Altman Admits OpenAI Underperformed Over the Past Year, Takes Primary Responsibility
Taylor Wilson
OpenAI CEO Sam Altman publicly acknowledged the company underperformed over the past year and took personal responsibility; meanwhile, skepticism about OpenAI's business model is spreading from one company to the entire AI infrastructure chain.
What exactly did Altman say?
On July 17, Altman posted on X: "The past 12 months were not our best year ever, and that's mostly on me."
He disclosed no specifics, but the wording was unusually direct — a tech CEO publicly shouldering blame is rare.
He also signaled a turnaround: AI's mission is to give people freedom and wealth, and new products are coming soon.
Are users and the market convinced?
Social-media reaction split: some praised the candor, others said words mean nothing without delivery.
User Jeff (@AllenTheDetails) put it sharply: "If it's just talk, users won't feel it. If it's better workflows and lower costs, they will."
This means → the bar for judging Altman has shifted from "what he says" to "what he ships."
"The AI bubble is really an OpenAI bubble" — where does that claim come from?
Long-time AI bear Ed Zitron recently made his most aggressive argument yet: OpenAI is the "credit anchor" of the entire AI valuation stack.
In plain terms = investors believe GPU demand will keep growing and data centers are worth the spend because they assume OpenAI will keep scaling fast; if that assumption cracks, the valuation logic for the whole chain needs recalculating.
Zitron's case rests on three points: inference costs remain too high, capex is outpacing cash-flow improvement, and the company will need external funding for years.
⚠ These are Zitron's personal views; OpenAI has not endorsed them.
Which companies face contagion risk?
Zitron argues that if demand from core clients like OpenAI falls short, Oracle and CoreWeave — infrastructure plays — would be hit first.
This reflects a deeper concern: those companies' rich valuations are built largely on the assumption that AI demand keeps exploding.
Anthropic and SoftBank are also in the frame — Anthropic burns cash and relies on big-tech compute support, while SoftBank's massive AI bets mean its risk exposure is the largest if the sector enters a valuation reset.
Where does Wall Street stand on this debate?
Whether AI has entered bubble territory remains unresolved.
Oaktree Capital's Howard Marks recently shifted his stance: from suspecting AI might be a bubble to acknowledging its long-term value, framing AI as a general-purpose platform akin to the internet.
Some academic research offers a more neutral read: the current AI market combines real technological progress with pockets of overvaluation — closer to "tech revolution plus localized bubble" than pure speculation.
What metrics should investors watch now?
The market's focus is shifting: from "how much money was spent" to "how much money is being made."
The numbers that matter most: enterprise AI revenue growth, paid-conversion rates, inference-cost declines, data-center utilization, and AI investment payback periods.
This means → Altman's promises will ultimately be tested against these metrics — they are the key benchmark for the market to reassess AI valuation logic.
Content is for reference only, not financial advice.