PodcastThe Monetary Matters NetworkEd Zitron

Ed Zitron: The AI Bubble is Bleeding Cash, Here Are The Receipts

The hosts argue AI at scale is burning vast sums — citing OpenAI's roughly $34 billion spend to earn $13.07 billion and a near‑$21 billion loss — and contend unprofitability runs through the value chain. They highlight unusually concentrated demand and capital: Anthropic and OpenAI drive roughly 89% of top‑tier demand, and hypothetical compressed raises (e.g., OpenAI $122 billion, Anthropic $95 billion) would dwarf historical cloud capex. That concentration, compressed timelines, and unclear commercialization create overbuilding and systemic‑risk concerns, with some executives reportedly prioritizing AI even at the expense of short‑term solvency.

Artificial IntelligenceCorporate EarningsInvestment StrategyMarket BreadthRisk ManagementAI Infrastructure

Preview · 16 more points in the app

Read the Full Summary

Key Takeaways

The hosts report OpenAI spent roughly $34 billion to generate about $13.07 billion of revenue, producing a net loss just under $21 billion, and they publicly dispute the Financial Times’ lower $8 billion loss figure as inaccurate.

The episode highlights unprecedented, highly concentrated capital flows — hypothetical six‑month raises of roughly $122 billion for OpenAI and $95 billion for Anthropic — that would dwarf historical cloud capex (AWS ~$53–$55 billion from 2003–2017) and even exceed some Amazon retail+AWS spending metrics.

Compute demand is concentrated in a tiny handful of buyers (the hosts single out OpenAI and Anthropic as capturing roughly 89% of top‑tier demand), meaning large data‑center buildouts are fragile unless 'two or three more OpenAI‑ or Anthropic‑sized customers' appear.

Profitability problems run through the AI value chain: the hosts assert 'almost every AI company is unprofitable,' most AI firms make barely more than $100 million a year, and the largest compute customers and their downstream customers are also loss‑making, creating a structural profitability issue.

TECHNOLOGY

OpenAI loss — Hosts say OpenAI spent about $34 billion to generate roughly $13.07 billion in revenue, producing a net loss of just under $21 billion. They use the example to illustrate extreme cash burn at scale.

“OpenAI spent $34 billion to make $13.07 billion. And, yeah, their net loss is just a little under $21 billion.”—

Loss discrepancy — Hosts note a discrepancy between their ~$21 billion loss figure and the Financial Times' $8 billion claim. They call the FT's framing 'laughable.'

“They are, of course, with the FT, given comments that suggest that they only lost $8 billion. I think that's laughable.”—

Data‑center risk — Hosts argue revenue is insufficient to justify the scale of data‑center construction. 'We need two or three more OpenAI‑ or Anthropic‑sized customers that do not exist,' they say, creating a risk of overbuilding.

“The amount of revenue to substantiate and justify the data centers that are allegedly under construction means that we need two or three more OpenAI and Anthropik -sized customers that…”—

Intro

OpenAI's $21 Billion Loss In 2025 - Huge

Advance In Model Capability From December 2025

Anthropic's Significant Increase In Revenue, Ed Reacts

Agentic AI: "It's Wrong"

Meta's Confusing AI Strategy

Anthropic, Fable/Mythos, and the U.S. Government

AI Bubble Will Pop In 2027? "A Safe Bet"

Related Episodes

Important Disclaimers

Not Investment Advice: For education only — not financial, tax or legal advice.

AI-Generated: Made by AI from source transcripts and may contain errors.

No Guarantees: Past performance does not predict future results; investing carries risk.

Our Commentary · Request Removal