Wall Street Just Said AI Spending Is Still Justified — Let Me Show You the Fine Print

Wall Street Just Said AI Spending Is Still Justified — Let Me Show You the Fine Print

Let me tell you something that's been sitting with me since yesterday's Bloomberg segment. Lazard Asset Management portfolio manager Celine Woo went on camera and said hyperscalers are still justified in ramping up AI spending because demand for cloud infrastructure keeps outpacing supply. And you know what? She's not wrong. That's the part that should scare you.

I've been running hosting infrastructure for over a decade, and I've watched this industry swing from "the cloud is dead" to "the cloud is everything" more times than I can count. But this is different. This isn't a marketing cycle. This is a $2.4 trillion question hiding behind a perfectly reasonable earnings call, and every independent hosting provider on the planet needs to understand what's actually being said.

The Bull Case — Demand Really Is Outpacing Supply

Let me give Woo her due first. She anchored her argument on Microsoft and Amazon's latest earnings, and those numbers are genuinely strong. Microsoft reported fiscal Q4 revenue of $90 billion, up 18% year over year, with operating income up 18% to $40.6 billion. Azure grew 43% — ahead of the 39.6% consensus — and crossed $100 billion in annual revenue for the first time.

Here's the line that matters, straight from Microsoft's own commentary: customer demand continues to exceed available supply. Not "demand is healthy." Not "demand is growing." Demand exceeds supply. That is the single most important sentence in the entire bull case, because it means the constraint right now is physical — there literally is not enough compute to sell.

Amazon told the same story. AWS grew 37% year over year to $42.2 billion, crushing the 31% expectation, and margins are climbing. CEO Andy Jassy raised the 2026 capex outlook to about $220 billion, up from the $200 billion guidance at the start of the year. The stock jumped 15% on the print. When both of the biggest cloud providers on earth are saying they can't build fast enough, a portfolio manager looking at that and saying "keep spending" is not being reckless. She's reading the same demand signals I read every day when I see colocation prices climbing and GPU lead times stretching.

The Numbers Behind the Confidence

So let's put the bull case in full context. The four largest hyperscalers — Alphabet, Amazon, Meta, and Microsoft — are on track to spend roughly $725 billion on AI infrastructure in 2026. That's up 77% from $410 billion in 2025. Combined, that's more than the GDP of most countries on earth.

And here's the thing the market keeps missing: the earnings are real. This isn't 2021, when WeWork-style storytelling was carrying valuations. Azure is producing actual revenue at $100 billion annual run rate. AWS grew 37%. The demand exists, the revenue exists, and the customers are lining up. If you only read the earnings headlines, Woo's conclusion is obvious and correct.

But I didn't run a hosting business for a decade by reading headlines. I read footnotes. And the footnotes to this bull case are where the story gets interesting.

The Fine Print — $2.4 Trillion of Promises

Bloomberg reported last week that the four largest players in the data center race have committed nearly $2.4 trillion in spending over the coming years. Not projected. Not hoped. Committed — signed purchase agreements, contractual obligations, and leases.

Break that down and it gets wild. Alphabet alone disclosed $902 billion of purchase commitments, contractual obligations, and leases that haven't even commenced. That's more than eight times higher than a year earlier. Meta reported nearly $700 billion in future spending — and about half of that is data center leases that haven't started yet, to be paid off over periods as long as 30 years.

Read that again. Thirty-year leases for infrastructure that doesn't exist yet, in an industry where the hardware becomes obsolete every two to three years. That's not a demand signal. That's a promise written in a financing document. Woo is right that demand outpaces supply today. But these commitments aren't today — they're a bet on what demand looks like in 2036.

The Counter-Argument — Negative Free Cash Flow at the Richest Companies on Earth

Here's the part nobody on the bull side wants to say out loud: Alphabet and Amazon have both tipped into negative free cash flow. Meta is expected to follow. The richest, most profitable companies in human history are now spending more cash than they generate, and they're doing it to build capacity they've already committed to.

Now, I've been around long enough to know negative free cash flow isn't automatically a death sentence. Amazon ran negative FCF for years and built a trillion-dollar business. But there's a difference between investing during growth and borrowing to fund promises. When you've got $902 billion in commitments on the books, and your free cash flow is negative, the only way to fund the gap is debt — and we've already seen what that does. Oracle got downgraded to BBB- this month, its stock down 64% from peak, with $165 billion in planned data center pipeline blocked. The bond market has started charging stressed borrowers 10% or more. That's not a blip. That's the credit cycle waking up.

The Zitron crowd calls this circular — OpenAI and Anthropic renting compute from the same hyperscalers who fund them, with UBS estimating the big labs are burning through $140 billion a year. They're not entirely wrong, but they're also not entirely right. The demand is real. The question is whether the demand can grow fast enough to fill $2.4 trillion of commitments before the financing costs eat the whole thing.

What This Actually Means for Independent Hosting Providers

First — keep your powder dry on capacity expansion. The bull case says demand is real today, and it is. But you're not Amazon. You can't absorb a 30-year lease that goes sideways. Expand into what you can fill in 12 to 18 months, not what a PowerPoint says you'll need by 2030.

Second — watch hyperscaler free cash flow like a hawk. That's the canary. Every hyperscaler earnings call, check the FCF line before the revenue line. When the giants stop generating cash, they do one of two things: they raise cloud prices, or they slow the buildout. Both change your competitive position. Price your services now against a world where cloud prices go up — because that's the world where independent hosting looks cheap by comparison.

Third — lock your hardware and power contracts while you still can. The same demand that justifies $725 billion of capex is also keeping GPU lead times long and colocation pricing high. If you've got a supplier relationship, extend it now. Don't wait for the 2027 pricing cycle, because if the hyperscalers keep spending, hardware stays tight — and if they cut, you'll be glad you locked in before the panic.

Fourth — position yourself as the capital-light alternative. This is your structural advantage and it's getting bigger every quarter. The giants are levered to the teeth with 30-year promises. You can respond to market reality in months, not decades. Sell that. Customers who watched their cloud bill double are already listening.

The Structural Reality — Both Things Are True at Once

Here's the honest truth, and it's uncomfortable for everyone: Woo is right, and the bears are right, at the same time. Demand for compute genuinely outpaces supply today. And the commitments made to chase that demand are structurally too big for the current financing model. Both statements are true. The market keeps trying to pick a side, and that's why you see these violent swings — Palantir up 93% one week, Oracle down 64% over a month, Google's stock dropping on a $205 billion capex raise that should have been good news.

For independent providers, this isn't a puzzle to solve. It's a weather pattern to navigate. The buildout isn't stopping — communities, regulators, and grid constraints are slowing it, but the money keeps flowing. What's changing is who's financing it and at what cost. The era of free money for AI infrastructure is over. The era of expensive money is here, and that's actually good news for anyone who doesn't need to borrow.

The Bottom Line

So yes, Lazard is right that AI spending is still justified. Demand is outpacing supply, the revenue is real, and the buildout continues. But "justified" is not the same as "safe," and "demand today" is not the same as "$2.4 trillion of commitments tomorrow." The fine print is where the next crisis lives, and the next opportunity too.

Keep your balance sheet clean, keep your contracts locked, and keep watching that free cash flow line, ent? Because the hyperscalers just told us they're going to keep spending. The only question is who's left standing when the music stops — and if you play it smart, it'll be you.

-- Allan Ali, Founder

This article was produced with AI-assisted research and editorial support. Reporting is based on sources cited in the article.