Four companies just committed more money to AI than the GDP of Sweden, Norway, and Denmark combined. The receipts are in. The returns are not.

The number that should make every investor nervous

On Wednesday April 29, Alphabet, Amazon, Meta, and Microsoft all reported earnings on the same night. The combined capex figure that emerged was staggering: $725 billion in AI infrastructure spending for 2026 alone. That is a 77% jump over last year's already record-breaking $410 billion. I have been watching tech earnings for years and I have never seen a number that large treated so casually.

My honest take: this is not confidence. This is a collective panic buy. Every hyperscaler is terrified of being the one that underspent, so they are all spending at a scale that defies any normal return-on-investment logic. The market is cheering. I think that is a mistake.

Here is the specific problem with Microsoft. CFO Amy Hood disclosed $190 billion in 2026 capex, of which $25 billion is purely from rising component prices. Memory costs have more than tripled since last year. IDC forecasts DRAM at $9.71 per gigabyte in 2026, up from $3.76 in 2025. Microsoft spent roughly $97 billion on infrastructure over the last four quarters and generated $37 billion in AI annual recurring revenue. That math does not close.

“For calendar year 2026, we expect to invest roughly $190 billion in capital expenditures, which includes approximately $25 billion from the impact of higher component pricing.”

— Amy Hood, CFO, Microsoft — Q3 2026 Earnings Call

Hood also warned that even after all this spending, Microsoft "expects to remain constrained at least through 2026." Read that again. They are spending $190 billion and still cannot meet demand. That is not a success story. That is a supply crisis dressed up in earnings-call language.

Google is the one winner here and that gap is widening fast

To be fair, one company actually showed its work. Google Cloud grew 63% year over year to $20 billion in a single quarter. Operating margins nearly doubled from 17.8% to 32.9%. The cloud backlog nearly doubled sequentially to $462 billion. Alphabet CEO Sundar Pichai said cloud revenue would have been even higher if they had more compute capacity. That is a genuinely good problem to have.

This is the one "good edge" moment in this whole story. Alphabet is proving the model works when you own your silicon roadmap. Google's custom TPU chips give it cost control that Microsoft, which relies almost entirely on third-party suppliers, simply does not have. That $25 billion component inflation hit at Microsoft is the direct cost of not owning your hardware stack.

Now for the counterpunch. The bulls will tell you demand is real and the spending is justified. Azure grew 40%. AWS hit $37.6 billion in quarterly revenue. Enterprise AI adoption is accelerating. Wedbush analysts say investors are still underestimating how fast these companies are monetizing AI. I do not fully buy that argument. Strong cloud growth is real. But cloud growth was already happening before the AI arms race began. The question is whether the incremental $300 billion in new spending this year produces incremental returns. Nobody has answered that yet.

OpenAI missed targets and the whole ecosystem felt it

The week got darker before earnings even dropped. A Wall Street Journal report revealed OpenAI may be on track to miss key revenue and user targets, pulling the Nasdaq down 1% overnight. Fortune reported that OpenAI's CFO Sarah Friar is worried the company is spending too much on data centers relative to the revenue it generates. This matters because OpenAI is the demand signal the entire AI infrastructure bet is built on.

When was the last time you saw a company's internal CFO publicly signal concern about its own spending trajectory and the stock market just shrugged? That is what happened here. The market is so committed to the AI narrative that even warning signs from inside the most important AI company alive barely registered.

Layoffs plus $725 billion in spending is not a contradiction, it is the strategy

Here is the part that should bother people more than it does. Meta announced 8,000 layoffs while simultaneously raising its capex forecast to $145 billion. Microsoft is cutting headcount year over year while committing $190 billion to infrastructure. These companies are not investing in people. They are investing in machines and betting that the machines will replace the people they are firing.

This is unserious as a long-term economic argument. You cannot fire your customers. The workers being laid off are also the consumers who buy software subscriptions, cloud services, and AI tools. The productivity gains from AI need to flow somewhere real or the demand side of this equation collapses.

Meanwhile, Amazon's free cash flow is heading negative as it chases $200 billion in capex. Barclays analysts are modeling negative free cash flow for Meta in 2027 and 2028. These are not small companies with reckless management. These are the most profitable businesses in human history choosing to burn cash at a rate that would terrify any normal CFO.

“We are now modeling negative FCF for '27 and '28, which is somewhat shocking to us but likely what we eventually see for all companies in the AI infrastructure arms race.”

— Barclays Analysts, note following Meta earnings

The war in Iran is making all of this more expensive and nobody is talking about it

There is a geopolitical layer here that Wall Street keeps trying to ignore. The U.S. began combat operations in Iran in late February, which spiked oil prices and disrupted helium supply chains critical to semiconductor manufacturing. Brent crude is now above $111 per barrel. Data center energy costs are rising. The Iran conflict is adding inflation to an already overheated infrastructure buildout.

I remember watching the dot-com bubble inflate in real time and thinking the numbers would eventually make sense. They did not. I am not saying this is 2000. The revenue is real, the cloud demand is real, and these companies have actual cash flows. But the gap between what is being spent and what is being earned is widening, not closing. That gap is the risk nobody wants to price.

The optimists say the only risk right now is underspending. I think the real risk is that $725 billion buys a lot of capacity that the world is not yet ready to use.