How to capture growth AND income in 2026’s surging orbital economy. TODAY 2PM ET. Click Here to register

You can buy every GPU NVIDIA makes.

You can pour the concrete. Sign the land lease. Order the servers.

If you can't energize the building, they're just expensive paperweights.

That's not a metaphor. It's what is actually happening right now across the United States, as the most aggressive infrastructure buildout since the interstate highway system collides with a four-year training program and a transformer backlog that stretches to 2031.

Every great technology supercycle hits a wall. The investors who get rich aren't the ones who identified the trend first — they're the ones who spotted the constraint that was about to be solved. The AI supercycle has already made that trade multiple times. It's about to do it again.

The Pattern That Has Made Fortunes — Over and Over

The question most investors are still asking is: "Is AI demand real?"

That's yesterday's debate. The only question that matters now is: "Can the world physically build the infrastructure fast enough to deliver the compute?"

Because mega-trends don't die when demand disappears. They stall when they hit a constraint. And the investors who make the real money aren't the ones who find the trend first — they're the ones who find the bottleneck that's about to be solved.

AI has already run this playbook in public, chapter by chapter:

Compute was the first wall.  CPUs were built to execute tasks sequentially. Training an AI model requires millions of parallel calculations simultaneously. NVIDIA had a chip designed for video games that happened to be exceptional at parallel computation. NVDA went from roughly $4 (split-adjusted) to over $130. That's what solving the bottleneck looks like.

GPUs got fast — then they starved for data.  Memory bandwidth became the chokepoint. HBM (High Bandwidth Memory) emerged as the answer. SK Hynix and Micron surged.

Then the chips couldn't talk to each other fast enough.  Copper cables hit their physical limits. Silicon photonics and optical networking became the trade. Coherent, Marvell, Lumentum.

Then the world ran out of places to put all this hardware.  Data center real estate and industrial REITs became the bottleneck play. EQIX, DLR, Iron Mountain.

Then people objected to data centers in their backyard. Now space is the new frontier.

Then there wasn't enough power delivery.  Transformers, switchgear, and grid-edge equipment became the constraint. Eaton, Vertiv, and Powell Industries made their investors very comfortable.

You'll notice the pattern: every solved bottleneck exposes the next one. The AI supercycle is a chain. And right now, the chain is stuck in a place that very few investors are looking.

The Bottleneck Map: AI's Full Stack, 2016 to Today

The New Constraint Isn't the Price of Power — It's the Ability to Deliver It

A major Wall Street research desk recently revised its AI and data center power demand forecast sharply higher — projecting 220% growth in global power consumption from this sector between 2023 and 2030. To put that in human-scale terms: the incremental AI power demand by decade's end will be roughly equivalent to adding an entirely new top-ten energy-consuming nation to the planet.

The hyperscalers — Microsoft, Google, Amazon, Meta — have already revised their combined 2026–2027 capital budgets upward by more than $300 billion. They are currently reinvesting approximately 87% of their operating cash flow back into capex and R&D. This buildout is not slowing down.

But here's the conclusion that should stop you cold — and that the market has almost entirely missed:

The most binding constraint on AI infrastructure buildout is not the cost of electricity. It is not nuclear permitting. It is not even transformer lead times, though those are real.

It is licensed electricians — and the four-year apprenticeship programs required to create them.

The US needs to add roughly 500,000 new workers to build out the generation and grid infrastructure AI requires. About 300,000 of those are on the generation side, where skill requirements are lower and the timeline is more manageable. The other 200,000 are in transmission and distribution — T&D — where licensed electricians are the critical path.

Here's the math that clarifies the problem: there are currently about 45,000 energy apprentices in the US pipeline. The annual flow of newly licensed electricians entering the workforce is what's short — not just the total count. Adding 20,000–25,000 new apprentices to the system today means qualified electricians four years from now. Capital cannot compress that timeline. A $10 billion infrastructure commitment does not produce a licensed electrician in six months.

This is not a problem that resolves in 2026. Or 2027. It is structural, and it translates directly into pricing power for the companies who already have these crews.

The "6 Ps" Framework: Drivers and Constraints on AI Power Demand

Pervasiveness: AI adoption breadth — a demand driver, not a constraint; enterprise and agentic machine-to-machine use cases are still in early innings

Productivity: Model efficiency gains reduce energy per token — but freed capacity historically gets consumed by new use cases (Jevons Paradox applies)

Price of Power: Rising, but manageable: even paying $40/MWh more for clean power across all global data center demand impacts hyperscaler 2030 EBITDA by only ~2.5%

Policy: NIMBY politics and state-level data center moratoria are real friction; "take-or-pay" contracts between hyperscalers and utilities are the emerging structural fix

Parts: Gas turbine lead times stretch to 2029–2031; nuclear meaningful only in the 2030s; solar + batteries + simple-cycle gas filling the gap now

People: The #1 constraint: 500,000 new US jobs needed; licensed electricians require 4-year training; the annual pipeline flow is structurally short

Two Market Dislocations Are Already Forming

The skilled labor shortage isn't a future problem waiting to arrive. It is actively reshaping where and how AI infrastructure gets built right now. Two dislocations are already visible to anyone looking closely.

Dislocation 1: Behind-the-Meter Power Becomes the Workaround

Hyperscalers cannot wait for 3–5 year grid interconnect timelines. So they're moving to behind-the-meter solutions — on-site natural gas generators sited directly at the data center, bypassing the public grid queue entirely. Simple-cycle gas generators are less efficient than combined-cycle plants. They consume more gas per megawatt. They're not the preferred long-term solution. But they can be operational in months, not years.

This is not a political statement about fossil fuels. It is a construction schedule. And it creates a structural floor under natural gas demand regardless of what the clean energy narrative says publicly.

Dislocation 2: T&D Contractors Become the Toll Collectors

If there aren't enough qualified crews, the ones who have them can name their price. In a world where everyone — utilities, hyperscalers, municipalities — is racing to energize megawatt-scale facilities simultaneously, a licensed electrical contractor with a trained workforce has something money can't instantly replicate.

That scarcity premium doesn't go away in a quarter. It compounds for the duration of the training pipeline gap — which, as we've established, doesn't close before 2029 at the earliest.

The Halo (not to be confused with Heavy Asset Low Obsolescence) Effect: How to Own the Chain Before It Gets Crowded

Every bottleneck in the AI chain has created what I call a halo effect — a ring of secondary beneficiaries that investors could own long before the primary trade became obvious and crowded.

When NVIDIA became the consensus trade, the halo plays were the memory and interconnect names. When data center real estate became consensus, the halo plays were the power equipment and grid names. Now that energy infrastructure is becoming consensus, the halo plays are the skilled labor and construction firms that wire it all together — and the natural gas producers whose product powers the bridge.

The energy mix for AI data centers through 2030 looks roughly like 60% thermal (predominantly natural gas) and 40% renewables — with nuclear playing a meaningful role only in the 2030s and beyond. This is a "yes and" environment, not an either/or. But in the near term, the money follows the gas turbine and the electrical crew.

The halo around this bottleneck has five rings:

1.  T&D construction: The companies that build grid expansion — crews with licensed electricians, commanding premium rates for a decade

2.  Electrical equipment: The switchgear and distribution hardware between the grid and the servers

3.  Thermal & power management: Inside-the-data-center power delivery and cooling as server density accelerates

4.  Gas turbines + natural gas: The near-term bridge fuel — both the machines and the molecules

5.  Dispatchable power ownership: Firm 24/7 electrons from nuclear operators — the long-term destination the grid is eventually heading toward

Winners and Losers

Note on the Losers column: labeling hyperscalers as "losers" refers specifically to the margin headwind from rising capex intensity and power logistics costs — not a fundamental bear thesis on AI demand. These businesses remain the engines of the trend. The issue is that at 87% cash flow reinvestment ratios, the easy multiple expansion phase has passed.

Bear Case: What Would Break This Thesis

Three scenarios that would materially undermine the bottleneck trade:

1.  AI efficiency gains outrun demand growth.  If model improvements like DeepSeek R1 represent a persistent trend toward doing more with dramatically less compute, the power demand curve flattens. The counterargument — supported by historical data — is that efficiency gains consistently generate more demand, not less (Jevons Paradox). But it's a risk worth watching.

2.  Hyperscaler capex reversal in recession.  A deep enough economic downturn could force Microsoft, Google, or Amazon to cut capex. At 87% cash flow reinvestment, there is room to cut. This would hit the entire AI infrastructure trade simultaneously and without warning.

3.  Political backlash accelerates beyond management.  State-level moratoria on data centers are already appearing. If consumer electricity rates spike visibly and get attributed to AI data centers in the press, regulatory action could slow permitting dramatically — compressing the timeline this thesis depends on.

What to Watch: Your Tripwires

You don't have to predict the next bottleneck. You just have to watch for these signals — they'll tell you when the thesis is accelerating, stalling, or breaking.

5 Key Takeaways

1.  The bottleneck framework is the most reliable repeating pattern in technology investing. AI's supercycle has moved from compute to memory to photonics to real estate to power — and each transition created enormous, time-limited opportunities for investors paying attention to the constraint, not the consensus.

2.  A major Wall Street research desk has revised AI power demand growth to 220% by 2030 — the equivalent of adding a new top-ten energy-consuming nation to the planet. Hyperscalers are reinvesting 87% of operating cash flow into capex and R&D. The buildout is not stopping.

3.  The market is pricing in electricity cost risk. The real binding constraint is licensed electricians who take four years to train and aren't in the pipeline in sufficient numbers. This creates structural, multi-year pricing power for T&D construction firms — a consensus-busting conclusion with direct investable implications.

4.  Behind-the-meter natural gas is the bridge fuel everyone is deploying and nobody wants to advertise. Simple-cycle gas generators are going in now, ahead of grid interconnects. Gas producers and turbine manufacturers benefit today — not when the idealized grid eventually catches up.

5.  Don't chase the headline. Follow the bottleneck. By the time NVIDIA is on magazine covers, the money is in the wires and the workers. The companies doing the unglamorous work of connecting AI to physical reality — Quanta, MYR, GE Vernova, Eaton, Powell, Vertiv — are where the next chapter of this trade lives.

The Moon Was Never About the Moon

Control of the high ground has always shaped what happens below it. Apollo was the
opening move. Artemis is the next — NASA sending crews back to the Moon and building
the infrastructure to keep them there. Satellites already run global supply chains,
communications, and economies. Nations are spending heavily to protect that access.
The Tuttle Capital Space Industry Income Blast ETF (SPCI) holds 11 pure-play names
across that stack. Could SPCI have a place in a diversified portfolio?

Learn More at incomeblastetfs.com/etf/spci.

Back in the dot com bubble you saw a number of companies add .com to the end of their name and their stock prices went parabolic. Of course a lot of them went bankrupt. Yesterday Allbirds, a shoe company, announced they were pivoting to AI…..

This happened……

At the same time the market closed at all time highs even though we are still in a war, oil is still elevated, and interest rates are still higher. I just saw a note that only 12 stocks in the S&P 500 are at 52 week highs, while the index itself is. There have only been a handful of similar readings, mostly in the late 90s and early 2000s. Are we in a bubble? Maybe. How do you handle it? While it’s easier to chase companies having parabolic up moves like the memory and optics names, you also have to balance that with companies making strong countertrend moves. The software names are popping again for example…..

Meanwhile, look at the move that SNDK has had this year……

Could it keep going? Sure, look at what I wrote above about bottlenecks. I’d just balance this out with names making countertrend moves, and of course the HALO type names I’ve been writing about. Oh, and always have hedges.

In much more important news, ASML reported earnings and sales jumped 13%. It also raised it’s full year outlook. This is the first AI related company to report earnings, so this is a good sign.

I talked about NVDAs quantum announcement and the quantum stocks yesterday, they are all up big again this morning.

What Iran Tells Us About UFO Disclosure

When governments confront unknown threats in their airspace, defense budgets surge
and the same aerospace and surveillance companies move hardest. On March 2nd,
Northrop jumped 6% and Lockheed 3.3% on the Iran news — and President Trump has
since ordered the formal release of government UAP files, with the Pentagon confirming
compliance. So if a conventional conflict can move these stocks this fast, what happens
when the bigger story breaks?

See the UFOD holdings: [thetruthisoutthereufod.com

$MEMY Holdings Update:

We replaced $RTX ( ▲ 1.9% ) $AU ( ▲ 3.6% ) and added $OKLO ( ▲ 11.57% ) $TTD ( ▼ 3.2% ) All 5% positions.


For a full list of MEMY holdings, visit:

https://incomeblastetfs.com/etf/memy

Distributor: Foreside Fund Services, LLC

$OKLO ( ▲ 11.57% )

In this environment I want to continue to find areas and stocks coming off bottoms and making buyable counter trend moves. OKLO made undercut and rally moves at most of it’s important moving averages, it’s also up 7%ish pre market.

The H.E.A.T. (Hedge, Edge, Asymmetry and Theme) Formula is designed to empower investors to spot opportunities, think independently, make smarter (often contrarian) moves, and build real wealth.

The views and opinions expressed herein are those of the Chief Executive Officer and Portfolio Manager for Tuttle Capital Management (TCM) and are subject to change without notice. The data and information provided is derived from sources deemed to be reliable but we cannot guarantee its accuracy. Investing in securities is subject to risk including the possible loss of principal. Trade notifications are for informational purposes only. TCM offers fully transparent ETFs and provides trade information for all actively managed ETFs. TCM's statements are not an endorsement of any company or a recommendation to buy, sell or hold any security. Trade notification files are not provided until full trade execution at the end of a trading day. The time stamp of the email is the time of file upload and not necessarily the exact time of the trades. TCM is not a commodity trading advisor and content provided regarding commodity interests is for informational purposes only and should not be construed as a recommendation. Investment recommendations for any securities or product may be made only after a comprehensive suitability review of the investor’s financial situation.© 2026 Tuttle Capital Management, LLC (TCM). TCM is a SEC-Registered Investment Adviser. All rights reserved.