Sam Altman Admits AI's Awkward Question: 'Where Is the Revenue?'

Answer-First Lead

Sam Altman is hearing the same question from CEOs worldwide: “Where is the revenue?” As companies pour billions into AI infrastructure, executives are seeing costs go “up and up” without meaningful productivity gains or financial returns. New data shows 82 cents of every $1 in AI spend never makes it to production — and 90% of executives report no meaningful productivity impact from AI tools.

🔍 THE BOTTOM LINE

AI’s economic promise is colliding with enterprise reality. Companies are burning through annual token budgets in months, but most of that spend goes to bug fixes, rework, and review processes — not shipped products. The question isn’t whether AI works; it’s whether the ROI justifies the bill.

The CEO Complaint

Speaking at a Commonwealth Bank AI event in Australia, Altman acknowledged the growing tension:

“The most common negative feedback I hear from companies centres on rising costs without obvious business outcomes. Spending on AI is going up and up, and they’re asking where the actual revenue and productivity gains are.”

Altman said he understood the criticism. AI is new, and companies need time to restructure workflows around it. But he also admitted he expected a stronger economic effect by now.

His caveat: He’d become “more concerned” if companies are still asking the same questions a year from now. That timeline gives the industry roughly until mid-2027 to prove the economics work.

The 82-Cent Problem

Here’s the uncomfortable math: For every dollar spent on AI, only 18 cents turns into shipped product. The rest — 82 cents — vanishes into:

Bug fixes and rewriting code
Rework and review processes
Tokens burned on dead ends
Infrastructure overhead

That’s according to startup EntelligenceAI, whose numbers align with broader industry reports of companies blowing through annual token budgets in months, not years.

If even directionally accurate, this is a major warning sign. Token waste can be fixed with better monitoring. But unproductive uses of AI? That requires finding productive uses — which is harder than it sounds.

The Data: 90% See No Productivity Gain

Altman’s anecdotal reports match survey data. A National Bureau of Economic Research study of 6,000 executives across the US, UK, Australia and Germany found:

90% saw no meaningful productivity impact from AI tools
Goldman Sachs noted there’s still “no strong economy-wide evidence” linking AI adoption to productivity growth
Uber’s COO said last week it was “tough to justify all the AI spending” without direct productivity gains

These aren’t laggards. These are sophisticated enterprises with dedicated AI teams and seven-figure budgets.

The Cancellations Have Started

When ROI doesn’t materialize, projects get cut. Recent casualties:

Starbucks scrapped its AI inventory tool across North America (May 2026)
Multiple enterprises are “rethinking or cancelling major AI initiatives” according to industry reports
Companies are implementing token rationing and budget caps mid-year

This isn’t a rejection of AI. It’s a rejection of unrestrained spending without accountability.

What Altman Gets Right

Altman made two important points that deserve attention:

1. AI remains very new. The desktop-to-mobile transition took a decade. AI’s infrastructure shift is compressing that timeline, but integration still takes time.

2. Demand for AI intelligence appears “effectively uncapped.” Australia could become a major data center hub, supplying low-cost computing power as global demand grows. The need is real — the economics just haven’t caught up yet.

The Take: This Isn’t a Bust, It’s a Reckoning

The panic over “token shock” is overblown. AI isn’t failing — waste is being exposed. That’s actually healthy.

The companies winning with AI aren’t the ones maxing out credit lines. They’re the ones:

Measuring what ships, not what’s spent
Building workflows around AI, not bolting AI onto broken workflows
Accepting that 82 cents of waste means their process needs fixing, not more tokens

Altman’s timeline — one year to show results — is reasonable. But it puts pressure on the industry to move from “AI is amazing” to “AI pays for itself.”

🔍 THE BOTTOM LINE

Sam Altman acknowledges CEOs are asking “Where is the revenue?” as AI costs soar with limited productivity gains. Data shows 82 cents of every $1 AI spend never ships, and 90% of executives see no meaningful impact.

This isn’t an AI bust — it’s a waste reckoning. Companies that fix their processes will see ROI. Those that just burn tokens won’t. Altman’s one-year timeline gives the industry until mid-2027 to prove the economics work.

❓ FAQ

Q: Is AI actually productive, or is it hype?

A: Both. AI is genuinely capable — but most companies aren’t using it productively yet. The technology works. The integration doesn’t. That’s a process problem, not a tech problem.

Q: What does “82 cents never makes it to production” mean?

A: Of every dollar spent on AI tokens, only 18 cents funds work that actually ships to customers. The rest goes to bug fixes, rewrites, reviews, and dead ends. It’s like paying for 100 hours of developer time but only getting 18 hours of finished code.

Q: Which companies are cutting AI projects?

A: Starbucks cancelled its AI inventory tool in May 2026. Multiple enterprises are “rethinking or cancelling” initiatives according to industry reports. The pattern: projects without clear ROI metrics get cut first.

Q: Should I be worried about AI for my job?

A: If your company is burning tokens without shipping product, AI layoffs might come before AI replacement. The risk isn’t that AI takes your job — it’s that your company spends millions on AI, sees no return, and cuts budgets (and headcount) as a result.

Q: What should companies do differently?

A: Measure what ships, not what’s spent. Build workflows around AI capabilities instead of bolting AI onto existing broken processes. Accept that waste means the process needs fixing, not that you need to buy more tokens.

📰 Sources: Sam Altman at Commonwealth Bank AI event (June 2026), Times of India coverage, Big Technology “The Token Reckoning is Here”, Axios “AI sticker shock hits corporate America”, NBER survey of 6,000 executives, EntelligenceAI token spend analysis.