year-over-year, acting as the primary engine for the business. Trading just above $48, early believers have notched massive gains over the last three years.

PL is up nearly 700% since I pointed out the opportunity to my followers on X almost a year ago:

But there could be more upside in store for PL investors as the space economy continues its rapid expansion.

The Play: Planet reports Q1 results on June 4 - - right in the dead-center of the SpaceX roadshow.

Expect fireworks.

2. BlackSky Technology - - The Tactical Intelligence Layer

BlackSky delivers rapid, AI-processed tactical satellite intelligence to global buyers. In Q1 2026, the company posted $20.8 million in revenue and promptly bumped its full-year guidance.

Financial Snapshot

• Q1 2026 Revenue: $20.8M

• Total Backlog: ~$380M (+32% YoY, with $90M converting in 2026)

• FY26 Guidance: $130M–$150M (Adjusted EBITDA of $12M–$24M)

BlackSky’s new Gen-3 satellite platform delivers razor-sharp, sub-50cm resolution imagery in under 40 minutes. CEO Brian O’Toole noted that dozens of early customer pilots are transitioning from six-figure test runs to massive seven- and eight-figure subscription contracts.

The Play: BlackSky sells geopolitical sovereignty. Foreign governments that used to rely on U.S. permission for satellite eyes can now simply buy their own capabilities. This demand won’t evaporate when the SpaceX IPO hype cycle slows down.

3. York Space Systems - - The Satellite Factory

York manufactures and operates modular small satellite platforms - - including spacecraft, launch, ground systems, and operations - - for national security, government, and commercial customers.

Since going public in January 2026, York has cleared the launchpad with ease. Its first quarter as a public company beat expectations, posting $116.3 million in revenue against the $110.7 million consensus.

Financial Snapshot

• Q1 2026 Revenue: $116.3M

• Total Backlog: $642.3M (+18% Quarter-over-Quarter)

• FY26 Guidance: $545M–$595M (70% already covered by backlog)

• Liquidity: $806M in total liquidity ($655.7M in cash)

The Play: York is positioning itself o be a low-cost, vertically integrated space-mission prime that delivers complete end-to-end solutions - - modular satellites, launch, ground systems, and operations - - at roughly half the price of competitors, primarily for U.S. defense and government customers.

York is also aggressively expanding via M&A. Alongside Q1 results, they announced the acquisition of Orbion (to accelerate satellite integration) and signed a deal to acquire ALL.SPACE, adding jam-resistant tactical communication terminals to their military portfolio.

4. Spire Global - - The Data Transformation Play

Spire operates a large constellation of nanosatellites to collect real-time ship tracking, aircraft data, weather patterns, and other Earth observations which it sells as analytics to governments and businesses worldwide.

Spire’s transformation is paying off beautifully. After shedding its lower-margin maritime business, the company beat the high end of its Q1 2026 guidance, bringing in $15.8 million. Core revenue grew 13% year-over-year, driven by heavy civil government weather data purchases.

Financial Snapshot

• Q1 2026 Revenue: $15.8M

• Non-GAAP Gross Margin: 44% (+5 percentage points YoY)

• FY26 Guidance: $75M–$85M (>50% ex-maritime growth, 76% under contract)

• Balance Sheet: Debt-free following a $70 million private placement in April

The upcoming catalyst to watch is an active $11.2 million NOAA radio occultation contract, with an expanded follow-on award expected later this year. Spire’s total active NOAA pipeline exceeds $150 million for 2026 bids.

The Play: Management is guiding for adjusted EBITDA breakeven by late 2026/early 2027. With the stock already up 140% year-to-date, analysts are aggressively raising price targets to account for Spire’s high-margin on-orbit data dominance.

5. Satellogic - - High-Risk, Asymmetric Underdog

Satellogic is the classic turnaround story: tiny revenue, high risk, but immense upside potential if it connects. Q1 2026 revenue jumped 80% to $6.1 million, and the company achieved a historic first: positive operating cash flow ($0.2 million).

Financial Snapshot

• Q1 2026 Revenue: $6.1M (+80% YoY)

• Cash Position: $121.9M

• Next Milestone: Merlin Constellation launch in October 2026

Satellogic’s AI-first, defense-focused Merlin constellation is fully funded by a $30 million contract from a strategic defense customer. The company also secured an $12 million sovereign defense contract and boosted its total pipeline to “just under $1 billion.” To cement its defense pivot, the company just added Vice Admiral Frank D. Whitworth III (former director of the NGA) as a strategic advisor.

The Play: This is still the riskiest stock on the list. But with accelerating revenue, narrowing losses, and a healthy $121.9 million cash cushion, it represents a highly asymmetric bet for risk-tolerant investors.

The Big Picture: Catalyst vs. Risk

The companies on this list don’t compete with SpaceX. Planet isn’t trying to build heavy-lift rockets; Spire isn’t trying to launch a global broadband internet network. They occupy highly specialized, highly profitable adjacent corners of the same cosmic ecosystem.

When institutional funds build out their mandatory “Space Portfolios” this June, they will need diversified sector exposure.

The window to buy these adjacent players at a discount is closing fast.

Find the Next Sector Leader with Equity Empire

At Equity Empire, we don’t wait for an S-1 filing to tell us where the market is going.

We do the deep-dive fundamental research to find explosive sector trends before they become mainstream news.

Our track record speaks for itself:

• Equity Empire subscribers have achieved major wins in Intel, Broadcom, Nebius, Applied Optoelectronics, Spire Global, Bloom Energy, Boost Run, and many more.

• The Space Sector Sector Play: Called before the SpaceX S-1 hit the newswires, returning 40% to 54% across the five names in my original report in just seven weeks.

The next major sector call will soon be on the way to our members’ inboxes.

Don’t watch history happen with the crowd while sitting on the sidelines.

The stock market in the coming year will be filled with life-changing profit opportunities.

The time to take action towards securing your financial future is now.

Join Equity Empire Today ↗

John McHugh co-founded Equity Empire with Colin Tedards to help small investors achieve financial freedom. He’s a former Wall Street analyst and hedge fund manager with more than 30 years of experience investing in small cap stocks. This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

To join the fast growing Equity Empire subscriber community, visit:

myequityempire.com/elite-yearly-subscription/

This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

By John McHugh, Equity Empire Small Cap Research

An emerging Nvidia competitor dominated the financial news cycle last week.

Fast growing AI Chipmaker Cerebras conducted its Initial Public Offering (IPO).

The IPO was 20 times oversubscribed.

And because of this overwhelming surge in demand, the CBRS IPO was a big hit - - the stock soared 68% on its first day of trading.

Yet, while all eyes were on Cerebras, almost nobody asked the obvious question:

Who is going to house and power the Cerebras chips?

The answer:

A little known company that’s building out a 40-megawatt AI data center campus in Columbiana, Alabama.

Purpose-built, Tier III certified, high-density, engineered from the ground up around the thermal requirements of next-generation AI accelerator hardware.

Just weeks ago, this little company signed a $1.1 billion, ten-year Master Services Agreement with Cerebras.

The company’s name is Digi Power X .

Few investors have heard of the company … yet.

But they will soon.

From Boring Bitcoin Miner to Fast Growing AI Infrastructure Operator

Digi Power X’s transformation is the real story here.

And it’s one that most investors are still catching up to.

The company - - formerly known as Digihost Technology Inc. - - spent years as a Bitcoin miner.

Steady, cyclical, and tied to crypto prices.

Then management made a bold decision that now looks visionary:

Pivot the entire business toward AI data center infrastructure, using the company’s most valuable assets - - secured power capacity and owned real estate - - as the foundation.

To execute that pivot, DGXX management hired Jagan Jeyapaul as Chief Technology Officer and Venkat Rangasamy as VP of AI Infrastructure.

Jeyapaul comes from Oracle, Equinix, and VeriSign, where he built and automated large-scale cloud and data center platforms across more than 200 facilities.

(Pictured above: DGXX CTO Jeyapaul next to a giant Nvidia B300 GPU rack)

Rangasamy is a former Engineering VP at Oracle who directed the migration of 6,000+ customers and 18,000 pods to Oracle Cloud Infrastructure.

These two datacenter and cloud technology pros are the kind of engineering talent you hire when you’re building a serious AI datacenter company.

And they’re assembling a team of experienced datacenter builders to execute the company’s rapid growth plans.

DGXX management recently made it clear that staffing for the Columbiana campus is actively underway, with hiring aligned to meet a Phase 1 launch target by year end 2026.

And with plans to open a new office in Silicon Valley in June, DGXX will likely bring on additional engineering talent to help Jeyapaul and Rangasamy execute on their ambitious expansion plans over the coming 12-24 months.

The Cerebras Contract: Why It’s a Game Changer for DGXX Investors

DGXX’s agreement with Cerebras has an initial ten-year term valued at approximately $1.1 billion, with a total potential contract value of $2.5 billion inclusive of renewal terms. Cerebras also holds an expansion option worth an additional $1.4 billion.

The campus will be delivered in two phases:

Phase 1 comprising 15 MW of IT load, targeted ready-for-service by December 15, 2026, and Phase 2 delivering an additional 25 MW by the end of Q1 2027.

The substation is already built. Grid interconnection is finalized. A power delivery agreement with Alabama Power is in place. The land is owned outright.

Once this capacity is up and running, it’s expected to generate over $100 million in annual revenues for DGXX.

The Cerebras agreement is structured as a modified lease model, providing stable, recurring colocation income without requiring the same level of GPU capital expenditure. The structure is designed to give Digi Power X long-term revenue visibility from day one of service.

NeoCloudz: The Company’s Second Revenue Engine

Colocation is only half of the model.

Digi Power X’s NeoCloudz platform delivers GPU-as-a-Service on dedicated bare metal NVIDIA $NVDA infrastructure.

In simple terms, this means that customers rent raw compute capacity without the overhead of a shared cloud environment.

NeoCloudz recorded its first AI revenue in May 2026, with an initial deployment of NVIDIA B200 and B300 GPUs at the Alabama facility.

The first GPU rental contract - - a 24-month agreement with SubQuadratic AI valued at $19.6 million - - went live in early May, delivered on time through coordination between NVIDIA, Supermicro , and Digi Power X’s Alabama team.

The company expects to deliver an additional 6 megawatts of GPU capacity by the end of 2026, with revenue potentially beginning in Q4 2026 or Q1 2027.

And this is just the start.

Put the two segments together and an attractive revenue picture for 2027 comes into focus.

Management’s guidance currently targets $250 to $300 million in total 2027 revenue across AI colocation, GPU-as-a-Service, and energy sales.

Against a current market cap of approximately $650 million, that implies the stock is trading at roughly 2x forward revenue.

Which is cheap for a company with a signed $1.1 billion anchor contract, 400 megawatts of secured power capacity across Alabama, New York, and North Carolina, zero long-term debt, and $125 million in cash and digital assets on hand.

DGXX is Trading at a Big Discount to its High-Flying Competitors

Here’s where the DGXX story gets genuinely interesting for investors paying attention to how peers are valued.

Compare DGXX to the names the market already loves in this space.

The valuation gap is striking.

DGXX is trading at roughly 1/10th the market cap per secured megawatt of its closest peers.

Even adjusting for execution risk and the fact that the company’s Columbiana facility is still under construction, the discount is extreme.

The revenue multiple tells the same story.

, and trade at 5x to 15x forward revenue on their 2027 estimates.

DGXX - - with a 2027 revenue target of $250 to $300 million anchored by a signed, billion-dollar contract - - currently trades at roughly 2.5x.

If DGXX delivers on its revenue growth targets and gets repriced to just half the per-MW multiple of its peers - - a conservative ask given the quality of the Cerebras anchor tenant and the company’s owned real estate and zero-debt balance sheet - - DGXX shares could more than triple from here.

The Risks Are Real

This isn’t a guaranteed outcome.

Phase 2 of the Cerebras build is conditioned on financing that hasn’t yet been secured. The share count has expanded meaningfully through ATM issuance and will likely expand further, creating ongoing dilution pressure. Revenue doesn’t begin in earnest until late 2026, meaning investors are being asked to pay for a story that is still being written.

Building a world-class Tier III data center operation also requires substantial investments in human capital.

And the broader AI infrastructure sector has to deliver on lofty growth expectations to justify current valuations across the board.

Add up these risks and DGXX is definitely a highly speculative investment.

So if you decide to invest after doing your own due diligence, size accordingly to account for these risks.

The Bottom Line: An Exceptional Growth Opportunity

Cerebras just went public as the most anticipated IPO of 2026.

The demand for its chips is validated by a $20 billion agreement with OpenAI and overwhelming investor interest.

The AI high performance computing wave it’s riding is real.

DGXX is building the house that holds Cerebras’ chips.

The company has:

• approximately 210 megawatts connected to the grid today

• a total secured capacity pipeline of nearly 400 megawatts

• a signed $1.1 billion anchor contract with first AI revenues live

• a Silicon Valley CTO from Oracle and Equinix overseeing a Tier III technology roadmap

• and a 2027 revenue target of $250 to $300 million.

In sum, DGXX looks like a company that the market has not yet properly priced.

The company’s peers are valued at 5 to 15 times their per-megawatt capacity.

DGXX is currently trading at a fraction of these valuation metrics.

But if DGXX management can execute efficiently on its enormous growth potential, this valuation gap will start to close.

Forward thinking investors will take action before this repricing happens.

The Window Is Open. It Won’t Stay That Way.

Equity Empire subscribers stay steps ahead of the crowd.

We called Nebius at 31.

We called Applied Optoelectronics at 10.70

Intel at 20.05

And Broadcom at $172.30

These weren’t lucky guesses.

They were the result of a repeatable process - - finding structurally mispriced, institutional-ignored companies before the crowd arrives.

Right now, that same process is pointing directly at DGXX.

The window is open - - but windows like this don’t stay open long.

Investors who act while a stock is still cheap are the ones who build real wealth.

I don’t want you to miss our next big wealth building opportunities.

Get the full Equity Empire DGXX report — PLUS a full year of our top research ideas like the ones above delivered straight to your inbox by clicking here: myequityempire.com/elite-yearly-subscription.

Join Equity Empire today and start building your own financial Empire!

John McHugh co-founded Equity Empire with Colin Tedards to help small investors achieve financial freedom. He’s a former Wall Street analyst and hedge fund manager with more than 30 years of experience investing in small cap stocks.

To learn more about Equity Empire and get ongoing access to John’s top small cap ideas, visit:

https://myequityempire.com/elite-yearly-subscription/

This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

By John McHugh, Equity Empire Small Cap Research

On December 5, 2024, I published a report on Nebius Group .

The stock was around $31.

No Wall Street analyst covered it.

Most investors had never heard of it.

I wrote it was one of the fastest-growing AI stocks very few people knew about - - and that Wall Street was way behind.

You can read the article here:

Eighteen months later, Nebius trades at $193.

The pattern I spotted back then - - elite business, structurally mispriced, ignored by institutions - - played out exactly as I forecast.

But I’m not here to talk about Nebius anymore. That ship has sailed.

What I want to talk about is a promising young company that just started trading on Nasdaq yesterday morning.

Because the setup looks almost identical to what I saw in Nebius in December 2024.

And the window to act is open right now.

What Made Nebius So Obvious to Me

The AI boom created a problem hiding in plain sight.

The world decided, simultaneously, that artificial intelligence would transform every industry on earth.

What nobody was saying loudly enough was that all of it - - every model, every agent, every inference - - ran on physical hardware.

Racks of Nvidia GPUs, housed in data centers, maintained by companies that built the infrastructure the AI revolution would run through.

The model builders got the headlines. The infrastructure operators got overlooked.

Nebius was one of those operators.

It built AI data centers, leased GPU compute to enterprises, and collected recurring revenue while demand grew faster than anyone could build supply.

When I wrote my report, the company had just raised $700 million with Nvidia as lead investor.

Annualized revenue was $170–190 million, the CEO had committed to $1 billion by end of 2025, and the balance sheet held nearly $3 billion in cash to make it happen.

I wrote: Wall Street is still way behind on this one.

My analysis was spot on.

The Same Opportunity Has Popped Up Again

Yesterday morning, Boost Run $BRUN began its first day of trading.

I’ve been waiting eighteen months for a setup that felt like this one.

Boost Run is an AI cloud infrastructure company - - same business as Nebius.

GPU clusters in data centers, on-demand compute for AI training and inference, recurring contracted revenue with exceptional margins.

Its customers span financial services, healthcare, defense, and frontier AI research - - industries that cannot afford infrastructure that underperforms.

The company trades at an enterprise value of $1.4 billion against $940 million in contracted customer revenue, with guidance for at least $375 million in annualized recurring revenue by year-end.

That’s a forward multiple of 3.4x.

Nebius today trades at approximately 14x its 2026 estimates.

That gap isn’t a reflection of business quality.

It reflects the fact that Boost Run just started trading yesterday.

CEO Andrew Karos previously built an algorithmic trading firm across thirteen data centers and seven countries before selling it to Galaxy Digital in 2020.

When he started Boost Run, he did something almost no one in this space has managed:

He made it capital efficient from day one.

No Series A financing.

No venture dilution.

Every dollar in hardware generates $0.70 in annual recurring revenue within twelve months.

EBITDA margins are projected at roughly 80%.

The company is built for rapid sales growth and profitability.

When Nvidia Picks a Winner, Smart Investors Pay Attention

The most powerful signal in my Nebius report was Nvidia’s direct investment.

My reasoning:

Nvidia’s chips power the vast majority of serious AI workloads on earth, and when supply is constrained - - as it has been for years - - Nvidia decides who gets hardware.

A nine-figure check into Nebius was Nvidia marking its preferred operator.

The market eventually agreed.

Boost Run carries the same quality mark but through a more rigorous mechanism.

In April 2026, it achieved NVIDIA Exemplar Cloud designation on Blackwell architecture - - a technical certification where Nvidia independently benchmarks a provider against real-world AI workloads and validates performance within 95% of its own reference standards.

The companies that have earned it:

CoreWeave , Nebius, Oracle , Microsoft Azure.

And Boost Run.

Five providers.

Out of the entire global AI infrastructure industry, Nvidia has certified five.

Nebius earned it first, on H200 GPUs and later the GB300 NVL72.

Boost Run now shares that designation.

Exemplar providers get chips, get priority access to new architectures, and attract enterprise customers who cannot afford unvalidated infrastructure.

It’s a structural moat, not a marketing badge.

And the fact that both Nebius in 2024 and Boost Run today hold it tells you something important about the kind of operator Boost Run already is.

The Contract Book Is Already Full

One risk with early-stage growth companies is that revenue is hypothetical - - future contracts, anticipated demand, management’s best guess dressed up in a slide deck.

Boost Run’s revenue isn’t hypothetical.

The $940 million in contracted customer revenue is largely already in production: customers signed, hardware deployed, cash flowing.

A two-year $127 million agreement with Fluidstack.

A $1.44 billion commitment with Dell.

Management expects to exit 2026 at a $375 million annualized quarterly run rate, with the SPAC transaction adding more than $120 million earmarked for $400 million in hardware deployment across Charlotte, Raleigh, Dallas, Seattle, and Eagan.

Deploy the capital, install the GPUs, collect the revenue.

Then watch the gap close.

Nebius went from $170–190 million annualized to $1 billion in twelve months.

The stock went from $31 to nearly $200.

Boost Run is starting on a similar path.

The Window Is Open. It Won’t Stay That Way For Long.

Within weeks of my December 2024 Nebius report, the stock started moving.

Today Nebius has a $46 billion market cap, ten Buy ratings, a $2 billion follow-on Nvidia investment, and multi-billion dollar contracts with Microsoft and Meta .

That window closed before most people knew it had opened.

Here’s what Boost Run looks like today:

$940 million in contracted revenue, 80% EBITDA margins, Nvidia Exemplar certification shared by only four other providers on earth, a founder with a proven nine-figure track record - - and a 3.4x forward multiple in a sector where the market pays well of 10x.

Every one of those facts is in the public filings.

The question is not whether Boost Run is a good business.

The filings show that it is.

The question is whether you act before the institutions do - - before the first analyst initiates, before the first major publication runs a feature, before the stock is up 50% and people wish they had bought on day one.

My paid subscribers at Equity Empire are already up more than 40% from my initial buy alert.

But they bought before the merger closed.

Today is Day One for Boost Run as a publicly traded company.

Investors who recognize Boost Run’s exciting growth story before Wall Street does will look back on today the same way Nebius readers look back on December 2024.

And they’ll smile brightly as they check their soaring brokerage account balance.

To get access to my full Boost Run report - - and a full year of my best small cap profit opportunities - - click the link below:

https://myequityempire.com/elite-yearly-subscription/

I believe the next 18 months will provide forward-thinking investors with an opportunity to achieve financial freedom.

Come join Colin, myself, and the fast-growing Equity Empire community to learn how.

John McHugh co-founded Equity Empire with Colin Tedards to help small investors achieve financial freedom. He’s a former Wall Street analyst and hedge fund manager with more than 30 years of experience investing in small cap stocks.

To learn more about Equity Empire and get ongoing access to John’s top small cap ideas, visit:

https://myequityempire.com/elite-yearly-subscription/

This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

By John McHugh, Equity Empire Small Cap Research

Back in December 2025, I published a piece called “The $100 Billion Bottleneck Nobody’s Talking About.”

The premise was simple:

Hyperscalers were spending hundreds of billions building AI datacenters, but the copper cables connecting the GPUs inside those facilities were becoming a fundamental constraint.

The financial media wasn’t covering it.

No major Wall Street firm had published a cohesive framework around it.

The stocks involved were climbing but still mostly under the radar.

To me, the opportunity was sitting in plain sight.

But no one was talking about it.

So I posted this article on Equity Empire’s Substack page:

Four months later, in April 2026, Goldman Sachs published a report titled “Global Tech: Optical Networking: The Next Mega Trend in AI Infrastructure.”

Their conclusions tracked what I had written about in December - - the same bottleneck, the same technology transition, many of the same companies.

This isn’t about taking credit.

It’s about what the timing of the two reports reveals:

That independent, ground-level research can identify major investment themes before they become institutional consensus.

And that the investors who act on that research early are the ones who capture the most value.

The Thesis, Built From the Ground Up

The core of my December argument was simple.

Copper cables simply cannot handle the bandwidth, latency, and power efficiency requirements of next-generation AI compute clusters.

I wasn’t bashing copper interconnect technology.

In fact, a year before launching Equity Empire, Colin and I wrote a 2000+ word report that highlighted the growth potential of copper interconnect technology developer Credo Technology .

And our subscribers who bought Credo after our report saw gains as much as 800% !

But even though we wrote about Credo’s growth potential in the report, we still forecasted the following:

The industry would have to transition to optical networking in stages.

And each stage would create identifiable investment opportunities for investors paying close enough attention.

Goldman’s April report confirmed this framing precisely.

Today, the connections between racks of AI chips, a domain that was 100% copper until recently, are opening up as a brand new optical market.

This is essentially the same argument I made four months earlier.

In the article, I organized the transition into three phases.

Phase One was the optimization of existing copper infrastructure through serialization technology.

Phase Two, which I identified as the active moment in December 2025, was the copper-to-optical transition, with Coherent Corp as a primary beneficiary.

Phase Three was what I called the real destination:

Full silicon photonics integration, co-packaged optics (CPO), and photonic integrated circuits weaving electronics and light into a single architecture.

Goldman’s April report maps the same arc.

They note that CPO technology has clear advantages in reducing latency and power consumption, with Nvidia planning mass production of CPO switches in early 2026 and Broadcom having already delivered related products to customers in late 2025.

These were the same technologies I had named in December as the logical endpoint of the transition.

The Numbers Tell the Story

In my December report, I estimated the optical opportunity to be around $100 billion.

Goldman subsequently formalized the numbers in its recent report:

They forecast the total addressable market expanding from approximately $15 billion to about $154 billion by 2027 - 2028.

That’s a roughly 90x increase !

My estimate, arrived at through bottom-up reasoning rather than a formal institutional model, landed squarely in the right range.

And you had the chance to read about it months before Goldman’s team published their report.

The companies I flagged for subscribers told the same story.

Coherent, Lumentum , and Marvell all appear prominently in Goldman’s subsequent coverage.

Each of these was a name Equity Empire subscribers were already familiar with before Goldman formalized its view.

Why Equity Empire Gets There First

Goldman’s April 2026 report is a significant institutional document.

It will move capital, re-rate valuations, and bring a new wave of investors into the optical networking theme.

But the investors who benefit most from that report aren’t the ones reading it in April.

They’re the ones who read the December 2025 Equity Empire analysis, understood the three-phase framework, and positioned themselves before the institutional money arrived.

This is the core value of independent research done well - - and it explains why Equity Empire exists.

Institutional research operates on a slower cadence.

Goldman likely had analysts tracking this trend well before April 2026.

But the report represents when they chose to formalize and publish their conclusions for clients - - not when they first arrived at them.

Equity Empire’s independent research doesn’t have publication time constraints.

When a thesis is clear, we publish it.

That structural timing advantage is real.

And for investors willing to act on it, it translates directly into returns.

That logic of the massive optical networking opportunity was accessible to anyone who read my December 2025.

The Equity Empire research process was built to show our subscribers megatrend opportunities early.

Not after a leading Wall Street firm has shown them to the entire world.

We Help You Stay One Up on Wall Street

Goldman’s April 2026 report is now an important marker in the optical networking story - - the moment the thesis moved from independent insight to Wall Street consensus.

There’s still profit potential in the trade.

But the most valuable window - - the one where investors could build positions before institutional recognition drove valuations higher - - opened in December 2025.

That’s when Equity Empire subscribers got to read about the opportunity.

I describe this timing advantage by referencing the words of hockey Hall of Fame hockey player Wayne Gretzky:

“I skate to where the puck is going to be, not where it has been.”

The ability to write about where the puck is going rather than constantly chasing after it is precisely where Equity Empire research provides the most value.

Every AI chip requires connections.

Every new generation requires more of them.

We saw that coming.

We wrote it down.

And our subscribers had four months to act on it before the big firms on Wall Street arrived.

Where’s the Puck Heading Next?

I recently issued a new report on a small cap stock that almost nobody is talking about today.

But I expect the stock to get lots of attention from both the big Wall Street firms and the financial media over the coming months and years.

I believe it has the potential for investors who act early to achieve exponential gains.

And you can have the chance to learn about the opportunity before the crowd rushes in.

To get access to the report - - and a full year of my best small cap profit opportunities - - click the link below:

https://myequityempire.com/elite-yearly-subscription/

I believe the next 18 months will provide forward-thinking investors with an opportunity to achieve financial freedom.

Come join Colin, myself, and the fast-growing Equity Empire community to learn how.

John McHugh co-founded Equity Empire with Colin Tedards to help small investors achieve financial freedom. He’s a former Wall Street analyst and hedge fund manager with more than 30 years of experience investing in small cap stocks.

To learn more about Equity Empire and get ongoing access to John’s top small cap ideas, visit:

https://myequityempire.com/elite-yearly-subscription/

This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

Their own CFO is worried they can’t pay for the data centers they ordered.

The Wall Street Journal called it the first crack in the AI trade.

They’re reading it backwards. This story is the cleanest validation we’ve gotten of the rule that quietly runs the entire AI trade - and once you see it, the rest of this cycle gets a lot simpler.

Today I'm going to walk you through the rule that's quietly run AI for two years, why the WSJ story confirms it instead of breaking it, and exactly which names are about to get repriced - for better and for worse.

The Rule That Runs the AI Trade

Here’s the rule.

95% of people who use AI never pay a dollar.

The other 5% pays for everything.

Roughly 900 million people use ChatGPT every week. About 50 million pay. That’s 5.5% of users carrying the entire $25 billion in annualized revenue. OpenAI’s internal conversion target is 8.5%. But they’ve been stuck at roughly 6% for a while.

Anthropic is even more concentrated. They hit $19 billion in annualized revenue in March 2026 - up from $1 billion just 14 months earlier. That money doesn’t come from hundreds of millions of users. It comes from roughly 300,000 business customers. Over 500 of those customers spend more than $1 million a year each.

In a world where over 1 billion people have “tried AI” … just a small few actually pay for it.

This isn’t a quirk. And I don’t expect this to change. This is the entire AI industry.

One Afternoon, One Year, Same Compute

The 5% doesn’t just pay more. They use orders of magnitude more.

A free ChatGPT user sends maybe 15 queries a day. In AI terms that’s only around 10,000 tokens. The financial cost to OpenAI is fractions of a penny.

A single developer using Claude Code could easily burn through 2 to 5 million tokens a day. Easily. That would cost: $6 to $12+ per day.

That’s a 200 to 500x difference per person.

This obviously gets more extreme at the enterprise level. Companies running agentic workflows burn through 10 million tokens per day per person.

The cleanest way to think about it: a single developer using Claude Code consumes more compute in one afternoon than a casual ChatGPT user will in an entire year.

And the gap is widening every quarter. Anthropic’s revenue tripled in the last three months. H100 rental prices surged 40% over the past 6 months, because the power users are outbidding each other for capacity.

This infrastructure isn't being built for the 95%. It's being built for the 5%. And the 5% is accelerating.

What the WSJ Story Actually Says

Now go back and re-read the WSJ headline. OpenAI Misses Key Revenue, User Targets in High-Stakes Sprint Toward IPO.

That’s not an AI demand problem. That’s an OpenAI problem.

OpenAI built its business on the assumption that ChatGPT could pull a massive number of free users up the funnel into paid plans. That’s the bet. But it’s not going to work.

Google and Meta will deliver “free AI” to the masses. They already have the users, and more importantly a business model that has been monetizing free users at scale for two decades. OpenAI is trying to build that from scratch, in real time, with hundreds of billions in obligations already on the clock.

Meanwhile companies that bet on the 5% directly - are only struggling to keep up with demand.

The WSJ frames OpenAI’s miss as the AI bubble cracking. The data says it’s the AI trade rotating.

There’s a big difference. And the difference matters for your portfolio.

The Task Doesn’t Go Back

Here’s the natural pushback: if the 95% bet is breaking at OpenAI, won’t the 5% crack next?

I wrote about why back in September, in a Substack piece called Your Job Isn’t Being Replaced. It’s Being Repriced.

The 95% aren’t building anything with AI. They’re playing with it. Settle a bet. Rewrite an email. Ask it a trivia question, close the tab, go back to doom scrolling Instagram.

That’s not demand - that’s curiosity.

And curiosity doesn’t pay.

That’s exactly what’s happening at OpenAI right now. The funnel from free chat to paid plan is breaking because the casual user never actually needed the upgrade.

The 5% is a different class of customer entirely. They’re builders. Coders. Founders. Researchers. Operators. People who use AI to compress decades of expertise into a weekend. People who run entire businesses out of a single Claude account. They’re not consuming AI for entertainment - they’re using it to do things with real economic output.

And once those tasks shift to AI - they don’t go back.

This is how every prior tool worked. Calculators didn’t go back to slide rules. Spreadsheets didn’t go back to ledger paper. Word processors didn’t go back to typewriters.

The task shifts once, and the labor cost gets repriced forever.

That’s why the 5% spend keeps compounding. Anthropic isn’t growing 19x because the 5% is curious. It’s growing because every task that’s been handed to Claude Code can’t be handed back to human hands.

95% of AI users are tourists. The other 5% are residents. Tourists leave when the weather turns. Residents stay for the long term.

The Repricing, Not the Reckoning

The takeaway isn’t complicated.

The 95% will keep calling AI overhyped because their experience with it is genuinely underwhelming. Nothing has changed there. They use ChatGPT the same way today as they did a year ago.

That’s not worth paying for.

The 5% is accelerating. Nothing has changed there either. The 3 generation old H100 rental spot market proves it. The compute layer is going to keep getting bid because the 5% is willing to keep paying for it.

What changed is that the market just learned not every lab gets to ride the 5% wave equally. OpenAI’s model - pull users up the funnel from free chat - is hitting friction. Anthropic’s model - sell API and Claude Code directly to the 5% - is compounding.

That’s a counterparty risk story for the names that wrote OpenAI a check. It is not a thesis-breaking story for the AI trade.

If you’ve been doing the work, you’ve been positioning into the right side of this rotation for months.

If you haven’t, the WSJ story is going to scare you out of names that are about to print money - and you’re going to confuse that fear with insight.

The first crack in the AI trade isn’t a crack. It’s a rotation. And the people who understand the 5% are about to get a clean entry into names everyone else is panic-selling.

Follow Colin on X: https://x.com/InvestorVideos

By John McHugh, Equity Empire Small Cap Research

It’s 2026, and the world is focused on artificial intelligence.

Data centers are popping up across the countryside.

The tech giants - - Amazon , Alphabet , Microsoft , Meta - - are pouring hundreds of billions of dollars into AI infrastructure.

Models that predict the weather.

Diagnose cancer before your doctor can.

Write code, design logos, generate entire websites in seconds.

But here’s the problem nobody fully saw coming when ChatGPT took the world by storm three years ago:

AI doesn’t run on dreams.

It runs on power.

Massive, unrelenting, grid-straining power.

The Demand Surge That No One Planned For

A single hyperscale AI data center can consume as much electricity as a mid-sized American city.

Hundreds of megawatts.

Sometimes gigawatts when you factor in training the next giant model and keeping the server racks from literally melting.

Now multiply that by the hundreds of data centers being planned, permitted, and built across the United States right now.

The U.S. electrical grid is already creaking from decades of chronic underinvestment.

And now it’s facing a demand surge that could double electricity consumption in some regions before the end of the decade.

The same grid that struggled during the Texas freeze of 2021.

The same infrastructure that loses roughly 5% of all electricity generated just in transmission.

A Department of Energy report released in July 2025 warned that most U.S. regions will face “unacceptable reliability risks within five years.”

Blackouts could increase by up to 100 times by 2030.

This isn’t just a distant theoretical risk.

It’s a structural crisis unfolding in slow motion.

Washington Wakes Up

For years, the power grid didn’t get much attention from policymakers.

That’s all about to change.

The most striking signal came directly from the Oval Office.

Invoking section 303 of the Defense Production Act of 1950 - - the same wartime authority used to mobilize industrial production during national emergencies - - the President formally determined that grid infrastructure and its entire supply chain to be resources that are essential to national defense.

Transformers, transmission lines, substations, high-voltage circuit breakers, power control electronics, capacitor banks, electrical core steel.

All of it was declared critical to national security.

The determination goes further, saying that without Presidential action, U.S. industry cannot reasonably be expected to provide these capabilities in a timely manner due to limited domestic production capacity, extended procurement timelines, foreign supply dependence, and insufficient capital investment.

Read that again.

The President has formally declared that America doesn’t currently have the industrial capacity to build its own grid fast enough.

That’s a legal determination carrying the full weight of federal emergency authority.

And it opens the door to government purchases and direct financial support for domestic grid manufacturing at a scale not seen since World War II.

The Bottleneck Has Shifted

Microsoft restarted Three Mile Island.

Google is funding next-generation geothermal.

Amazon is securing nuclear contracts years before the plants come online.

The bottleneck in the AI arms race is no longer about who makes the fastest chips.

It’s about who can deliver the juice - - reliably, cleanly, and fast enough to keep the buildout from stalling entirely.

And the companies solving that problem quietly are where the next big trade lives.

Three Little-Known Stocks Worth Paying Attention To

Forgent Power Solutions is the kind of company that gets overlooked until suddenly it isn’t.

Forgent manufactures transformers, switchgear, power distribution units, automatic transfer switches, and substations.

Its the exact kind of equipment that the President specifically named as critical to national defense when invoking the Defense Production Act.

So Washington just put a spotlight on this company’s product catalog.

After bottoming near $27, the stock has quietly ground back toward $37, building the kind of base that precedes a real move.

The market hasn’t fully found this one yet.

That’s the opportunity.

SOLV Energy operates in the critical middle layer:

The engineering, procurement, and construction work that has to happen before a single megawatt flows into a data center.

Every new power facility, every substation upgrade, every transmission interconnection project requires the specialized expertise MWH brings to the table.

MWH’s services are absolutely essential.

And with the Defense Production Act invoked and federal dollars flowing into domestic grid manufacturing, firms doing the actual construction work like MWH are positioned to capture that spending directly.

Eos Energy Enterprises represents the final and most strategically important piece:

Energy storage.

Generating power is only half the problem.

Keeping a data center running with zero interruption requires energy storage that is fast, scalable, and cost-effective.

EOSE’s zinc-based battery technology is purpose-built for stationary, long-duration storage.

The Defense Production Act determination specifically calls out foreign supply dependence in grid components - - and domestically produced storage technology sits at the heart of what Washington is now mobilizing to build.

EOSE remains well below the radar of most institutional investors.

It won’t be long before they catch on.

The Gold Rush Analogy Has Never Been More Timely

During the California Gold Rush, the miners got the glory.

We’re at an identical inflection point today.

Washington has declared the grid a matter of national defense.

The DOE is deploying billions.

The biggest tech companies on Earth have signed legal commitments to fund the infrastructure they demand.

This is what a full policy mobilization looks like.

It doesn’t happen often.

But when it does, the investment implications are generational.

The grid couldn’t keep up.

Now everyone - - from the Oval Office to the server farm - - is racing to fix it.

The smartest investors will already be in position when they do.

Want Access to More Frequent Small Cap Investment Ideas Like These?

John McHugh co-founded Equity Empire with Colin Tedards to help small investors achieve financial freedom. He’s a former Wall Street analyst and hedge fund manager with more than 30 years of experience investing in small cap stocks.

To learn more about Equity Empire and get ongoing access to John’s top small cap ideas, visit:

https://myequityempire.com/performance/

This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

By John McHugh, Equity Empire Small Cap Research

There’s a moment in the history of every great technological transition when the people who understand it can barely get anyone to listen.

But those who do listen - - and, more importantly, take action - - end up securing massive stock market gains.

We’re at the start of one of those moments right now.

Big Tech is spending hundreds of billions on AI infrastructure.

The problem is that data inside today’s AI data centers is not moving at the speed of light.

Not yet, at least.

When a hyperscaler - - Amazon , Google , Microsoft , Meta - - trains or runs an AI model, it’s not using just one chip.

It’s using thousands, sometimes hundreds of thousands, all constantly passing enormous amounts of data back and forth.

For years, those connections were made of copper.

And for years, that was fine.

But copper has limits. Physics has limits.

When the bandwidth demands of next-generation AI processors started to outpace what copper could carry, something had to give.

Three Phases, Three Fortunes

The story of data center interconnect technology over the last five years can be told in three chapters.

Chapter One was the age of SerDes - - serialize/deserialize technology - - a method of packing more data onto copper cables.

The winners weren’t obvious at first.

Marvell Technologies had been developing the technology but they missed the $30 billion opportunity.

In 2008, two engineers left Marvell and founded Credo Technology Group , spending years quietly advancing a patented version of SerDes that almost nobody on Wall Street cared about.

Then, in early 2024, the hyperscalers started placing enormous orders for Credo’s SerDes technology.

Colin and I noticed this development early and recommended Credo for our subscribers at our prior firm in March 2024 at $22.50 per share.

The stock went sideways for six months.

Then revenues grew 272% in a single quarter and have been growing at a rapid clip ever since.

Our subscribers who acted on our initial recommendation scored gains of as much as 800% !

Chapter Two is happening right now.

The industry is transitioning from copper to optical - - replacing electrons with photons.

Coherent Corp , Ciena , Lumentum , and Astera Labs have soared, pricing in the inevitable wave of demand.

The opportunity is real, but much of it has already been captured by those who were early.

I provided my followers on X with a good early look at the optical interconnect opportunity:

But chasing optical now is like chasing Credo in late 2025.

Chapter Three, however, is where everything gets genuinely interesting.

And it’s where generational money will be made.

The Marvell Pivot

Chapter Three starts with Marvell - - the same company that accidentally helped the Credo growth story get started by letting its genius SerDes engineers walk out the door in 2008.

Marvell has long operated as a distant second to Broadcom , serving as a smaller but active behind-the-scenes enabler for companies building the world’s most powerful computing infrastructure.

Recently, however, the company has recently started to attract some serious attention.

This week, The Information reported that Google is in talks with Marvell to co-design its next-generation Tensor Processing Unit (TPU) AI chip, optimized specifically for AI inference.

Google has designed its own TPUs since 2015, relying primarily on Broadcom as its main design partner - - a relationship Broadcom has locked in through 2031.

But it appears that Marvell is now being pulled in as a potential third partner, focused on a memory processing unit that could reach production as early as 2027.

Marvell isn’t going to replace Broadcom as the leader in custom silicon designs for AI chips any time soon.

In fact, they probably won’t even get close.

That said, Marvell is starting to attract the attention of the AI chip industry leaders.

Nvidia invested $2 billion in Marvell in late March 2026, partnering through NVLink Fusion to integrate Marvell’s custom chips with Nvidia’s own interconnect fabric.

So now Marvell appears close to positioning itself inside both the GPU ecosystem and the custom ASIC ecosystem simultaneously.

The $5.5 Billion Bet on Light

Marvell shares have soared more than 50% since the Google rumors started swirling.

But here’s where the story takes a turn that most investors haven’t noticed.

In December 2025, Marvell acquired a company called Celestial AI for up to $5.5 billion.

Most analysts treated it as a defensive move to bolster photonic interconnect capabilities.

I believe these analysts are missing the depth of what Marvell actually acquired.

Celestial AI had built a Photonic Fabric - - a genuine architectural platform designed to enable large AI clusters that scale both within and across server racks using a high-bandwidth, low-latency, low-power optical fabric.

It delivered more than twice the power efficiency of copper at significantly higher bandwidth.

Marvell CEO Matt Murphy was clear about the company’s goal with the acquisition:

To build “the industry’s most complete connectivity platform for AI and cloud customers.”

Celestial AI wasn’t a bolt-on acquisition.

It was a statement of intent.

Marvell intends to own not just the chips that power AI, but also the photonic fabric that connects them.

And this is where a tiny company enters the story in a very large way.

A David-and-Goliath Story

POET Technologies certainly isn’t a household name.

Its market cap is a tiny fraction of Marvell’s.

Most institutional investors have never heard of it.

But for years, POET has been quietly building what could be the key to next-generation AI datacenter interconnect technology.

Much like Credo did with SerDes before the big datacenter orders came flooding in.

POET has developed its Optical Interposer - - a chip-scale manufacturing process that integrates electronic and photonic devices into a single chip using standard semiconductor wafer-level techniques.

In simpler terms, POET figured out how to put the plumbing and the wiring on the same piece of silicon.

This eliminates the costly, labor-intensive process of assembling optical components by hand.

The result is a product that is smaller, cheaper, more power-efficient, and more scalable than conventional approaches.

POET builds high-speed optical engines running at 800 gigabits per second, 1.6 terabits per second, and beyond.

Its second product line - - arguably more consequential - - is a family of light source products designed specifically for co-packaged optics (CPO), the architecture where light replaces electrons at the most fundamental level of computing infrastructure.

Here’s the critical connection:

Celestial AI’s Photonic Fabric - - the platform Marvell just paid $5.5 billion for - - was built upon POET’s Optical Interposer.

POET’s technology is not a component supplier relationship.

It’s the foundational layer on which Celestial built its entire photonic fabric.

The integration is deep.

The dependency is real.

And now that Marvell owns Celestial, POET’s relationship runs directly through a company with $8.2 billion in annual revenue, 18 hyperscaler design wins, a $2 billion Nvidia investment, and active Google TPU negotiations.

The market is just starting to see the big growth potential in POET’s relationship with Marvell.

But at the Optical Fiber Communications Conference (OFC) in Los Angeles last month, POET representatives told me that they believe the company’s growth opportunity in addition to Marvell could be substantially larger.

In September 2025, POET announced a strategic collaboration with Sivers Semiconductors to develop external light source modules.

These light sources are a critical component in any co-packaged optics architecture, solving the thermal management problem created when high-power electronics and heat-sensitive lasers are packed together.

POET demonstrated their Blazar and the POET Teralight technologies publicly at the OFC Conference.

These aren’t future product roadmap slides.

These are products in customer testing, being integrated into next-generation module designs.

And most importantly, with commercial purchase orders already secured.

What Comes Next?

None of this is a guarantee.

Co-packaged optics manufacturing is complex and standards aren’t mature.

Supply chains for advanced optical materials carry geopolitical risk.

POET itself is still a development-stage company in many respects - - serious technology, real purchase orders, meaningful execution risk.

So this is not a bet for investors with a low tolerance for risk.

But the companies that win in the next phase of a technological transition are almost never the obvious names from the previous phase.

They are the companies building the infrastructure of the next infrastructure.

Credo didn’t build the AI chip. It built the cables that made the AI chip useful.

POET isn’t building the data center switch. It’s building the light source that will make the photonic switch possible.

The custom AI silicon market will likely exceed $100 billion by 2030.

The photonic interconnect market will be similar in scale.

The transition from development stage to production is nearing - - and the opportunity to score extraordinary returns once it arrives will be enormous.

POET is positioned to be a leader in next generation datacenter interconnect technology.

Our subscribers at Equity Empire got an early look at the opportunity - - and they are already up more than 120% since my initial buy alert on POET.

POET is still a highly speculative investment.

But if the company can execute on its enormous growth potential and get its technology integrated into the hyperscalers’ supplier ecosystems, I believe that POET could be the next Credo in the coming years.

The only question is whether you will act before the crowd rushes in and drives the stock significantly higher.

Want Access to More Frequent Small Cap Investment Ideas Like These?

John McHugh co-founded Equity Empire with Colin Tedards to help small investors achieve financial freedom. He’s a former Wall Street analyst and hedge fund manager with more than 30 years of experience investing in small cap stocks.

To learn more about Equity Empire and get ongoing access to John’s top small cap ideas, visit:

https://myequityempire.com/performance/

This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

Equity Empire subscribers have already captured a 92% gain on a little known battery storage stock.

We timed the trade well.

Because the stock sold off more than 60% after we exited our position.

But now, I believe the selloff has created an exciting second chance for investors to double their money in the same stock.

In this article, I’m going to show you how.

To make the most of the opportunity, you’ll want to read this article until the end.

History Doesn’t Repeat, But It Often Rhymes

• Mark Twain

There’s a pattern that repeats itself throughout the history of technology, markets, and human behavior, and once you learn to see it, you cannot stop seeing it.

But once you see the pattern in the stock market, you can make a tremendous amount of money.

Malcolm Gladwell called this pattern the tipping point.

It’s the singular, almost invisible moment when something that has been gathering force beneath the surface suddenly breaks through and reorganizes the world above it.

The tipping point doesn’t announce itself.

It doesn’t ring a bell.

It just happens - - and then everyone agrees it was inevitable.

That moment, I believe, is approaching for zinc.

Not the zinc you find in a vitamin bottle or coating a chain-link fence.

I mean zinc as the foundational chemistry behind a battery technology that no analyst is talking about, no Reddit thread is hyping, and no CNBC segment has devoted more than thirty seconds to.

One little-known company is using this zinc battery technology to provide AI datacenters with a consistent source of power.

And the company’s stock has the potential to make forward-thinking investors big profits.

That’s precisely why you want to pay attention to this little-known company’s story.

The Overlooked Element at the Center of Everything

Eos Energy Enterprises makes a battery called the Znyth.

The name is almost comically unassuming.

So is the raw material.

Zinc is the forty-second most common element on Earth.

It’s abundant, domestically sourced, non-flammable, and completely free of the geopolitical entanglements that have made lithium-ion supply chains one of the defining anxieties of the 21st century.

The Znyth battery’s architecture is deceptively elegant.

At its core, the system uses a zinc-bromine aqueous electrolyte chemistry - - water-based, not flammable - - cycling between charge and discharge states with no meaningful degradation over thousands of cycles.

Unlike lithium-ion cells, which lose capacity as the active materials inside them physically expand and contract during charging, zinc-bromine chemistry operates in solution.

The electrodes don’t degrade the same way.

That’s not a minor distinction.

That’s the difference between a battery that performs reliably over a ten-year project finance horizon and one that doesn’t.

The Indensity platform - - Eos’s newest modular system unveiled in January 2026 - - takes this chemistry and reimagines its physical form.

Instead of sprawling ground-level installations requiring vast acreage and crane-dependent maintenance, Indensity stacks twelve units high in a containerized format.

The footprint fits inside an urban power substation.

It can be serviced by a standard forklift, not specialized equipment.

It can also be deployed in dense neighborhoods where a lithium-ion installation would never survive a zoning permit hearing.

Because lithium burns, and zinc doesn’t.

The system is engineered for eight hours or more of continuous discharge - - what the industry calls long-duration energy storage, or LDES.

Lithium-ion batteries, for all their dominance in electric vehicles and consumer electronics, were not designed for this use case. They excel at short bursts.

The grid doesn’t need short bursts. The grid needs eight hours of overnight storage to smooth out the solar generation curve. It needs twelve hours of backup for a data center that cannot go dark for a single second.

It needs, in short, exactly what the Znyth battery was built to provide.

The Tipping Point Was Always Coming. Nobody Paid Attention.

The conditions have been accumulating for years.

China controls the dominant share of global lithium refining.

The Democratic Republic of Congo supplies the majority of the world’s cobalt under conditions that have attracted sustained human rights scrutiny.

Rare earth dependencies, forged over a decade of supply chain decisions made when geopolitical risk felt abstract, are now very real.

And now tariffs are arriving, sharp and fast, on lithium-ion batteries manufactured outside the United States.

Meanwhile, the grid is straining.

And soon it may not be able to meet the insatiable energy needs of the nation’s AI datacenter operators.

AI data centers consume electricity the way a desert absorbs rain - - relentlessly, completely, with no patience for shortages.

Amazon , Microsoft , and Google are each committing hundreds of billions of dollars to data center infrastructure buildout at a pace that has no modern precedent.

These hyperscalers are not price-sensitive about energy storage.

They’re delivery-sensitive. They’re reliability-sensitive.

And they’re location-sensitive - - because not every data center gets built in a Nevada desert where there’s room for a lithium farm.

Some get built in cities. In dense suburbs. In places where your storage solution needs to be compact, stackable, and non-flammable.

Because of the versatility of Eos batteries, the company’s technology is attracting a wave of new potential buyers.

Data center leads at Eos grew 50% quarter-over-quarter in Q4 2025.

The active data center pipeline grew 40%.

Twenty-two percent of the company’s $23.6 billion total pipeline is now tied directly to AI infrastructure buildout.

The Department of Energy found Eos’s technology compelling enough to write a $90.9 million loan guarantee.

Three major U.S. utilities have deployed it in the field.

The Department of Defense is running a system at a national laboratory.

These are not promotional partnerships.

These are operating deployments, with contractual obligations and performance requirements attached.

The Miss, The Confession, and What Came Next

Despite Eos’ tremendous $23.6 billion pipeline of potential deals, the company recently experienced its first real growing pains.

Eos missed its 2025 revenue guidance. They had told investors to expect $150 to $190 million. They delivered $114 million.

The stock tumbled.

At its low point just a few weeks ago, EOSE shares had dropped as much as 75% from their January 2026 highs.

But here’s what happened next, and why it matters to opportunistic investors like you.

CEO Joe Mastrangelo stood in front of analysts and said something that CEOs almost never say:

‘The bottom line is we missed our guidance, and that falls on me.‘

No macroeconomic hedging. No industry cycle deflection.

He named three specific root causes - - a supplier that went dark for a week, bipolar automation equipment that couldn’t hit quality targets, and battery line downtime running at 35% when world-class operations run at 10%.

He explained exactly what was being done to fix each one.

And then he put his money where his mouth was and bought a big chunk of EOSE stock.

A month later, on April 9th, 2026, Eos released its Q1 operational update.

Shipments were up 17%. Battery output was up 10.4%. Bipolar automation yields - - the metric that measures how many parts come off the line correctly the first time, without rework - - were up 22% sequentially. Shipments were tracking in line with customer contractual commitments.

That last sentence is promising news for EOSE investors.

It’s the answer to the question every serious investor has been holding since February:

Is the $701 million backlog, the 2.8 gigawatt-hours of contracted business, going to be delivered?

On April 9th, the answer was looking like a yes.

The Second Production Line, and Why It Changes the Math

Eos’s second production line - - Line 2 - - has cleared factory acceptance testing and is scheduled for initial production by the end of June.

It was engineered with a single-piece flow design that cuts internal material travel distance by 86% and eliminates the production redundancy that caused the downtime issues in Q4.

When it comes online, Eos doubles its manufacturing capacity and removes the single largest operational risk still hanging over the company.

Full year 2026 revenue guidance is $300 to $400 million.

Q1 alone is tracking $56 to $57 million.

After a brief scare, the math is starting to feel achievable again.

The stock closed recently just above $7, up nearly 30% on the Q1 news.

In January it was at $19.86. Six weeks ago it was at $4.37.

At Equity Empire, we rode this roller coaster once - - and got off the ride with a 91% gain.

We’re now looking at another entry point that’s lower than our first.

That’s even after the company has significantly improved its financial structure with a massive billion dollar equity and convertible note financing late last year.

I believe EOSE can return to its January highs near $20 over the coming 12 to 18 months.

For the investor with patience, conviction, and a tolerance for volatility — this is the bet.

The Bottom Line

EOSE is undeniably speculative.

The company is not profitable. The losses are real.

The history of guidance misses is real and not to be dismissed.

Execution risk remains.

But the demand is real. The technology is real. The government backing is real. The pipeline is not only real, but also massive.

And it looks like the operational recovery - - evidenced across four simultaneous production records on April 9th - - is under way.

The conditions have been accumulating invisibly: the right product, the right market moment, the right operational fix.

But something has shifted, and the company’s growth story will soon become more obvious to the investing public.

Smart investors will want to beat the crowd.

Eos Energy is on the verge of a massive growth cycle over the coming years.

At Equity Empire, we booked a 92% gain on EOSE shares once.

This time, I believe we’re going to score even greater gains.

The question is whether you’re going to join us.

John McHugh co-founded Equity Empire with Colin Tedards to help small investors achieve financial freedom. He’s a former Wall Street analyst and hedge fund manager with more than 30 years of experience investing in small cap stocks. This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

To learn more about Equity Empire, visit:

myequityempire.com

SpaceX is issuing shares to the investing public.

Elon Musk confirmed it.

Goldman Sachs and Morgan Stanley are running the books. The S-1 for the IPO is reportedly being filed in the coming weeks, targeting a mid-June listing at a valuation approaching $1.75 trillion.

That would make it the largest IPO in the history of capital markets.

Not just the largest tech IPO.

The coming SpaceX IPO could be the biggest event in the history of capitalism.

And most investors are thinking about this completely wrong.

They’re focused on whether they can get an allocation in SpaceX itself.

That’s not the trade.

The trade is what happens to the rest of the space industry before SpaceX goes public.

Here’s why.

The AWS Moment Nobody Sees Coming

When Amazon Web Services (AWS) went from a footnote in earnings calls to a clearly defined, clearly profitable business unit around 2013-2014, something interesting happened.

Investors didn’t just buy Amazon .

They repriced every software company on Earth.

Salesforce . ServiceNow . Twilio . Shopify . Snowflake . DigitalOcean . Atlassian .

Once the investment community saw the economic potential of AWS, these companies’ valuations soared because investors finally had a mental model for what cloud infrastructure was worth at scale.

And the run lasted more than a decade.

Here’s how the First Trust Cloud Computing ETF performed in the ten year period 2012-2022:

The SpaceX IPO is the AWS Moment for Satellite Infrastructure Companies.

When SpaceX files its S-1, portfolio managers who have never modeled a satellite business will suddenly need one.

Space ETFs will see inflows - - much like the cloud software ETFs did in 2012.

Analysts at major banks will build full coverage universes overnight.

The space sector will get a spotlight it’s never had.

The window between now and mid-June is the window where the smaller names in the space industry are still relatively undiscovered.

That window is closing fast.

Here’s where I’d focus:

Planet Labs : The Daily Imaging Machine

Planet Labs operates the world’s largest Earth observation satellite constellation. Every day, it photographs the entire landmass of Earth.

For years, this was an impressive technology in search of a sustainable business. Then something changed.

In Q3 of fiscal year 2026, Planet reported $81.3 million in revenue - - up 33% year-over-year. Adjusted EBITDA swung positive. Free cash flow turned positive. And the company disclosed a remaining performance obligation backlog of $672 million, representing 361% year-over-year growth.

That means customers aren’t just buying Planet’s data. They’re signing long-duration contracts, locking in future revenue streams. Defense agencies. NATO. The German government. Insurance companies. Commodity traders.

The company also sits on $677 million in cash. The survival question, which haunted Planet for most of its public life, has been answered.

When SpaceX goes public and institutional investors build their space frameworks, Planet is one of the first names they encounter.

It’s the broadest, most established pure-play in Earth observation. The AI-in-orbit work it’s doing with Google and NVIDIA positions it as a fellow traveler in the emerging space-as-infrastructure narrative.

That’s a potential catalyst for a higher share price.

BlackSky Technology : The Tactical Intelligence Layer

If Planet is the wide lens scanning the globe, BlackSky is the tip of the spear.

BlackSky’s business is high-revisit imagery of specific, high-value locations. Its customers aren’t scientists. They’re intelligence professionals and military planners who need to know whether something changed in the last four hours, not the last twenty-four.

BlackSky’s third-generation satellite platform delivers sub-50 centimeter resolution - - the ability to distinguish individual vehicles, identify aircraft on runways, detect camouflage netting over equipment. The first Gen-3 achieved on-orbit operations within hours of deployment. By March 2026, BlackSky had commissioned its fourth.

International customers now represent more than half of BlackSky’s total revenues, having grown over 50% year-over-year. Its contracted backlog sits at $345 million - - up 32% - - and approximately 91% of that backlog comes from international contracts.

That’s not a detail. That’s a thesis.

Governments that previously couldn’t access this class of intelligence without relying on US intelligence-sharing arrangements can now simply buy it. BlackSky is selling sovereignty. The ability to see your own borders, your own threats, your own neighbors - - without asking permission from Washington.

Management guided 2026 revenue to $120-145 million, roughly 24% growth at the midpoint. The risks are real - - net losses widened in 2025 as Gen-3 deployment costs ran through the income statement. But the backlog is growing, the hardware is generationally better, and the international demand story is just getting started.

York Space Systems : The Satellite Factory

Most people outside the defense space industry have never heard of York Space Systems.

That’s about to change.

York’s strategy is simple: they build satellites the way Toyota builds cars.

That’s not just marketing copy. It’s a business model.

Where traditional defense primes take 30 months to build a satellite, York delivered one program from contract signature to orbit in seven months.

Where competitors charge for specialized engineering on every unit, York builds on standardized, modular platforms using common components and a production line that can scale to 1,000 satellites annually.

The foundation is the the US military’s next-generation missile warning and tracking constellation, now also known as Golden Dome.

In 2025, York became the first prime contractor to complete an on-orbit delivery under the program, deploying 21 satellites. All made contact within hours of separation.

Revenue grew 52% in 2025 to $386.2 million.

Management guides to $545-595 million in 2026 - - roughly 48% growth - - with positive adjusted EBITDA expected for the first time.

Over 70% of projected 2026 revenue comes from existing funded backlog.

When SpaceX files and investors start building space infrastructure frameworks, York is one of the first names they’ll encounter as a public pure-play in defense satellite manufacturing. It shares SpaceX’s industrial manufacturing plan: build at scale, faster and cheaper than the legacy primes.

Institutional investors who understand the SpaceX story will understand the York story immediately.

Spire Global : The Transformation Play

Here’s a sentence that doesn’t appear in most investor materials.

Spire Global’s satellite constellation can listen to the atmosphere.

Not photograph it. Listen to it.

When GPS signals pass through the Earth’s atmosphere on their way to Spire’s receivers, they bend in predictable ways based on temperature, humidity, and pressure.

Spire collects those bending patterns from orbit and sells the resulting data to weather forecasting agencies, aviation operators, and governments as one of the most accurate atmospheric sensing datasets available anywhere on Earth.

This is unglamorous.

It’s also potentially irreplaceable.

2025 was a year of deliberate transformation.

Spire divested its maritime tracking business in April, generating cash and sharpening focus on its highest-margin segments: weather, aviation, and Space Services.

Strip away the maritime noise and the continuing business posted 44% year-over-year growth in Q4 2025.

Management guides to $75-85 million in 2026, representing over 50% growth in the continuing businesses.

In late 2025, Spire secured a position on the US Missile Defense Agency’s SHIELD IDIQ contract, which carries a shared ceiling of $151 billion.

That’s a huge number. But Spire has to compete with a long list of much larger competitors to get a slice of this giant pie.

Space reconnaissance, atmospheric sensing, RF data collection - - these are capabilities with obvious defense intelligence value.

Spire’s remaining performance obligations exceed $200 million, with approximately $70 million expected to convert in 2026. The balance sheet is debt-free with $81.8 million in cash.

Spire’s atmospheric sensing data isn’t challenged by SpaceX. Its defense positioning is genuinely complementary.

And it’s trading at a significant discount to larger competitors like Planet while growing its continuing business even faster at 50%.

Satellogic : The High-Risk, High-Reward Underdog

Let me be direct about something upfront.

Satellogic is not competing on the same financial playing field as the other companies in this report. With $17.7 million in 2025 revenues, it is an early-stage business making a bet that high-resolution, affordable satellite imagery - - offered on a subscription basis at prices that were once accessible only to superpowers - - will find enough paying customers before the cash runs out.

For most of its public life, that bet looked shaky.

Then something shifted in the back half of 2025.

Q4 revenue grew 94% year-over-year to $6.2 million - - the largest quarterly haul in company history. Full-year revenue grew 38% to $17.7 million. The EBITDA loss shrank by 48%. And Satellogic completed a $90 million public offering in October 2025 followed by a $35 million registered direct offering in January 2026, ending the year with $94.4 million in cash.

For a company that spent several years flirting with the survival question, that balance sheet is oxygen.

Now the company is deploying that capital into the Merlin constellation - - a next-generation platform designed to deliver sub-meter resolution at commercially competitive price points. Management describes Merlin as fully funded. If that holds, Satellogic enters a new competitive tier: the class of satellite operators that can win contracts Planet and BlackSky go after.

Here’s the honest outlook.

Satellogic has to execute on its growth potential. The revenue base is too small, the competitive environment too brutal, and the company has a history of guiding and missing. Planet has a dramatically larger constellation. BlackSky has deeper defense relationships. New entrants keep arriving.

But in a world where satellite imagery has become a commodity - - where the buyers are no longer just superpower governments but city planners, insurance actuaries, logistics companies, and commodity traders - - the market is large enough for multiple operators. Satellogic is building infrastructure for that world at a price point that could prove genuinely disruptive.

If the Merlin constellation launches on schedule, if the revenue ramp materializes, and if the SpaceX IPO floods the sector with institutional attention, the current valuation could look backward-looking twelve months from now.

That’s a lot of ifs. But that’s what early-stage investing looks like.

Satellogic is the highest-risk name on this list.

But the opportunities with the biggest reward potential often also carry the biggest risks.

A Look at Different Scenarios

Let me say what the bulls will tell you, and also what the bears will tell you.

The bulls are right that a $1.75 trillion SpaceX IPO is the single most powerful legitimizing event the commercial space industry has ever experienced.

Every institutional investor on Earth will need a view on space.

The sector will get repriced.

The bears are also right that SpaceX’s valuation will dwarf every other public space company combined.

When investors allocate to “space,” a meaningful percentage of every dollar will flow directly to SpaceX - - potentially at the expense of the smaller names.

That’s how every mega-cap IPO has worked.

The reality lives between the scenarios.

Planet and Spire benefit most from the rising tide - - they’re the names institutional investors encounter first when building space frameworks.

BlackSky’s tactical intelligence positioning doesn’t overlap with SpaceX’s commercial broadband focus, and 91% international defense backlog means demand is driven by geopolitical necessity, not investor sentiment.

York is the most straightforward case - - a defense satellite manufacturer with PWSA contracts that directly serve the same Golden Dome priorities powering SpaceX’s Starshield business.

Satellogic is the high risk, high reward early-stage speculation.

The SpaceX IPO won’t just be a capital markets event.

It will be a catalyst for money to pour into the entire space sector.

Every institutional investor who reads that S-1 will come out of it knowing more about the commercial satellite industry than when they started.

And they’ll realize that they need exposure to it.

Investors in the SpaceX IPO may make money.

But the investors who build positions in Planet, BlackSky, York, Spire, and Satellogic before the SpaceX IPO captures the headlines may be the ones who look smartest in the months ahead.

The window is open.

It won’t be open long.

John McHugh co-founded Equity Empire with Colin Tedards to help small investors achieve financial freedom. He’s a former Wall Street analyst and hedge fund manager with more than 30 years of experience investing in small cap stocks. This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

To learn more about Equity Empire, visit:

myequityempire.com

Even Jim Cramer was tweeting about it this weekend.

Some believe this might cause the stock market to completely unravel.

But if you’re like me … you’re not totally sure what’s actually happening. You just know it sounds bad.

Well, I’ve spent the last few weeks pulling SEC filings, reading short seller reports, and tracing where the money actually flows.

And what I found is shocking.

It’s a system that works in layers in order to make executives rich.

And once I show you how it works, the headlines are going to make a lot more sense.

Because here’s the thing. The architecture behind private credit isn’t new.

In fact we’ve seen it unravel before.

In 2008, the private credit machine broke and it took the global economy down with it.

And that same architecture is running again.

But this time not with mortgages. But With private credit.

And the cracks you’ve been hearing about are already showing.

To understand what’s happening right now in private credit, you have to understand how the machine works.

Because it works in layers and each layer has a job.

Layer One: The Originators

Let’s begin with Layer One: the originators.

These are Companies like SoFi Upstart and LendingClub

These are fintech lenders. They often use AI-driven underwriting to quickly and cheaply approve personal loans.

This is unsecured debt, no collateral required. It’s literally just SoFi depositing money into your account with the idea you’ll pay them back.

SoFi alone grew personal loan originations by 66% over the past year.

And Their charge-off rate? … Under 3%.

At least, that’s what they told investors. More on that in a minute.

Wall Street loved the story. SoFi’s stock went parabolic.

The narrative was simple: technology made lending smarter, default rates stayed low, and money was being made.

But here’s the thing about fintech lenders:.

They don’t end up holding most of these loans. They originate them, collect a fee, and sell them.

Layer Two: The Packagers

Which brings us to layer two: the packagers.

Companies like Blue Owl Capital you may have heard of them. Blue Owl runs funds that buy these loans from the fintechs like SoFi.

Blue Owl then pools the loans together, slices them into tranches, and sell shares to investors.

One of Blue Owl’s funds, their Alternative Credit Fund, has over a billion dollars in assets.

I spent the weekend reading the entire SEC filing. Nearly half the portfolio is consumer personal loans.

Now here’s where you - the investor comes into the picture.

Layer Three: The Investors

That’s Layer three: the retail investors.

The people buying shares in these Blue Owl funds, chasing dividend yield, trusting that the managers know what they’re doing.

In the case of Blue Owl, if the loans they bought from SoFi start to go bad at a higher rate than was expected,

The entire fund will lose value.

This should sound familiar

Because In 2006 and 2007, Countrywide originated mortgages.

Lehman and Bear Stearns packaged them into something called CDOs.

Pension funds, investors, and insurance companies bought the paper.

But when the mortgages started to go bad the entire system collapsed.

And just like last time - the cracks didn’t start at the top. They started at the bottom.

On March 17th, Muddy Waters Research, one of the most well-known short sellers on the planet, dropped a 28-page report on SoFi.

The title: “A Financial Engineering Treadmill Leaving Management Fat, Shareholders the Biggest Losers.”

The allegations were specific.

Muddy Waters claims SoFi’s actual loss rate on personal loans is 6.1% more than double the 2.89% the company reports.

They allege that SoFi is making its loan portfolio appear to be worth two billion dollars more than it actually is.

And that profit number Wall Street keeps celebrating at SoFi …

Muddy Waters says the real number is much lower.

If that’s true - shares of SoFi should be worth far less.

But here’s the part nobody’s connecting.

If even a portion of what Muddy Waters alleges is true

If the loss rates on these personal loans is meaningfully higher than what SoFi claims

Then the assets they’ve sold to Blue Owl funds are worth far less than what’s being reported.

So I want to zoom out now, because this is important.

First … what’s happening at Blue Owl right now?

Their stock has crashed over 40% this year alone.

Short sellers betting against Blue Owl just hit an all-time high.

And in February, the company did something that sent shockwaves through the entire private credit industry.

They permanently cut investors off from selling out of their fund.

Investors saw what was happening and wanted their money back

Withdrawal requests surged 200% … and Blue Owl said no.

Let me put that in perspective.

Imagine you put money into an investment.

You were told you could get it back periodically. And then one day the manager of the fund just says no.

That’s what happened.

Now, this is not a normal event.

It’s a bank run in slow motion - except Blue Owl is not a bank.

It’s a private credit fund. And there’s no government backstop. No FDIC insurance here.

Investors panicked.

Now this is where most people stop paying attention.

The headlines sound scary, Blue Owl, Sofi, and Upstart stocks are tanking, and the instinct is to look the other way.

But this is exactly when the story gets interesting.

It’s called Level 3 accounting. And nearly all of Blue Owl’s funds fall into this category.

Let me explain what that means because it’s actually pretty simple.

Imagine you own a house.

If your house is in a neighborhood where homes sell every week, you - or your real estate agent - knows exactly what it’s worth.

You look at what your neighbor’s house sold for, and you’ve got a rough idea what your house would sell for.

That’s Level 1 - there’s a real market price.

Now imagine your house is in a quieter neighborhood, where the homes are all custom.

It’s harder to find a comp - but not impossible. You can still get a pretty good estimate.

That’s Level 2.

But what if your house is completely unique? One of a kind. Nothing like it has ever sold.

There are no comparables anywhere. In this scenario - YOU get to decide what it’s worth.

You pick a number you think your house is worth and you list it for sale.

Nobody can tell you you’re wrong because there’s really nothing to compare it to.

That’s Level 3.

And that is what Blue Owl is doing with over a billion dollars worth of consumer loans inside this fund.

These loans don’t trade on any exchange. There’s no stock ticker. There’s no daily price.

Blue Owl’s own internal models decide what they’re worth.

They pick the assumptions - things like how many borrowers they think will default

And the assumptions Blue Owl makes drives the entire valuation of the fund.

Now connect that back to what Muddy Waters just told us.

If SoFi’s actual default rates are double what’s being reported — but Blue Owl is using those lower numbers in their models — then the math doesn’t work.

The loans are worth way less than Blue Owl says they are.

And here’s why that matters way beyond just one fund.

If the assumptions about the loans are wrong, the value of these Blue Owl funds are wrong.

If the values are wrong, the total assets Blue Owl manages are way overstated.

Now, Why would Blue Owl want to overstate the value of the assets they hold?

Blue Owl collects fees on those assets.

And the higher the assets - the more fees the company collects!

This is where the story gets really uncomfortable.

In its first year of operation, Blue Owl’s private credit fund is already losing money.

Investors who wanted to sell their shares had to do it below the original price.

Think about that … sophisticated investors are already trying to get out at a loss after less than a year.

But here’s one that should bother every investor listening to this.

Blue Owl is paying dividends to the investors in this fund. Sounds great, right?

Except a portion of those dividends aren’t coming from profits the fund earned.

Instead, they’re coming from the money investors put in.

Blue Owl is taking a piece of your original investment, handing it back to you, and calling it a dividend.

That’s not a return ON your money.

That’s a return OF your money.

And that’s not what investors signed up for.

Now I want to be clear about something.

I am not calling this the next Lehman Brothers. The scale isn’t the same here.

Personal loans are not mortgages … but in some ways they are worse because at least a mortgage has a home sitting somewhere as collateral.

But the lesson from 2008 was never really about mortgages.

It was about what happens when you build a system where the people making the loans aren’t the ones holding the risk.

Where the people in the middle get to decide what those loans are worth.

And where the people at the top collect fees based on those valuations - whether the numbers are real or not.

That is the system running right now in private credit.

SoFi and Upstart make the loans - and sell them.

Blue Owl buys them, marks them at whatever their models say, and collects fees on those marks.

And retail investors, people like you and me, are the ones left holding the bag if the music stops.

The good news, investors aren’t letting these greedy executives off the hook.

Bets against Blue Owl at all-time highs. Shares of SoFi and Upstart have tanked.

The biggest news networks in the world are asking whether private credit is the next crisis.

Morgan Stanley is warning that default rates could quadruple.

The headlines are finally catching up. And now you know what’s behind them.

Follow Colin On X: https://x.com/InvestorVideos

Subscribe On YouTube: https://www.youtube.com/@EquityEmpireResearch

Imagine if Google Maps took a new photo of every inch of Earth’s surface.

Every. Single. Day.

Now imagine those photos were so sharp you could spot a military vehicle, identify a crop disease, or track a cargo ship - - all before it ever made the news.

That’s not science fiction.

That’s Planet Labs.

And most investors don’t even know the company exists.

Here’s the setup.

Planet Labs $PL operates the largest commercial constellation of Earth-imaging satellites ever built - - over 200 spacecraft circling the globe right now, capturing a complete image of every landmass on Earth every 24 hours.

Think about that for a second.

The entire surface of Earth. Every day.

Let me put that in perspective:

The old way of doing satellite imagery was expensive, slow, and one-directional.

You’d commission a satellite to point at a specific target.

You’d wait.

Then you’d hope it wasn’t cloudy.

Planet flipped that model on its head.

Instead of expensive, one-off satellites, they built cheap, modular “Dove” microsatellites - - shoebox-sized, mass-produced like smartphones, and launched in swarms.

The result?

A continuous, real-time picture of the planet. No waiting. No gaps.

It’s like the difference between hiring a private detective to tail someone once versus installing a camera that never blinks.

But here’s where it gets interesting for investors:

For years, Planet was the classic “incredible technology, terrible business” story.

Revenue was growing, but so were the losses.

The stock went from $10 to under $3.

Analysts shrugged and retail investors moved on.

Then something changed.

Governments woke up.

Phase One of Planet’s Transformation: Defense Becomes the Growth Engine

In fiscal year 2026 (ended January 2026), Planet’s defense and intelligence segment became its single largest revenue driver - - generating $41.9 million in a single quarter, up 41% year-over-year.

That’s not a blip. That’s a structural shift.

Consider the contracts stacking up:

• A €240 million multi-year deal with the German government

• A contract renewal with NATO

• A partnership with the European Space Agency

• A $13.5 million NASA task order

• Renewed data supply to the U.S. Navy

These aren’t pilot programs.

These are long-term, recurring commitments from institutions that don’t switch vendors lightly.

And Planet is uniquely positioned to win them - - because nobody else has the 10+ year global archive Planet has built.

That archive is the moat. And it can’t be replicated.

Phase Two: The AI Multiplier Nobody Is Pricing In

Here’s the thing about satellite imagery.

Raw images are useful.

But AI-processed insights are transformational.

This is where Planet is quietly building something extraordinary.

Think of it this way:

Raw satellite data is like having the world’s largest library of books - - in a language nobody speaks yet.

AI is the translator.

Planet is now partnered with both Alphabet $GOOGL and NVIDIA $NVDA to build what amounts to a “GPU-native AI engine for planetary intelligence.”

The idea: instead of selling images, they sell decisions.

Instead of sending a defense analyst a photo of an airfield, Planet’s system flags the anomaly automatically, cross-references it with historical data, and delivers an alert.

The Pelican satellites - - Planet’s next-gen high-resolution constellation - - are also equipped with NVIDIA Jetson AI chips onboard, meaning the satellite itself can identify a ship, a vehicle, or a construction project before the data even hits the ground.

That’s not just a satellite company. That’s an intelligence platform.

And here’s the kicker:

Planet has also partnered with Anthropic (the company behind Claude) to make these insights accessible to non-experts.

Think of it as giving every government analyst, hedge fund researcher, or supply chain manager the ability to ask a question and get a satellite-backed answer in seconds.

The market they’re targeting?

Over $100 billion in defense and geospatial intelligence spending.

Planet’s numbers tell a compelling story.

For the full fiscal year 2026, the company delivered:

• $307.7 million in revenue - - a record, up 26% year-over-year

• $900 million in backlog - - up a staggering 79% year-over-year

• Its first-ever full year of adjusted EBITDA profitability

• 98% recurring annual contract value - - meaning almost all revenue renews automatically

That last number deserves a moment.

98% recurring.

That’s a SaaS business metric hiding inside a space company.

And that $900 million backlog?

It represents contracted future revenue that is already locked in.

Planet’s business is building visibility in a way that early-stage growth companies rarely achieve.

Phase Three: Data Centers in Space

If everything above sounds exciting, this part might sound a little crazy.

But stay with me.

Planet has announced an R&D partnership with Google called Project Suncatcher - - a project to explore the concept of data centers in space.

The goal is to launch two prototype satellites equipped with Google’s custom TPU (Tensor Processing Unit) AI chips by early 2027.

The implication?

Planet isn’t just collecting Earth data.

It’s exploring whether the next frontier of AI compute infrastructure could operate in orbit - - where solar energy is abundant, heat dissipation is easier, and latency to global customers could be radically reduced.

This is still years from commercialization. But it signals something important about where this company is heading.

Planet isn’t thinking like a satellite company.

It’s thinking like a platform company with satellites as the infrastructure layer.

The Risk: Is This A Defense Bubble?

Let’s be honest about the bear case.

Planet’s explosive growth is heavily tied to government and defense contracts.

That’s a double-edged sword.

Government spending is subject to budget cycles, political shifts, and procurement delays.

A single contract cancellation can torpedo quarterly numbers.

The company is also still posting significant GAAP net losses - - $152.5 million in Q4 FY2026 alone (though a large chunk of that was non-cash warrant revaluation from stock price appreciation, which is actually a good problem to have).

And at roughly 28x enterprise value to revenue, the stock is priced for a perfect execution.

The market is expecting growth to continue accelerating. Any stumble will be punished.

The Bottom Line

Here’s the pattern that keeps repeating in tech investing:

A company builds a unique infrastructure layer. The market ignores it for years because it’s losing money. Then the infrastructure becomes critical - - and suddenly everyone needs it.

We saw it with Amazon’s AWS.

We saw it with Palantir .

And now we’re watching it play out with Planet Labs.

The combinations of the factors listed below creates a setup that looks very different from the broken growth story it appeared to be just 18 months ago:

• An irreplaceable data archive with a decade-long head start

• A defense and government customer base committing to multi-year contracts

• A $900 million backlog growing 79% annually

• AI processing partnerships with Google, NVIDIA, and Anthropic

• A first-ever move toward EBITDA profitability

Winning Investors Stay Ahead of the Crowd

I’ve been following Planet’s progress since it became a publicly traded company in late 2021.

Last year, I highlighted Planet’s promising stock chart formation on the X platform when the stock was trading at $6.51 :

And since my post on Planet nine months ago, the stock has soared 507% !

Now, after seeing Planet hit new highs when most stocks are rolling over, the investment community is starting to pay attention.

Most investors missed the big move over the past year.

But if the AI-powered intelligence platform thesis plays out - - and the defense and sovereign space tailwinds continue - - Planet could continue its powerful climb.

The eye in the sky is open.

The question is whether investors are watching.

This Little-Known Satellite Stock Looks Like Planet Labs Did 18 Months Ago

If you you missed my post on X before Planet Labs stock rocketed 500%+, there’s no need to worry.

The stock market is stocked with new profit opportunities.

And you have another chance to achieve gains similar to Planet.

That’s because a setup is forming in another under-the-radar space stock that looks eerily familiar.

You just read about Planet.

The 500% run from the lows.

The $900 million backlog. The defense contracts piling up. The AI partnerships with Google and NVIDIA.

Here’s the question you should be asking yourself right now:

What does the next Planet look like - - before it makes its big move?

Because I think I’ve found it.

The Pattern Is Repeating

Think about what made Planet such a compelling setup 18 months ago.

It had a unique, hard-to-replicate technology that most investors didn’t understand. It was losing money, so Wall Street ignored it. Then governments started signing long-term contracts. The backlog exploded. The stock followed.

That exact pattern is forming again.

In a different satellite company.

With a different technology.

But the same DNA.

Here’s what I can tell you: this company doesn’t take pictures of Earth.

It listens to it.

While Planet’s satellites capture light, this company’s constellation captures radio signals - - from ships, aircraft, weather systems, radar installations, and communication infrastructure - - bouncing them off the atmosphere to extract intelligence that cameras simply cannot see.

Think about that for a second.

A camera can’t see through clouds.

A camera can’t detect a radar system powering up 500 miles away.

A camera can’t measure atmospheric conditions at 30,000 feet over the middle of the Pacific Ocean.

This technology can.

And right now, the U.S. military, NOAA, NATO, and a growing list of allied governments are starting to realize they can’t live without it.

The Numbers Are Starting to Turn Higher

For years, this company was in the same boat as Planet - - extraordinary technology, messy financials, ignored by the market.

Then something changed.

It made a bold strategic decision: it sold off a major division to focus exclusively on its highest-value, highest-margin business.

The result?

44% revenue growth in its most recent quarter in its core business.

Management’s revenue guidance for the core business in 2026 calling for over 50% growth.

75% of next year’s projected revenue already contracted and signed.

A path to EBITDA breakeven by year-end.

And a balance sheet that is completely debt-free.

Sound familiar?

This is almost exactly where Planet was before the market woke up.

The Defense Catalyst

Planet’s re-rating came when defense and intelligence agencies started writing big checks.

The same thing is happening here.

This company has a technology that can track radio frequency emissions from military hardware - - day or night, rain or shine, anywhere on the planet - - and deliver the intelligence in near real-time.

Governments aren’t just buying a subscription.

They’re building national security infrastructure around this data.

And here’s the part that makes this really interesting:

The company’s ability to produce this intelligence is about to increase 15-fold over the next year as new satellite clusters come online.

15x the data production.

With the contracts already in place to buy it.

Why Most Investors Will Miss This (but you don’t have to)

The stock is tiny.

This is not a household name. It’s not covered on CNBC. It’s not in any major ETF with meaningful weight.

And that’s exactly the point.

The gains in Planet came before the crowd arrived.

The gains in most of my best picks come before the crowd arives.

Camtek climbed 244% before the crowd arrived.

Credo soared 885% before the crowd arrived.

Applied Optoelectronics ran 126% before the crowd arrived.

The pattern is always the same:

Unique technology → market adoption → backlog inflection → stock re-rating.

We are sitting at step three right now.

The Full Research Report

I’ve put together a complete deep-dive on this company - - the technology, the financials, the key catalysts, the risks, and the price target.

In the report, I describe the company’s unique growth story much like I did with my past reports on CAMT, CRDO, and AAOI.

If you want access to the new full report - - including the ticker -- it’s available right now to Equity Empire subscribers.

Get the full report at myequityempire.com

The trend-following crowd will find this stock eventually.

The question is whether you’ll be in it before they do.

John McHugh cofounded Equity Empire with Colin Tedards to educate investors and help them achieve financial freedom. He’s a former Wall Street analyst and hedge fund manager with more than 30 years of experience investing in small cap stocks.

This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

Something remarkable happened on March 4, 2026.

And it has created a big profit opportunity for savvy small cap investors.

The CEOs of Amazon, Google, Meta, Microsoft, Oracle, OpenAI, and xAI walked into the White House and signed a pledge.

Not a press release.

Not a blog post.

A formal commitment, witnessed by the President of the United States, that Big Tech will build, fund, or buy every watt of power their AI data centers consume.

So American families don’t see their electric bills explode.

Trump called it “a colossal expansion of U.S. energy.”

Energy Secretary Chris Wright said these companies are smart, powerful, and strong - - and now they’re on record to prove it.

Here’s what most investors missed:

The same day that pledge was signed, Babcock & Wilcox received full notice to proceed on a $2.4 billion contract to deliver 1.2 gigawatts of power for AI factory campuses.

That is not a coincidence. That’s a roadmap.

Why the Pledge Changes Everything for BW

Before March 4, Big Tech’s power strategy was a patchwork of utility negotiations, grid connection requests, and years-long permitting battles.

The Ratepayer Protection Pledge blows that up.

These companies are now on the hook - - publicly, politically - - to own or directly fund their power generation.

That means they need Engineering, Procurement, and Construction (EPC) partners who can actually build power plants.

Fast.

At scale.

And with a track record that justifies writing a billion-dollar check.

The list of companies that can do all three is very, very short.

Babcock & Wilcox is one of these rare companies.

They’ve been manufacturing power generation equipment since 1867.

BW has more than 400 gigawatts of installed generating capacity around the world.

Thousands of boilers designed and installed.

Decades of large-scale project management experience.

And critically unlike GE Vernova - - which has a four-to-five year backlog for datacenter turbines - - B&W has the manufacturing capacity to move now.

Why the Story Is Still Being Written

I know what you’re thinking.

BW has already had a massive run. Did we miss it?

BW has certainly been a strong performer over the past several months.

And our subscribers at Equity Empire who acted on my buy alert in late November are already sitting on some fat profits.

But here’s why the opportunity is still live:

• The $12 billion pipeline is real. BW has identified over $12 billion in AI data center power opportunities. The Applied Digital project is the first domino. The White House pledge is the catalyst that accelerates every domino behind it.

• The option doubles the contract. Base Electron - - the power producer behind the current project - - is evaluating an option for an additional 1.2 gigawatts. That’s another $2+ billion sitting on the table, subject to a single decision.

• Analyst coverage is still thin. Only four analysts cover BW. None have fully modeled the AI data center growth potential into their forward estimates. When the big institutions arrive - - and they will - - they won’t be doing so quietly.

• The balance sheet is strengthening. BW retired its February 2026 bonds on schedule and is targeting December 2026 next. A $2.4 billion contracted backlog doesn’t just generate revenue - - it generates the credibility that changes how lenders and equity investors price the stock.

• The stock is still reasonably priced. As BW starts building the new $2.4 billion project, revenues will start to ramp dramatically over the next few years. And a good chunk of those revenues will fall to the bottom line as profits. I believe BW has earnings power of more than $1.00 per share (EPS) in the next 12-24 months.

BW shares finished trading last week at $13.10. That means BW shares are trading at a forward price to earnings (P/E) ratio of just 13x, still cheap for a company with growth prospects of more than 100% in the coming years.

The Bottom Line

The Trump pledge didn’t create demand for dedicated AI power generation. That demand already existed.

What it did was formalize the obligation - - publicly, with the biggest tech companies on earth as signatories.

Every Big Tech company that walked out of that White House now has a political and reputational incentive to get moving.

And when they move, they need someone who can build.

Babcock & Wilcox has been building for 159 years.

And they just found their biggest customer base in history.

The biggest gains in investing don’t go to those who arrive after the crowd.

They go to those who recognize the inflection point while it’s still unfolding - - and act swiftly.

The BW growth story is still in its early chapters.

And you still have a chance to get in before everyone starts talking about this quiet giant of the emerging AI power boom.

Best wishes in your trading and investing in the week ahead.

John McHugh cofounded Equity Empire with Colin Tedards to help investors achieve financial freedom. He’s a former Wall Street analyst with more than 30 years of experience investing in small cap stocks. This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

For access to John and Colin’s current buy alerts and full research portfolio, visit:

myequityempire.com/elite-yearly-subscription/

Millions of drones.

All running on a two-dollar chip made by companies you’ve probably never thought about as defense stocks.

Last week, the U.S. fired its first-ever one-way attack drone during strikes on Iran. The Pentagon just launched a billion-dollar program to build attack drones at five thousand dollars each. Ukraine is producing over a million drones a year. Iran is launching waves of them as we speak.

When researchers tore apart an Iranian-made Shahed drone, they found about fifty-two electronic components inside. Roughly eighty percent of them were American-made. Not by defense contractors. By analog chipmakers.

Premium Equity Empire subscribers already know about Texas Instruments — the company whose parts showed up nearly two dozen times in those teardowns. I recommended the stock back in December. But TXN is just one of five semiconductor companies whose chips appear in virtually every military drone on the planet.

Here are the other four. Plus three pure-play drone makers worth having on your radar.

Analog Devices (ADI) — The Precision Play

Where Texas Instruments dominates on volume and breadth, Analog Devices makes the higher-performance chips. Guidance systems. Radar. Electronic warfare. Signal processing. If a drone needs to do something precise - lock onto a target, process radar returns, jam an enemy signal - there’s a good chance ADI silicon is involved.

Of the five companies in these teardowns, ADI has the most direct and acknowledged defense exposure. Their aerospace and defense segment just hit a record quarter. Revenue surged thirty percent in Q1 2026.

The drone angle: ADI’s chips handle the sophisticated stuff - the sensing, targeting, and electronic warfare capabilities that separate a smart drone from a flying pipe bomb. As drones get more autonomous and more contested by electronic warfare, demand for this kind of precision silicon only grows.

NXP Semiconductors (NXPI) — The Secure Microcontroller

NXP microprocessors were found inside Shahed drone teardowns. They’re Dutch-headquartered but U.S.-listed, and about sixty percent of their revenue comes from automotive. But here’s where it crosses over - their secure microcontrollers and radar processing chips have direct military applications.

Think drone navigation systems. Identification-friend-or-foe modules. Encrypted communication links. NXP builds the kind of secure, hardened processors that military systems require when they need to verify what’s friendly and what isn’t - which matters a lot when you’ve got swarms of autonomous drones in contested airspace.

The drone angle: NXP’s automotive radar expertise translates directly to drone sensing and navigation. Their secure element chips are designed to prevent tampering and spoofing - critical when adversaries are actively trying to jam or hijack your drones.

ON Semiconductor (ON) — The Eyes and the Power

ON Semi plays a completely different role than the other four. They don’t make the brain of the drone. They make the eyes and the power system.

Their silicon carbide and gallium nitride chips handle power conversion in harsh environments. That’s critical for drone motor controllers and battery management - extracting maximum flight time and range from a small lithium-polymer battery. Their CMOS image sensors are literally the cameras on these drones. When an FPV pilot is flying a four-hundred-dollar drone into a Russian tank, the video feed coming back is running through ON Semi silicon.

ON Semi is less about showing up in Shahed teardowns and more about being essential to the next generation of autonomous systems that need better power efficiency and better vision.

The drone angle: Two critical functions — power management and imaging. As drones push toward longer range, longer endurance, and autonomous target recognition, both of these capabilities become more important, not less.

Microchip Technology (MCHP) — The Flight Controller Brain

Microchip Technology makes the microcontrollers and FPGAs that serve as flight controllers and mission computers in smaller drones. Their PIC and AVR microcontroller families are ubiquitous in embedded systems — the kind of cheap, reliable, low-power processors that run the basic functions keeping a drone in the air.

They explicitly serve aerospace and defense as a named end market. And Citi actually has MCHP as their top semiconductor pick for 2026, partly on the cyclical recovery thesis.

The drone angle: When the Pentagon says it wants drones at five thousand dollars or less, the flight controller needs to be dirt cheap and rock solid. That’s Microchip’s sweet spot. Their embedded processors are battle-tested across millions of industrial applications — and now they’re going into millions of drones.

The Pure-Play Drone Makers

The semiconductor companies are the picks-and-shovels layer. But there are also companies building the drones themselves. These are higher risk, higher reward - and more directly tied to defense procurement cycles/contracts.

AeroVironment (AVAV) — The Switchblade

AeroVironment is probably the most prominent pure-play drone investment right now. They make the Switchblade family of loitering munitions — essentially kamikaze drones that a soldier launches from a tube, flies to a target via live video feed, and detonates on impact.

The U.S. Army just awarded them a hundred and eighty-six million dollar order for next-gen Switchblade 600 and 300 variants, under a broader contract ceiling near a billion dollars. The Switchblade has been battle-tested in Ukraine and proven effective against armored vehicles and personnel.

Stock: ~$260, up 71% over the past year The catch: Trades at roughly 150x trailing earnings. That’s a steep premium even for a high-growth defense name. You’re paying for the future here.

Kratos Defense & Security Solutions (KTOS) — The Drone Dominance Contender

Kratos is in the Pentagon’s new Drone Dominance program - the billion-dollar initiative to mass-produce cheap attack drones. They build target drones, tactical unmanned systems, and have a long history with the military’s drone programs.

What makes Kratos interesting is they already know how to build drones at scale for the military. Their Valkyrie autonomous combat drone has been in development with the Air Force. The Drone Dominance program could be a significant revenue catalyst if they’re one of the final twelve vendors selected.

The angle: Less pure-play “drone stock” and more diversified defense technology company with meaningful drone exposure. Lower valuation risk than AVAV.

Red Cat Holdings (RCAT) — The Small Drone Winner

Red Cat won a major Army contract in late 2024, replacing Skydio as the military’s small reconnaissance drone supplier. These are the handheld drones that soldiers carry in their packs and launch for immediate situational awareness.

This is the smallest and most speculative name on this list. But they’re positioned in the fastest-growing segment - small tactical drones that every infantry unit will eventually carry as standard equipment.

The angle: High-risk, early-stage company that just landed a meaningful contract. The small drone category is where volume will be highest - but execution risk is real.

The Supply Chain Angle

One thing that ties all of these companies together - semiconductor and drone maker alike - is the supply chain.

China controls ninety percent of the global commercial drone supply chain. Carbon fiber frames, rare-earth magnets, lithium-ion cells, gallium-nitride chips, flight controllers, sensors. In September 2024, China imposed export restrictions on drone components to both Ukraine and Russia. Then in April 2025, they suspended exports of critical minerals and magnets entirely.

The Pentagon now requires “Blue UAS” compliance - or no Chinese components in anything sensitive. That’s a reshoring tailwind for every American semiconductor and particularly Texas Instruments which manufactures the chips domestically.

Same playbook as the CHIPS Act for advanced semiconductors. But this time it’s for the low-end analog and embedded chips that go into every drone flying today.

The Bottom Line

The companies behind the drone boom aren’t Lockheed and Raytheon. They’re analog chipmakers and small defense firms whose products show up in every drone on earth.

This is an idea generation piece - and other than Texas Instruments (and possibly NXPI) none are on my buy list. Each of these companies has a different risk profile, a different exposure level, and a different reason to be watching them. But the thesis is simple: drones are the defining weapon of modern warfare, their production is scaling to levels nobody anticipated even two years ago, and the companies supplying the guts deserve to be on your watchlist.

Seven names. One tailwind. Do your own research from here.

For more on Texas Instruments specifically — including why I recommended it to premium subscribers back in December — watch the full video breakdown

Nvidia CFO Colette Kress, Feb 25, 2026

The AI Boom Is Having Its Model T Moment

In the early 1900s, Henry Ford’s Model T rolled off assembly lines, promising mobility for the masses.

But before the public would get access to Ford’s four-wheel wonders, there was a problem that had to be addressed first.

Roads were scarce and gas stations even scarcer.

Without massive investments in highways and refineries, Ford’s Model T’s would have sputtered into obscurity.

So in the 1920’s, the nation had to build.

In addition to building cars - - it also had to build the broad ecosystem to fuel them.

We’re at a similar inflection point today.

Not with automobiles, but with artificial intelligence.

AI isn’t just a tool.

It’s a voracious consumer.

And it’s devouring electricity at a scale that outpaces our electric grids.

If we don’t ramp up power plant construction (natural gas, nuclear, solar, wind … whatever it takes), we risk stalling the most transformative technology since the internet.

This isn’t hyperbole.

It’s the consensus emerging from the very architects of the AI boom.

Consider the words of Alphabet CEO Sundar Pichai.

In a candid earnings call, he pinpointed the core challenge:

“All constraints - - be it power, land, supply chain constraints - - how do you ramp up to meet this extraordinary demand for this moment?“

Pichai isn’t mincing words.

Pichai’s ambitions - - from search enhancements to outright AI supremacy - - are slamming into real-world limits.

He’s even floated putting servers in space to tap unlimited solar power, admitting the “immense” energy needs of AI.

It’s a sci-fi solution to an earthly problem:

The US energy grid wasn’t built to handle the insatiable power needs of AI.

Echoing Pichai is Amazon CEO Andy Jassy, whose AWS is the cloud backbone powering much of the world’s AI.

Jassy describes the situation like this:

“It is so unprecedented how much compute is being consumed right now.“

And Jassy calls power “the single biggest constraint,” warning of shortages that could hobble growth.

Amazon is pouring billions into data centers.

But without more electricity, those investments will sit idle.

It’s like building a fleet of electric cars without having enough charging stations.

Microsoft CEO Satya Nadella adds a layer of urgency, framing it as a societal imperative.

“The biggest issue we are now having is not a compute glut, but it’s [getting access to] power,” he says, revealing stacks of unused GPUs waiting for “warm shells” to plug into.

Nadella warns that AI’s “vast amount of energy” is “putting a lot of pressure” on grids, and without proving its value, we might lose “social permission” to consume such scarce resources.

His point?

This isn’t just about tech.

It’s about earning public buy-in by delivering breakthroughs in healthcare, education, and beyond.

Meta CEO Mark Zuckerberg also sees energy as the ultimate bottleneck.

“I think we would probably build out bigger clusters than we currently can if we could get the energy to do it,” he admits.

Meta is planning “tens of gigawatts this decade, and hundreds of gigawatts or more over time,” envisioning facilities the size of Manhattan to fuel AI like Llama models.

Without power plants to match, however, these plans are nothing more than blueprints.

NVIDIA founder Jensen Huang, the visionary chipmaker fueling the AI frenzy, calls AI data centers “factories” that transform energy into intelligence.

“We should invest in just about every possible way of generating energy,” he urges, predicting a “decade of buildout” where power limits everything.

Huang is warning that energy - - not chip production - - now limits how fast intelligence can expand.

What ties the voices of these Big Tech titans together is a shared revelation:

AI has tipped us into an energy revolution.

Like the tipping point from the horse and buggy to the internal combustion engine automobile, we’re at a cusp.

Data centers could consume 12% of U.S. electricity by 2028, tripling today’s levels.

This development has enormous implications for the nation’s financial, social, and political future.

President Trump even addressed the issue in his State of the Union address on February 24.

“We have an old grid. It could never handle the kind of numbers, the amount of electricity that’s needed. So I’m telling them [the major tech companies], they can build their own plant. They’re going to produce their own electricity. It will ensure the company’s ability to get electricity, while at the same time lowering prices of electricity for you [the people] and could be very substantial.”

• President Donald Trump, 2026 US State of the Union Address

The Bottom Line

So here’s the bottom line.

The world’s largest tech industry leaders are all saying that access to reliable power is one of the biggest limiting factors to building out the nation’s AI infrastructure.

Without new plants - - perhaps trillions in investments - - US citizens face blackouts, stalled innovation, and lost economic advantages.

So just this week, President Trump told the world that tech companies with operations in the US need to build more power plants.

And that’s presenting a once in a lifetime investment opportunity for forward-thinking investors.

The Trillion Dollar Opportunity

Here are several companies who are ideally positioned for growth in the coming years driven by increased spending on new power infrastructure:

GE Vernova is a leading global provider of power generation equipment, specializing in gas turbines, wind power, grid solutions, and decarbonization technologies. It has seen explosive demand for its natural gas turbines from data centers and utilities building or co-locating power plants to meet AI electricity needs, with massive backlogs (including slot reservation agreements) and partnerships like those with Chevron and Engine No. 1 for multi-GW “power foundries.”

Fluor Corporation is an engineering, procurement, and construction (EPC) giant specializing in large-scale energy projects, including power plants, transmission infrastructure, and industrial facilities. It benefits from the wave of new power generation and grid upgrades required for AI-driven demand, with expertise in executing complex builds for utilities and private power developments.

Caterpillar Inc. is a leading maker of industrial engines, generators, and power systems, including diesel and natural gas reciprocating engines used for backup, peaker plants, and on-site generation. It benefits from the push for reliable, distributed power solutions at data centers and new facilities where grid constraints drive demand for modular, quick-deploy generation equipment.

Bloom Energy specializes in solid oxide fuel cell technology that provides efficient, on-site power generation for data centers and industrial users. It partners with investors and tech firms to deploy fuel cells as distributed power solutions, directly addressing the “build your own power plants” trend for AI facilities to avoid grid strain.

Quanta Services is a premier infrastructure construction and engineering firm specializing in electric power transmission, substation buildouts, and renewable/gas plant construction. It is seeing strong tailwinds from utilities and tech firms investing in grid modernization and new power plant tie-ins to support surging AI-related electricity loads across the US.

These five companies are integral pieces of a much bigger puzzle.

But they’re not the whole picture.

And because these companies are already widely followed by the investment community, their share prices already reflect a lot of future growth.

In my research at Equity Empire, I’ve identified two additional smaller companies in this space that most investors haven’t found yet.

I’m a small cap research analyst with more than 30 years of experience finding investment opportunities in emerging growth stocks before the crowd rushes in.

In the past year, I’ve shared several power-related stock picks with Equity Empire subscribers.

And in the past six months, we’ve closed out several large winning trade alerts including:

Eos Energy (EOSE) +91%

Bloom Energy (BE) +89%

Cipher Holdings (CIFR) +73%

Babcock & Wilcox (BW) +57%

Hondius (HOND) +55%

And I’ve recently issued two new power-related buy alerts for Equity Empire subscribers that I believe have similar - - if not greater - - price appreciation potential.

Most investors have never heard of these two stocks … yet.

That’s exactly the point.

The best opportunities in the market are always found before the crowd arrives - - not after.

If you want access to both company profiles, the trade thesis behind each one, and my best stock picks throughout 2026 as the AI power buildout accelerates - - Equity Empire is where the world class research lives.

Colin and I built Equity Empire to deliver the kind of deep, experienced analysis that used to be reserved only for institutional investors.

My previous employers paid me hundreds of thousands of dollars for my unique investment insights.

Now these insights can be available to you for a tiny fraction of the cost.

Click on the link below to get immediate access:

myequityempire.com/elite-yearly-subscription/

The AI power trade is early.

The window is open.

Fortunes will be made.

But the opportunity shrinks fast once the crowd shows up.

Best Regards,

John McHugh

This week, Citrini Research published one of the most compelling pieces of AI bear fiction produced thus far.

It’s a fake macro memo from June 2028 - written as if the worst-case AI scenario actually played out.

S&P down 38%

Unemployment at 10.2%.

White-collar workers driving Ubers.

A mortgage crisis triggered by AI displacement.

It’s well-written. I’m not here to trash it.

I’m here to tell you we’ve heard this story before. About a completely different technology. And the people who believed it missed one of the greatest wealth creation events in financial history.

The Year Was 2014

Bitcoin was trading around $400. Ethereum hadn’t launched yet. And the smartest people in every room were telling you, with absolute conviction, that cryptocurrency was going to change everything.

Marc Andreessen wrote in the New York Times that Bitcoin offered: “a sweeping vista of opportunity to reimagine how the financial system can and should work.”

The predictions were specific.

• Bitcoin would replace the dollar.

• Smart contracts would make banks obsolete.

• Decentralized finance would kill Visa and Mastercard.

• Crypto would dethrone gold as the global store of value.

• Intermediaries that skimmed 2-3% off every transaction were dead men walking.

Every single prediction was wrong.

What Actually Happened

Visa traded around $25 in 2015. Today it’s north of $300.

Mastercard? 20x over 15 years.

JPMorgan Chase - Assets nearly doubled from $2.4 trillion to $4 trillion.

Gold? Somewhere Peter Schiff is laughing.

Gold was supposed to be replaced by Bitcoin as the global store of value. Instead, gold went from $1,077 an ounce to over $5,000. Central banks are buying more of it than they have in decades.

The dollar was supposed to die. It’s still 58% of global reserves.

None of the doomsday predictions came true. Even the one’s that called crypto a scam or worthless.

And here’s the part that matters for investors: if you listened to no one … you won.

Bitcoin went from $430 to over $100,000 (!!!!!!)

A 21,000% return.

Stablecoins - which nobody predicted - now settle $23 trillion in annual volume.

Financial stocks ripped higher. Visa & Mastercard got bigger. The dollar is still king.

The doomers were wrong. The utopians were wrong. Everyone from the haters to the hodlers were wrong.

The investors who tuned it all out - made generational wealth.

The Same Script, Different Technology

Read the Citrini piece again with this lens. The structure is identical.

• Crypto was going to kill banks

• AI is going to kill SaaS companies.

• Crypto was going to eliminate payment networks

• AI is going to eliminate white-collar jobs.

• Crypto was going to destroy intermediation

• AI is going to destroy intermediation.

• Crypto was going to cause a financial crisis through correlated bets

• AI is going to cause a financial crisis through correlated bets.

Even the details rhyme. Citrini’s scenario has AI agents routing around Mastercard’s interchange fee using stablecoins on Solana.

The logic in the Citrini piece is sound. Just like the crypto logic was sound.

But sound logic doesn’t mean the conclusion is right. We learned that with crypto already.

It’s human nature to take a trend to its most extreme logical conclusion. We’ve been doing it with every technology since the printing press. And we’re almost always wrong.

The Timeline Problem

You don’t even need the crypto analogy to see the pattern. Just look at every technology disruption in the last 50 years.

ATMs were introduced in 1969. They were supposed to eliminate bank tellers. Instead, bank teller employment doubledover the next 40 years. ATMs made branches cheaper to operate, so banks opened more of them.

E-commerce has existed for 25 years. Amazon is the biggest retailer on the planet. And e-commerce is still only 20-22% of total retail sales.

Cloud computing has existed for two decades. Analysts predicted near-total enterprise migration by now. On-prem still holds 37% of workload capacity. Side note: I’m predicting that gets larger.

In total, disruption took 2-3x longer than the smartest people predicted. The incumbents adapted rather than died.

But here’s the bit takeaway for investors: The biggest wealth creation happened in the infrastructure that enabled the disruption - not in the destruction of the old guard.

We don’t even need to go back decades for proof. We have examples from the last 13 months.

The Predictions That Already Failed

Perplexity was supposed to kill Google. That was the consensus in early 2025. AI-native search would make traditional search obsolete. Perplexity peaked at 2.4% search market share. As of this month, it’s declining — down to 2.0%. Google recovered to over 90% market share.

DeepSeek was supposed to kill the capex cycle. In January 2025, a Chinese AI lab released a model that appeared to match frontier performance at a fraction of the cost. The narrative was instant: hyperscalers were overspending, the AI infrastructure buildout was a bubble, and capex would collapse.

A year ago I told you not to panic.

Thirteen months later, hyperscaler capex for 2026 is tracking $635-665 billion. That’s up 67-74% from 2025. NVIDIA’s revenue is up 62% since the DeepSeek announcement.

The incumbents didn’t die. They adapted.

Conclusion

Every doomsday prediction about a transformational technology follows the same arc. The logic is compelling. The timeline is aggressive. The conclusion is catastrophic. And then reality happens.

Crypto didn’t kill banks. But Bitcoin went from $400 to $100,000. The people who were busy worrying about the death of fiat missed a 21,000% return.

AI won’t cause the 2028 Global Intelligence Crisis. But it will reshape industries, create new infrastructure, and generate wealth at a pace we haven’t seen since the internet.

The doomsday predictions aren’t the risk. Believing them is.

I’ve been an individual investor and computer programmer for over 25 years. I’m currently building, coding, and experimenting with AI on a daily basis. Yes, I have a OpenClaw installed on a Mac that’s absolutely doing amazing things for me. I’ve sold my entire stock portfolio twice to buy houses … and I’m looking to do it again when it hits $5M. Join me on my journey by following along.

Subscribe now

While the world debates whether AI will boost productivity or eliminate jobs, a team of brilliant engineers in San Francisco just pulled off something extraordinary.

And forward thinking investors stand to make big profits from the development.

Anthropic’s $30 Billion Fundraise: The Second-Largest Private Tech Deal in History

On February 12th, the leadership team at fast growing AI startup Anthropic convinced some of the planet’s savviest investors - - Singapore’s sovereign wealth fund, Coatue, Founders Fund, D.E. Shaw, MGX, plus tech titans Microsoft and Nvidia - - to write checks totaling $30 billion.

Not to build factories.

Not to launch rockets to Mars.

But to keep creating increasingly sophisticated versions of something called Claude.

What Is Claude?

Claude is an AI thinking partner - - one that handles your everyday tasks: writing emails, analyzing documents, answering questions, and solving problems through conversation.

Imagine having a brilliant colleague on call 24/7 - - ready to brainstorm, edit your work, or explain complicated topics in plain English.

Planning a trip? Generating creative ideas? Trying to boost your work or personal productivity?

Claude makes it remarkably simple.

The Democratization of Intelligence

For centuries, advanced knowledge and tools belonged exclusively to those who could afford them.

Only big corporations had supercomputers. Only the wealthy could hire expert consultants at thousands of dollars per engagement. Specialized training created gatekeepers to crucial information.

AI tools like Claude are dismantling those barriers.

Today, a small business owner gets sophisticated market analysis without the consultant fees.

Students access personalized tutoring instantly.

Writers get professional-grade editing help.

Aspiring coders have an expert teacher available around the clock.

Parents can research their child’s medical diagnosis without waiting weeks for a specialist.

Remember how Google democratized information in the early 2000s?

Suddenly everyone had the world’s knowledge at their fingertips - - no encyclopedia collection or library card required.

AI takes this further.

It’s democratizing not just information, but intelligence itself - - the ability to analyze, create, solve problems, and make informed decisions.

The playing field is leveling.

Good ideas, hard work, and creativity now matter more than closed-door connections or deep pockets.

Everyone gets access to a value-creating, time-saving thinking partner.

The Insatiable Hunger for Computing Power

Creating legions of these smart AI assistants - - these digital fountains of near-infinite knowledge - - demands massive computing power.

In a recent podcast with Dwarkesh Patel, Anthropic’s CEO Dario Amodei predicted the amount of compute capacity the tech industry is building will triple every twelve months over the next several years: from 10-15 gigawatts in 2026 to 300 gigawatts by 2029.

An Exceptional Investment Opportunity Emerges

The average person’s eyes will glaze over once you start talking about electricity and gigawatts.

But here’s where the AI compute spending story gets interesting for investors.

Each gigawatt of compute costs $10-$15 billion to produce.

So if Amodei’s forecast is even close, within three to four years the tech industry could be spending trillions annually on computing infrastructure.

And that means investors in the companies who supply the critical components of the expanding AI infrastructure stand to make HUGE profits.

But there’s a catch.

America’s electric utilities can’t keep pace with tech’s explosive demand for power.

At Cisco’s AI Summit on February 3rd, Amazon AWS CEO Matt Garman put it bluntly:

“[Utilities] don’t scale at the level we need them to scale...We’re going to have to fund that ramp up.”

This wasn’t a throwaway comment.

It’s a signal that the world’s largest tech companies are about to pour serious money into power generation.

And that creates a massive opportunity to invest in the companies designing, supplying, constructing, and delivering that power.

Our Latest Pick: Perfectly Positioned for the AI Power Boom

Last week, I issued a new buy alert to Equity Empire subscribers on a company ideally positioned for this power generation investment trend.

This company leads in infrastructure services for the power industry: engineering, procurement, construction, testing, commissioning, operations, maintenance, and repowering.

Since 2008, this little known company has constructed over 500 power plants representing 20 gigawatts of generating capacity.

Between 2014 and 2024, they built one out of every nine megawatts of utility-scale solar in the US.

The company was 2024’s second-largest builder of battery energy storage systems, according to Solar Power World.

They currently service 146 operating power plants generating over 18 gigawatts.

And with tech heavyweights like Amazon starting to pour money into building out their power generation capabilities, it looks like this company could be poised for a sustained growth surge.

Despite this promising growth outlook, few investors have heard of this company yet.

That’s because the company issued its initial public offering (IPO) of shares last week when most market participants were distracted by news of broad AI-related business disruptions.

But I expect investor interest in this stock to increase dramatically over the coming 12 to 24 months.

As investors wake up to this accelerating and sustainable growth opportunity, I expect this stock to be a strong performer as the year unfolds.

Get Full Access to Our Best Ideas

Want access to this pick? Plus all the most promising investment opportunities that Colin and I share with our subscribers throughout 2026?

Join the rapidly growing Equity Empire community:

Subscribe to Equity Empire

Subscribers get research and picks delivered directly - - including early alerts on which stocks to buy AND when to sell for maximum gains.

Just in the past year, our research gave subscribers early entry to:

• Nebius NBIS (226% gain)

• Applied Optoelectronics AAOI (127% gain)

• Ondas Holdings ONDS (70% gain)

• Eos Energy EOSE (91.6% gain)

• Bloom Energy BE (92% gain)

Many of our subscribers bought into these stocks well before the “gurus” caught on.

The team of research experts at Equity Empire is laser-focused on a single mission:

We want to help you make money. Period.

That’s why over 1,000 investors have joined us in just the past six months.

We believe the coming year will be the biggest opportunity to become financially free in the last decade.

Join the Equity Empire subscriber community and let us help you secure your financial future.

Subscribe to Equity Empire

I’ll return with more market and small cap insights soon.

Until then, best wishes in your trading and investing this week.

Best Regards,

John McHugh

I’ve spent years tracking investment opportunities in AI datacenters, and the returns have been staggering.

My subscribers who followed my early calls on Camtek and Vertiv are sitting on gains of 196% and 455% respectively.

But the real home run came in early 2024 when I uncovered a little-known small cap that most investors had never heard of.

That company was Credo Technology Group .

Less than two years later, it soared over 800%.

Now, I’m seeing a new opportunity emerging in this space - - one that could dwarf everything that came before it.

But to understand where we’re headed, we need to understand how we got here.

The Problem Nobody Saw Coming

Back in early 2024, I published a report titled “The Multibillion-Dollar Problem: A Looming AI Data Traffic Jam.”

The thesis was simple:

As AI models grew exponentially, datacenters were hitting a critical bottleneck.

They needed to move massive amounts of data between servers and switches, but the infrastructure couldn’t keep up.

That’s where Credo came in.

The company made high-speed cables and chips that solved this exact problem.

Their technology used less power than traditional copper cables while moving data faster - - essentially the high-tech plumbing that keeps AI systems like ChatGPT running smoothly.

The market was skeptical. Wall Street had low expectations.

They were wrong.

Yesterday, Credo shattered forecasts once again.

The company announced third-quarter revenue guidance of $404-408 million - - more than Wall Street’s original estimate for the entire year when I first wrote about them.

Even better, management expects sequential mid-single-digit growth leading to over 200% year-over-year growth in fiscal 2026.

The stock has been a monster performer, which leads to the question my readers keep asking:

What’s the next CRDO?

From Copper to Light: The Optical Revolution

The datacenter industry is already moving to the next phase: optical interconnects.

As I detailed in my recent article on the $100 billion AI bottleneck, the world’s largest tech companies are transitioning from copper cables to light-based technology.

It’s faster, more efficient, and essential for the next generation of AI workloads.

Credo is managing this transition brilliantly and could remain strong for the next 12-18 months.

But there’s a new group of players emerging as the pure-play leaders in optical technology:

Lumentum designs and manufactures optical and photonic products - lasers, modules, and subsystems - - that enable high-speed datacenter connectivity. Think of them as making the key light-based components that help massive datacenters move information at blazing speeds.

Coherent Corp is a major photonics player developing engineered materials, lasers, and optoelectronic components. They’re experts in using light to create high-performance parts for everything from AI datacenters to advanced manufacturing.

Ciena builds the “highways” for internet and data traffic using advanced fiber-optic technology to connect cities, countries, and massive datacenters - - especially for cloud and AI workloads.

Celestica is the behind-the-scenes partner assembling and scaling high-tech gear for big tech companies and hyperscalers, helping turn designs into reality quickly and globally.

Astera Labs specializes in semiconductor-based connectivity solutions that solve bottlenecks in rack-scale AI infrastructure. They make smart controllers that let AI servers, memory, and networks communicate faster inside massive datacenters.

Corning manufactures the specialty glass and optical fiber that forms the backbone of modern communications infrastructure.

All of these companies are experiencing rapid growth right now.

And given Big Tech’s plans to dramatically boost spending on AI infrastructure this year, the datacenter demand for optical interconnect technology will be off the charts in 2026 and 2027.

But here’s the catch:

Much of that growth is already priced into their shares after massive recent runs.

I wouldn’t chase them here.

Instead, wait for pullbacks to find better entry points.

And I’d start thinking about the coming transition to next generation datacenter interconnect technologies.

The Next Frontier: When Electronics Meet Photonics

This is where things get really interesting - - and risky.

The next evolution beyond optical interconnects involves merging electronics with photonics at the chip level.

Technologies like co-packaged optics (CPO) and silicon photonics integration promise enormous gains in bandwidth, lower latency, and dramatically reduced power consumption.

But this transition won’t be easy.

The industry faces several major hurdles through 2026-2030:

Manufacturing complexity. Integrating optical components directly with chips requires advanced, precise fabrication processes that are far more expensive and yield-challenged than current solutions. Supply chain constraints for high-speed components could persist through 2029.

Thermal challenges. Packing optics and high-power electronics together generates intense heat in dense AI clusters. This demands advanced cooling solutions and risks degrading performance if not managed perfectly.

Reliability risks. Unlike easily swappable pluggable optics, CPO embeds components into the switch or GPU package. A single failure could require replacing an entire expensive module, leading to higher downtime and repair costs.

Standardization gaps. The industry currently relies on interchangeable pluggables from multiple vendors. CPO often locks customers into specific ecosystems, reducing flexibility until broader standards mature.

Integration complexity. Efficiently connecting fibers to chips remains technically difficult, especially at high density. Managing fiber routing and system-level design adds layers of complexity in hyperscale environments.

Despite these challenges, the big tech companies are already planning the transition.

They have no choice.

Nvidia’s next-generation GPUs will demand data transmission speeds that current interconnect technologies simply can’t support.

The stakes are enormous - - and so are the potential rewards.

The High-Risk, High-Reward Plays

For aggressive investors with high risk tolerance, three companies are working on breakthrough next-generation photonic technologies:

POET Technologies has developed a proprietary Optical Interposer platform that enables seamless wafer-level integration of electronic and photonic devices into single chips. Their technology creates advanced “bridges” that combine light-based and electronic components to handle massive data flows efficiently - - supporting speeds from 400G up to 3.2T and beyond.

Lightwave Logic is developing proprietary electro-optic polymers and photonic devices that convert electrical signals to optical signals for ultra-fast data transmission. Their special polymer materials act as super-efficient “switches” for light in fiber-optic networks, targeting the next wave of 800G and 1.6T transceivers.

Aeluma is building advanced light-detecting sensors and communication chips using proprietary technology that combines compound semiconductors with large-scale silicon manufacturing. Their applications span mobile devices, automotive, AI infrastructure, defense, AR/VR, and quantum computing - - aiming to make high-end photonics affordable and mass-producible.

The Bottom Line

These three companies are working on genuinely exciting next-generation datacenter technologies.

If their innovations become widely adopted by the industry, the returns could be extraordinary.

But let me be clear:

Given the technical and market challenges I outlined, these are highly speculative investments. They should only be considered by investors with an extremely high tolerance for risk.

We’ve seen what can happen when you identify the right technology at the right time.

Credo proved that.

The optical plays are showing similar potential now.

The question is this:

Which of these next-generation photonics companies will be the next 800% winner?

Only time will tell - - but the race is already underway.

You Can Get Wall Street Research for a Fraction of the Cost

Want access to my complete research on AI datacenter investments, including detailed analysis, price targets, and real-time updates on these opportunities?

Join Equity Empire, my premium stock research subscription service where I share in-depth reports on emerging tech companies before Wall Street catches on.

Members get exclusive access to my full investment thesis, risk assessments, and timely alerts on when to buy, hold, or take profits.

Click here to join the fast growing, Street-smart subscriber community at Equity Empire:

myequityempire.com/elite-yearly-subscription/

I’ll be back with more research and market insights over the coming weeks here on Substack.

In the meantime, wishing you the best in your trading and investing in the week ahead.

John McHugh co-founded Equity Empire with Colin Tedards to help small investors achieve financial freedom. He’s a former Wall Street analyst and hedge fund manager with more than 30 years of experience investing in small cap stocks.

To learn more about Equity Empire and get ongoing access to John’s top small cap ideas, visit:

myequityempire.com/elite-yearly-subscription/

This is not financial advice. Past performance does not guarantee future results. Always conduct your own due diligence before investing.

In 2023, Colin and I recommended a little-known stock that changed the trajectory of our subscribers’ portfolios.

Since that alert, it has rocketed 244%.

Take a look at the chart:

Today, I’m pulling back the curtain on a new opportunity.

It’s another under-the-radar player with the same DNA as our first big winner - - and the potential to deliver similar life-changing returns over the coming years.

If you have an appetite for small-cap stocks with massive appreciation potential, you’ll want to pay close attention to the shift happening right now.

The Blueprint: Profiting from the “Nvidia Effect”

In August 2023, we published a research report titled “The Nvidia Effect.”

While the rest of the world was chasing Nvidia , we took a different path.

We realized that the hundreds of billions being poured into AI infrastructure created a “halo effect.”

To make Nvidia’s chips work, the world needed a massive ecosystem of supporting technology to improve speed and efficiency.

We focused on an emerging new technology called High Bandwidth Memory (HBM).

AI training requires data access speeds that traditional memory simply cannot handle.

HBM was designed to meet AI developers’ constantly expanding bandwidth requirements.

But its complexity also required more precise testing to ensure high quality.

So we concluded that HBM manufacturers would need an explosion of new equipment to inspect and test these chips before they ever reached an Nvidia motherboard.

Our pick was Camtek .

At the time, Camtek was an undiscovered gem selling testing equipment to giants like TSMC , Samsung, SK Hynix, and Micron Technology .

In our report, we noted:

“Camtek is at the forefront of a massive wave of HBM spending that will dramatically accelerate revenue and profit growth. Management expects to receive additional orders over the coming months and years.”

If Camtek hits our revenue estimate in the next year, we expect it to earn a higher multiple of at least 10x. That would deliver a 76% gain. Forecasting out to 2026, that would deliver a 176% gain. That’s our base case.”

Our forecast was conservative.

Subscribers who acted on that research are currently sitting on 244% gains.

Now, the “Nvidia Effect” is moving into its next phase … and it has led us to a new target.

The Next AI Infrastructure Megatrend: Optical Data Transmission

In 2023, HBM was a new technology that was just getting rolled out.

In early 2026, surging demand for HBM has created a high bandwidth memory bottleneck.

But a new AI infrastructure bottleneck is also emerging … and its creating an exciting opportunity for savvy small cap investors:

High speed data transmission and connectivity.

As AI data centers scale to 100,000+ GPU clusters, traditional copper wiring is failing to handle the oncoming AI data tsunami.

Copper creates too much heat and loses signal integrity over distance.

To solve this, the industry is aggressively pivoting to optical interconnects - - using light instead of electricity to move data.

Read our article on the transition to optical interconnects here:

Every single one of these high-tech optical components must be tested for flawless performance.

If the “highway” inside the data center fails, the multi-billion dollar AI cluster grinds to a halt.

This is where Viavi Solutions comes in.

Viavi Solutions: The Quality Control King of AI Data Transmission

Viavi provides the specialized measurement and validation tools that form the backbone of modern AI factories.

As data centers jump to 800G and 1.6T speeds, Viavi’s instruments ensure that every fiber and transceiver performs at peak capacity.

In short: Viavi is the high-tech “quality control expert” for the fiber-optic highways of the future.

Much like CAMT is the “quality control expert” for HBM manufacturers.

The setup with VIAV today is strikingly similar to Camtek in 2023:

• The Catalyst: A massive shift from copper to optical interconnects.

• The Momentum: Viavi recently crushed analyst estimates, sending the stock to multi-year highs.

• The Pivot: For years, Viavi was tied to sluggish telecom cycles. Now, its growth is being fueled by the insatiable demand of hyperscale AI data centers.

Viavi’s CEO, Oleg Khaykin, recently confirmed this trajectory, noting that strong momentum in the data center ecosystem will continue to drive growth through 2026:

“We expect the strong momentum and demand in the data center ecosystem and aerospace and defense applications to continue driving our anticipated growth this calendar year.”

The Bottom Line

Equity Empire subscribers have already seen a 35% gain on VIAV since our original buy alert, but we believe the VIAV growth story is just getting started.

Just as Camtek rode the HBM wave to triple-digit returns, VIAV is now positioned to catch the massive wave of optical infrastructure spending.

The “Nvidia Effect” is just starting to take hold here, and for investors who missed the move in CAMT starting in 2023, VIAV is your second chance.

Smart Investors Jump on Megatrends Early.

I haven’t had time to mention it in this report, but the second company in our “Nvidia Effect” report published in mid 2023 was up even more than CAMT.

That company is Vertiv - - the leading provider of datacenter cooling equipment.

VRT soared as much as 382% from our original recommendation:

Additionally, one of our later “Nvidia Effect” recommendations - - datacenter interconnect provider Credo Technologies - - climbed as much as 885% after our initial buy alert.

Past performance like this is not a promise of future results.

But the team at Equity Empire has been delivering exceptional results for the past several years.

And we’ve been delivering Wall Street quality large and small cap research to our subscribers for a fraction of Wall Street’s cost.

I’m currently working on several small cap opportunities that have similar, if not greater, profit potential than CAMT and VIAV.

If you’d like access to these opportunities and our best research throughout 2026, click on the link below for more information:

myequityempire.com/performance

I’ll be back with more small cap insights here on Substack in the days and weeks ahead.

In the meantime, wishing you the best in your trading and investing in the week ahead.

Your personal small cap research analyst,

John McHugh

You can also follow me on X here: x.com/satoritrade

And my chart analysis videos here:

yoube.com/@EquityEmpireJohnMcHugh

But this time it’s different.

An open-source AI agent called Clawdbot went viral over the weekend.

The pitch: a “digital employee” that runs 24/7 on your local computer, handling email, calendars, messaging, and tasks autonomously.

It can browse websites, save files, build spreadsheets, control your calendar … all while you relax comfortably on the sidelines.

Tech Twitter exploded. It reminded me of DeepSeek a year ago.

But the biggest surprise was the preferred hardware.

Not a GB300 super computer with the latest from Nvidia.

But instead, a Mac Mini.

One of Apple’s cheapest computers.

Within days, Mac Mini sales started trending as developers see what’s coming next.

Most investors will see it as a quirky tech trend.

They’ll miss what it actually signals.

The Hardware Shift Is Happening

For two years, AI has been cloud-based. Rented intelligence. You pay OpenAI per token. Your data lives on their servers.

Clawdbot - and tools like Claude Cowork - represent something different.

AI moving to the edge. Running (somewhat) locally. Owned, not rented.

Is Clawdbot overhyped?

Yes.

It’s not the full AI stack on a Mac mini …

But that’s missing the point.

The Implementation Doesn’t Matter. The Direction Does.

Remember Bitcoin mining in 2019?

I’m going to guess you don’t.

That’s because only tech geeks on the edge were into it.

But it proved where things were going.

It didn’t matter that early Bitcoin mining was complex and impractical for most people.

What mattered was the shift.

From centralized to decentralized, running on personal hardware.

AI is hitting the same inflection point.

The trend is unstoppable and in some ways needed.

AI moving from strictly the cloud to locally-owned, always-on agents.

And Wall Street is missing the infrastructure play underneath.

Layer 3.5: The Infrastructure Nobody Sees

Nvidia CEO Jensen Huang recently laid out five layers of AI infrastructure:

Layer 1: Power

Layer 2: Chips

Layer 3: Datacenters

Layer 4: LLMs

Layer 5: Software applications

Most investors understand these layers.

But there’s a critical layer he didn’t mention.

Layer 3.5 - the infrastructure software between datacenters and applications.

When you use ChatGPT, you don’t see:

• The database storing conversation history

• The monitoring systems tracking performance

• The cybersecurity protecting against attacks

• The API layer integrating multiple systems

That’s Layer 3.5.

And when AI agents replace humans, this layer explodes in demand.

The Agentic Workflow Problem

Software was designed for humans.

A person opens a spreadsheet. Makes changes. Saves. Closes.

Limited interactions. Manageable database load.

Now imagine Clawdbot running 24/7:

• Reading/responding to hundreds of emails daily

• Modifying dozens of documents in minuets

• Writing to databases constantly

• Integrating calendar, messaging, file storage simultaneously

One AI agent generates exponentially more database activity than a human.

Multiply that by millions of agents across organizations.

Four Infrastructure Categories That Win

1. Databases AI agents need constant data storage and retrieval.

2. Monitoring Autonomous agents running 24/7 need performance tracking and error detection.

3. Cybersecurity Agents with access to everything need protection against compromised systems and data exfiltration.

4. API & Integration Agents require seamless connections across multiple systems.

Why Wall Street Doesn’t Get It

Investors see software as one category.

Salesforce struggling. Adobe facing headwinds. Conclusion: “AI will replace all software.”

But that’s backwards.

The application layer faces disruption. The infrastructure layer faces explosive demand.

What Most People Miss

I can hear the pushback already: “But most people won’t run local AI agents.”

True.

Most people didn’t mine Bitcoin either.

The technology doesn’t need mass adoption to drive infrastructure demand.

One company deploying 10,000 agents creates more strain on compute resources than 10,000 individuals at an organization.

The Bottom Line

Clawdbot might be overhyped. The security might be is sketchy.

None of that matters.

What matters: AI is moving to the edge.

From rented cloud compute to owned local hardware. From centralized to distributed.

That shift creates two opportunities:

Hardware: Apple, Nvidia, ARM benefit from edge AI devices.

Infrastructure Software: benefit from agentic workflows

Wall Street debates whether Salesforce survives.

They’re missing Layer 3.5 - the infrastructure software about to see explosive demand.

The home hardware moment for AI is here.

Position before Wall Street figures it out.

Have you subscribed to my podcast?

I’m recording podcast everyday about topics and questions sent in by the listeners.

Here’s an episode where I talk more about this infrastructure 3.5 layer in depth:

Apple Podcasts: Equity Empire Podcast