Ed Zitron says AI needs trillions a year, and HN can't refute the math

Ed Zitron's math: AI needs $2–3 trillion a year by 2030, and nobody can find it

Ed Zitron's latest broadside, a 3,000-word newsletter post, hit 456 points and 478 comments on HN — and the top twenty comments are arguing about Zitron, not the data. That's the real story: a financial argument that even AI skeptics on the thread concede is "factually correct," wrapped in rhetoric so caustic that half the discussion is about whether you can take the messenger seriously at all.

The core claim: the infrastructure bill has no customers

Strip away the tone and Zitron's spine is arithmetic. The roughly 190GW of planned AI data centers, he argues, cost $9.5–15 trillion at current rates and require $500 billion to $1 trillion a year in fresh debt issuance — against the ~$250 billion the market currently absorbs annually. To service that, the AI sector would need to be generating $2–3 trillion in annual revenue by 2030. Anthropic and OpenAI together project around $358 billion in 2029 revenue against a need closer to $1.75 trillion from data-center usage alone.

His sharpest new evidence is what happened after both labs moved to token-based billing in Q1 2026 and customers could finally see the meter:

• Uber burned its entire annual token budget in a single quarter, then capped spending at $1,500 per employee per month
• Brex engineers were limited to $500 a week in tokens
• T-Mobile temporarily capped usage at $2,000 monthly per user
• Only 26% of companies say they have a comprehensive view of their AI costs; 22% report no visibility at all

The thesis: AI demand was inflated by opacity, and the moment the bill became legible, customers recoiled. He closes by teasing a story "out in the next two weeks" that will "confirm the absolute worst fears of the AI industry."

The thread can't refute the numbers, so it litigates the man

The highest-signal pushback didn't dispute a single figure — it attacked the framing. A WebAssembly co-founder and former V8 compiler tech lead quoted Zitron's own line calling AI believers "liars, cowards, imbeciles, craven boosters":

"Who writes like this? When you lead with 'everyone who doesn't agree with me is a lying cheat coward imbecile' I think we should just turn the volume down on you to zero." — WebAssembly co-founder, ex-V8 compiler lead

Others went after the track record directly:

"The writer has been posting popular but consistently wrong takes for 2+ years … arguing that AI is failing, is a waste of money, is bad, will never work." — an engineer linking Zitron's March 2024 'Peak AI' post

"Zitron is begging for a collapse at this point. Yes, his macro analysis correctly identifies a massive financial risk but his incessant pessimism completely misses the incredible ground-level utility." — a commenter conceding the macro case while rejecting the conclusion

The Uber counter, and why the bears don't blink

The recurring rebuttal is historical: we've heard this before. One commenter pulled out the obvious analogy — the same screeds were written about Uber, which now posts $50bn in revenue and turns a profit.

"I'm reminded of identical screeds I read about Uber a few years ago … Now it's 2026. Uber still exists, has revenues of $50bn and is apparently a highly profitable business." — a skeptic of the skeptics

But that comparison got picked apart in the replies — Uber subsidized fares to kill local taxi monopolies and then raised prices on a captive market, a playbook that doesn't obviously map onto a sector where cheaper open-weight Chinese models are actively undercutting the incumbents. And the bears had a clean counter to the "utility" defense:

"If the AI companies need $X billion in revenue to stay afloat, it doesn't matter if 0.5% or 5% or 50% of that revenue is from transforming the State of the Art … what matters is that these companies won't have the income to pay their bills." — a commenter separating utility from solvency

Even a self-described AI user who finds Zitron "obsessive and biased" landed on the same place: approaching AI "from the angle of economics seems correct," given the scale of investment.

"His rhetoric is a bit obsessive and frankly biased against AI. That said, I think his voice is useful as a counter to the mainstream opinion." — a founder working on a solo-dev framework

The asterisk: a teaser, not a bombshell

The most deflating read came from a commenter who actually checked the sources against the headline:

"Buried lede … the sources don't seem to back the title either … this takes clickbait to new lows. The headline is trying to sell the teaser here, with very limited meat in the middle." — a reader who chased the citations

That is the honest summary of where this lands. The financial argument — circular financing, a revenue gap measured in trillions, and customers flinching the instant billing turned transparent — is the part even Zitron's critics won't touch. The collapse he keeps promising is still two weeks away, as it has been for two years. Whether you read "AI is slowing down" as analysis or as a man who has built an identity around a crash depends almost entirely on whether the next NVIDIA earnings call proves the math right before the rhetoric exhausts its audience.