OpenAI's CEO Sam Altman has publicly acknowledged that the company's performance over the past year fell short of expectations, attributing the shortcomings primarily to himself, while also stating that the coming 12 months will be the company's best year yet. This rare admission comes as external scrutiny of OpenAI's business model and the return on investment in AI continues to intensify.

On July 17th, Altman posted on the social media platform X: "The last 12 months were not our best year, and that is largely my fault, but we are about to have our best year yet." He concurrently emphasized that the mission of AI is to grant more people freedom, autonomy, and wealth, rather than relying on fear to drive user choices.

Simultaneously, debates surrounding the AI business model continue to escalate. According to prior reports, long-time AI skeptic Ed Zitron recently labeled OpenAI as the "systemically important institution" of the current AI investment cycle, suggesting that if its business model encounters problems, the shockwaves could spread to data centers, AI infrastructure, and even global tech stocks. Altman's public statement has brought this discussion back into the market spotlight.

Altman's Introspective Apology and User Expectations for Tangible Results

Altman did not disclose specific details of the missteps, but his wording was notably direct, attributing the past year's underperformance to responsibility at his leadership level. He stated that the team is working on "amazing work" and hinted at upcoming new products that will satisfy users.

This statement elicited mixed reactions on social media. Some users appreciated his candor, while others argued that more crucial than public statements is whether the company can deliver on promises regarding product quality, system stability, and commercialization efficiency over the next 12 months.

User Jeff (@AllenTheDetails) commented: "If it's just talk, users won't feel it; if it's better workflows and lower service costs, they will." Another user, @onecloudtech, stated: "Users still have confidence in the product, but trust needs to be built with each product release."

Is the AI Bubble an "OpenAI Bubble"? Zitron's Long-Form Critique Raises Radical Doubts

The backdrop to Altman's post is increasing external skepticism about OpenAI's business model forming a new wave of public pressure. Long-time AI critic Ed Zitron recently published a lengthy article, presenting his most radical assessment to date: the real AI bubble is essentially an "OpenAI bubble."

He argues that since ChatGPT's launch in late 2022, OpenAI has effectively become the "credit anchor" for the entire generative AI era—investor confidence in the long-term growth of hyperscale data centers, GPU demand, and the eventual profitability of large model companies is predicated on OpenAI's continued rapid growth.

Zitron's doubts focus on three key points: first, inference costs remain prohibitively high, and user growth may lead to proportionally increasing losses; second, the pace of capital expenditure expansion far outpaces cash flow improvement, with many data center projects taking years to recoup costs; third, OpenAI will continue to rely heavily on external financing for years to come, and if financing conditions tighten, its business model will face greater pressure.

It is worth noting that these views represent Zitron's personal opinion and have not been endorsed by OpenAI. However, these critiques do reflect genuine recent market debates surrounding AI return on investment (ROI).

Hidden Risks in the AI Infrastructure Frenzy: If Demand Expectations Falter, the Supply Chain Faces Valuation Adjustments

Zitron's concerns extend beyond OpenAI itself to the entire AI infrastructure ecosystem.

Over the past two years, the US tech industry has witnessed an unprecedented wave of data center construction. Hyperscale cloud providers like Microsoft, Google, Meta, and Amazon have consistently increased capital expenditures, while companies like Oracle and CoreWeave have taken on an increasing share of AI compute construction, with many projects heavily reliant on long-term leases, project financing, private credit, and corporate debt.

Zitron believes that if demand from core clients like OpenAI falls short of expectations, or if capital markets reassess AI ROI, companies like Oracle and CoreWeave, whose valuations are heavily dependent on anticipated explosive AI demand growth, could be the first to feel the impact.

Furthermore, Anthropic and SoftBank were also included in the discussion. Zitron pointed out that while Anthropic's approach differs from OpenAI's, it similarly requires continuous massive funding and relies on major tech firms for compute support. SoftBank, due to its large-scale bets on AI infrastructure, chips, and model companies, would see its extensive AI portfolio face market scrutiny if the industry enters a valuation correction cycle.

Bubble or Not? Market Focus Shifts from "Money Spent" to "Money Earned"

The debate on Wall Street about whether AI has entered a bubble phase has persisted for some time, with no definitive conclusion yet.

Oaktree Capital co-founder Howard Marks recently stated that he has shifted from initial skepticism that AI might be just a bubble to a greater recognition of its long-term value. He believes modern AI's demonstrated reasoning, contextual understanding, and interactive capabilities possess unprecedented characteristics, making it unsuitable for simple comparisons to historical speculative bubbles. He positions AI as a general-purpose technology platform akin to the internet or electrification.

Some academic research offers a more neutral conclusion: the current AI market exhibits both genuine technological advancement and localized issues of overheated valuations and premature capital expenditure, resembling a "technological revolution superimposed with a local bubble" rather than pure speculative mania.

For investors, the metrics truly worth tracking are shifting from the scale of capital expenditure to another set of data: corporate AI revenue growth, AI product paid conversion rates, the pace of inference cost reduction, data center utilization, and AI investment payback periods. Whether Altman's promises are validated across these dimensions may become a key reference point for the market's re-evaluation of AI transaction valuation logic.