데이터센터 대환상: 월가가 세고 있는 기가와트는 절대 실현되지 않는다

THE DAILY H.E.A.T. Torching Wall Street’s Obsolete Playbook. Delivered straight to your inbox 5 days a week. Subscribe Home Posts The Great Data Center Mirage: Wall Street Is Counting Gigawatts That Will Never Exist The Great Data Center Mirage: Wall Street Is Counting Gigawatts That Will Never Exist Apr 15, 2026 I’ve been a trader and investor for 44 years. I left Wall Street long ago—-once I understood that their obsolete advice is designed to profit them, not you. Today, my firm manages around $4 billion in ETFs, and I don’t answer to anybody. I tell the truth because trying to fool investors doesn’t help them, or me. In Daily H.E.A.T. , I show you how to Hedge against disaster, find your Edge, exploit Asymmetric opportunities, and ride major Themes before Wall Street catches on. Table of Contents Live Webinar: The Space Trade H.E.A.T. News vs. Noise: What’s Moving Markets Today ETF News In Case You Missed It Live Webinar: The Space Trade How to capture growth AND income in 2026’s surging orbital economy. April 16th, 2026, 2PM ET. Click Here to register H.E.A.T. Yesterday, Open AI exited a Project Stargate data center project in Norway. Five days ago it paused a project in the UK, and last month it paused an expansion in Texas. If this was a public company, this news would be reverberating through the market. A trillion dollars of AI orders won't matter if the building can't get power. That is not a metaphor. It is a physical constraint that is playing out right now across 140 data center projects scheduled to open in the United States this year. And it is the most important thing the market is not pricing. Here is the most dangerous word on Wall Street right now: "announced." Every major hyperscaler — Amazon, Microsoft, Google, Meta — has made headlines promising a combined $650 billion-plus in 2026 data center capex. Analysts have dutifully modeled out the earnings. Infrastructure stocks have been bid up on the premise. The AI buildout narrative has become consensus gospel. There is just one problem. Half of those data centers will never open. Not in 2026. Many not ever. That is not hyperbole. That is the Sightline Climate 2026 Data Center Outlook — the most comprehensive ground-level tracking of US data center construction — and it says this plainly: of the 16 gigawatts of capacity slated to come online this year across 140 projects, 30 to 50 percent will be delayed or outright canceled. As of today, only 5 gigawatts are actually under construction. The market is pricing in the announcement. It has not priced in the friction. That gap is where your opportunity lives. The Consensus Gospel: Capex Equals Capacity The institutional narrative on AI infrastructure has been elegantly simple: the hyperscalers are spending historic sums, therefore historic compute capacity is coming, therefore every company in the supply chain will print record revenues. Buy the shovels. This thesis is not wrong. It is incomplete in a way that is going to be very expensive for investors who have not looked past the press releases. Capital commitment and physical capacity are not the same thing. Between a CEO's earnings call promise and a data center actually going live stands a gauntlet that Wall Street has collectively chosen to ignore. The numbers look impressive on paper. JPMorgan estimates the full AI infrastructure cycle will require $5 trillion in total investment. The US government faces a funding gap of over $1 trillion just to close the grid modernization deficit. Hyperscalers have committed to deploying more capital in 2026 than at any point in the history of corporate America. And yet, as Canaccord Genuity analyst George Gianarikas writes, the American data center boom is hitting a formidable wall of logistical friction. The question is not whether the money exists. The question is whether the transformers, the permits, the power grid, the community acceptance, and the supply chains can convert that money into actual silicon. They cannot. Not at the pace the market is assuming. The Crack in the Foundation: Five Friction Points the Market Has Missed 1. The Transformer Is the New Chip — and You Can't Get a Slot For years, power transformers were a boring utility line item — the steel box in the substation that nobody in tech bothered to think about. That era is over. Modern hyperscale data centers require transformers dramatically larger than anything the industry ordered a decade ago. Before 2020, high-power transformers arrived 24 to 30 months after an order was placed — a timeline GE Vernova CEO Philippe Piron calls "totally manageable in the old world." Today, the spike in demand from data centers combined with grid electrification from EVs and heat pumps has pushed delivery windows to as much as five years on some components. AI companies need delivery inside 18 months. The math does not work. But here is what makes this worse than a standard supply-demand mismatch: you are not just ordering equipment. You are competing for a slot in a manufacturing queue that is measured in years, not quarters. And without that slot — without that transformer — nothing else matters. The racks are installed. The GPUs are racked. The fiber is run. And the building cannot turn on because the steel box the size of a truck has not arrived. The chips do not run. The revenue does not show up. The earnings miss lands eighteen months after the capex was spent. Because domestic manufacturing was systematically offshored over the past three decades, the US now finds itself in an absurd position: dependent on imports — including from China — to win the AI race against China. The tariff environment compounds this. Some builders, like Crusoe, have resorted to refurbishing transformers pulled from shuttered power plants as a stopgap. When your cutting-edge AI infrastructure depends on repurposed hardware from decommissioned facilities, you do not have a temporary bottleneck. You have a structural problem. 2. Grid Connection Is a Five-Year Wait, Not a Six-Month Process Only 53 percent of planned 2026 data center capacity is grid-connected. Another 25 percent has not even disclosed its power strategy. This is because connecting to the grid — in many parts of the country — now requires joining an interconnection queue that stretches years into the future. Transmission upgrades triggered by new industrial loads have backlogged interconnection studies across every major regional grid operator. This reality has birthed a new model: the gigascale, grid-independent campus. New Era Energy & Digital's 7-gigawatt project in Lea County, Homer City's 4.5-gigawatt coal-to-gas redevelopment in Pennsylvania, Crusoe's 1.8-gigawatt natural gas and renewables complex in Wyoming — these are not data centers in the traditional sense. They are small power utilities that happen to run AI workloads. The capital required to fund your own generation plant is staggering, and it is only available to a handful of players. The rest are waiting in line. 3. Permitting Has Become a Political Battleground Community resistance to data centers is no longer a fringe concern. Maine's House of Representatives passed an outright moratorium on large-scale data centers until 2027 by an 82-to-62 vote. The goal was not to kill innovation — legislators were explicit about that — but to force a planning pause to evaluate environmental and resource impacts. Variants of this conversation are now happening in Virginia, Texas, and Arizona. The sociopolitical headwind is deeper than zoning fights. A Quinnipiac University poll shows public sentiment has shifted materially against AI's integration into daily life, healthcare, and education. The firebomb attack on Sam Altman's home last week — whatever its specific motivation — is a symbol of a broader cultural backlash. When AI simultaneously drives up electricity bills and competes for jobs, you create a combustible combination. Wall Street has not modeled the regulatory response to a Luddite resurgence. That is an analytical gap with real earnings consequences. 4. The 2027 and 2028 Pipelines Are Even More Exposed If the 2026 situation is concerning, the forward pipeline verges on fiction. Of the 21.5 gigawatts announced for 2027, only 6.3 gigawatts have broken ground. For 2028 through 2032, the numbers are worse: 37 gigawatts of announced infrastructure have not received a firm completion date, with only 4.5 gigawatts of that total having begun any construction work. Think of the following table as a directional map, not a precise forecast: announced pipeline versus realistic energized capacity given interconnection queues and equipment lead times. The gap is the trade. Year Announced GW Under Construction GW Likely Delay Rate Effective Gap 2026 16.0 GW 5.0 GW 30 – 50% 5 – 8 GW at risk 2027 21.5 GW 6.3 GW Est. 55%+ 12+ GW at risk 2028 – 2032 37.0 GW (firm date) 4.5 GW Est. 70%+ 30+ GW at risk Sources: Sightline Climate 2026 Data Center Outlook; Canaccord Genuity; Bloomberg; Futurism 5. Trump's Nuclear Renaissance Is Still Purely Rhetorical The administration has promised a nuclear revival as the long-term answer to AI's power needs. So far, not a single new nuclear plant has broken ground under this policy posture. Small modular reactors — the genuine ray of hope in the power generation landscape — remain years away from commercial deployment at any meaningful scale. The gap between political promise and installed generation capacity is measured in gigawatts and grows wider every quarter. The Mechanism: How Mispriced Friction Flows Through the Market The market has made a category error. It has treated announced capital expenditure as a proxy for completed infrastructure. These are not the same variable, and the delta between them is large enough to matter for earnings estimates across multiple sectors. Consider the chain reaction. A hyperscaler announces $20 billion in data center investment. Analysts model the revenue implications for server manufacturers, cooling companies, power distribution providers, and data center REITs. Stocks move in anticipation. But if 40 percent of that capacity does not open on schedule — because the transformer slot is two years out, or the permitting board voted no, or the interconnection queue is full — then every downstream revenue estimate is wrong. Not marginally wrong. Wrong by the percentage of announced capacity that never materializes, compounded by the 12 to 18 month lag between the miss and when it appears in earnings. Goldman Sachs Executive Director Shreeti Kapa captured the disconnect after a recent dinner with technology investors: the consensus among sophisticated Valley insiders is that every major player is "acutely compute constrained" and the bottlenecks are "real and here to stay." The question she posed is the right one — if this is already consensus in Silicon Valley, is the rest of the market catching up, or does the scale of AI demand so thoroughly dominate investor imagination that the physical constraints never fully register? The answer is both. The demand for AI compute is real and growing. But the supply-side frictions are also real and growing faster than the market's model. The spread between those two curves is the trade. SPOTLIGHT: The Power-Up Infrastructure Trade — Two Years Later Two years ago, the thesis was straightforward: the hyperscaler capex wave would ultimately be gated by power generation and grid infrastructure. Buy the enablers, not just the AI darlings. That basket — utilities, grid equipment manufacturers, high-voltage contractors — outperformed both the AI data center basket and the TMT AI basket through 2024 and much of 2025. The underperformance since then reflects a market that began to doubt whether US energy infrastructure could ever scale to meet demand. That doubt has become consensus — which is precisely when it becomes investable again. The friction is not shrinking. It is institutionalizing. That means the premium for companies who can actually deliver electrons to chips is just beginning to be priced. In 2026, the company that controls the transformer slot controls the AI timeline. Winners, Timing Traps & Pressure Points Category Company / Ticker Why It Matters WINNER Eaton Corp (ETN) Largest domestic transformer and power management manufacturer. Every delayed data center is a future backlog item. The slot shortage is Eaton's pricing power. WINNER GE Vernova (GEV) Grid electrification leader. CEO Philippe Piron quoted directly in the Sightline report. When lead times stretch to five years, GEV's backlog swells accordingly. WINNER Quanta Services (PWR) High-voltage grid buildout contractor. Grid complexity is revenue. Every interconnection upgrade, every transmission line, every substation expansion runs through Quanta's labor force. WINNER Vistra Energy (VST) On-site and hybrid power demand surges when grid connection queues stretch beyond five years. Gas and nuclear independent power producers collect the premium. WINNER Constellation Energy (CEG) Nuclear baseload is the only 24/7 carbon-free answer to gigascale data centers that need guaranteed uptime. Regulatory moat deepens as permitting gridlock worsens. WINNER Comfort Systems USA (FIX) Mechanical and electrical contractor for data center buildouts. Smaller relative to ETN and GEV means more price leverage as demand outstrips labor supply. TIMING TRAP Digital Realty (DLR) REIT valued on a capacity pipeline that is 30-50% at risk of delay. Capex is immediate. Revenue is deferred. That spread has not been priced into the multiple. TIMING TRAP Equinix (EQIX) Same dynamic as DLR. International exposure adds the OpenAI Stargate UK pattern: committed capex, paused deployment, earnings impact arrives before analyst estimates adjust. TIMING TRAP Super Micro Computer (SMCI) Server revenue tracks data center openings, not announcements. If half of 2026 builds delay, near-term revenue misses follow with roughly 12-18 months of lag. TIMING TRAP Vertiv Holdings (VRT) Data center cooling and power distribution revenue is tied to new center completions, not signed LOIs. Valuation reflects announced pipeline, not the energized version. TIMING TRAP NuScale Power (SMR) SMR is the hoped-for nuclear fix — but years away from practical scale is not a 2026 or 2027 catalyst. Long-dated optionality priced as near-term operating reality. PRESSURE Amazon (AMZN) AWS capex is committed and not reducible. Delays don't cut spending — they compress return on that spending and push revenue timelines out. Watch guidance, not capex. PRESSURE Microsoft (MSFT) $80B 2025 capex for Azure. Permitting and equipment shortages do not distinguish hyperscalers from co-location operators. The constraint is physical, not financial. PRESSURE Alphabet (GOOGL) Google Cloud. Part of the $650B+ 2026 combined hyperscaler commitment. Compute capacity risk translates directly to cloud revenue growth risk in the outer quarters. BEAR CASE: What Could Make This Thesis Wrong Emergency reshoring works faster than expected. If the administration invokes the Defense Production Act or similar tools to accelerate domestic transformer manufacturing, lead times could compress faster than the current trajectory suggests. The political will exists; the industrial capacity does not yet. Hyperscaler scale rewrites procurement rules. When a company is spending $80 billion a year on infrastructure, it has the capital to vertically integrate supply chains, manufacture its own transformers, and build proprietary generation assets. Amazon and Microsoft are large enough to create their own physics — but only for themselves, not for the broader market. Efficiency gains reduce the power requirement. Breakthroughs in model efficiency — like the disruption caused by DeepSeek earlier this year — could meaningfully reduce the compute intensity required per dollar of AI output. If the next generation of models runs on a fraction of the power, the grid constraint loosens and the thesis softens. Five Takeaways for Investors • 1. Announced is not built. Nearly half of 2026's planned US data center capacity will not come online on schedule. Downstream revenue models that treat the announced pipeline as the deployed pipeline are systematically overstated. • 2. The transformer is the critical path. Delivery windows stretching to five years on large power transformers are the single biggest physical constraint on the buildout. You are not ordering equipment — you are competing for a slot. Companies holding those slots, primarily Eaton and GE Vernova, have structural pricing power that is only beginning to be reflected in their multiples. • 3. The Power-Up infrastructure trade is re-entering its window. After underperforming the AI tech basket in late 2025 as the market began to doubt the buildout thesis, grid infrastructure and power generation stocks are returning to levels that look compelling relative to the certainty of the underlying demand. The doubt created the entry. • 4. Data center REITs are carrying a timing trap, not a demand problem. DLR and EQIX do not face a question of whether tenants want space. They face a question of when that space can be energized. The market has priced the demand. It has not priced the delivery schedule risk. Capex is immediate. Revenue is deferred. That spread matters. • 5. The sociopolitical risk is not a tail event. State-level moratoriums, attorney general investigations, and a measurable polling shift against AI infrastructure are observable trends, not hypothetical scenarios. Maine's House passed a data center moratorium 82 to 62. Model accordingly. The AI revolution is real. The capital is real. The demand for compute is real. But between the announcement and the ribbon-cutting stands a gauntlet of transformer slots, interconnection queues, permitting boards, community opposition, and supply chains that the market has collectively chosen to treat as background noise. Wall Street is counting gigawatts that will never exist — at least not on the timeline embedded in current valuations. The friction is not temporary. It is structural. And structural friction, correctly positioned, is the best kind of investment thesis. The companies that move electrons from the grid to the chip are going to be worth a great deal more than the market currently believes. The companies that merely count on the chips arriving on schedule are going to disappoint. AI isn't constrained by compute anymore. It's constrained by permission slips and metal boxes. Make sure you're on the right side of that. 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. News vs. Noise: What’s Moving Markets Today Market’s continued to rally yesterday. One of my favorite sayings is “trade based on what you see not what you think”. I have a hard time believing that the market should be back where it was before the war while oil is in the 90’s, interest rates are higher, and private credit issues are still lingering. I shared some thoughts with Bloomberg yesterday…… - BNN Bloomberg Matthew Tuttle, CEO and CIO of Tuttle Capital Management, joins BNN Bloomberg to discuss the markets amid oil price uncertainty. BNN I shared on Twitter some thoughts about tail risk hedging, which hopefully won’t come in handy….. A bright spot yesterday was PPI coming in slightly cooler than expected. It is in an uptrend thought and yesterday’s reading was the highest in 3 years. We are sitting on a couple of potential powder kegs, and stagflation is a big one. Nothing to worry about yet, but something to keep an eye on. One area that jumps out to me is quantum, NVDA announced this yesterday….. NVIDIA Ising — AI Models for Quantum Computing Open-source AI accelerating quantum calibration & error correction NVIDIA That got the quantum stocks running and they are all up big this morning as well. 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 ETF News $MEMY Holdings Update: We replaced $BTU ( ▲ 1.23% ) and added $TEM ( ▲ 13.89% ) All 5% positions. For a full list of MEMY holdings, visit: https://incomeblastetfs.com/etf/memy Distributor: Foreside Fund Services, LLC A Stock I’m Watching $INFQ ( ▲ 6.23% ) This is moving off the NVDA quantum news. All sorts of entry points, you had an undercut and rally, then a couple of moving average undercut and rallies. Up another 6.75% pre market so probably a bit extended. Full disclosure, we own this in $MEMY ( ▲ 1.85% ) In Case You Missed It 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. Keep Reading Mar 31, 2026 • 18 min read The Shadow Bank Runs America's Corporate Debt — And Nobody Is Minding the Store Apr 7, 2026 • 17 min read The Strait Doesn't Reopen.Now What? Mar 24, 2026 • 11 min read The Market May Have Just Handed You a Gift. Most Investors Are Too Confused to Pick It Up. View more