Losing $12.3 billion in one quarter, OpenAI is setting investors' money off like fireworks.

Q2 revenue was $6.7 billion, up 18% quarter-on-quarter, which sounds good. But its operating loss was $12.3 billion, $3 billion more than the $9.3 billion loss in Q1 last year. Revenue rose 18%, while losses grew 32%, making the ROI (return on investment) negative—and increasingly so. In any traditional industry, a company like this would have been kicked out by investors long ago. But OpenAI is different: it's AI, it's OpenAI, it's Sam Altman—and so it can keep raising money, keep burning cash, and keep drawing up grand visions.

But the bigger the vision gets, the thinner the substance becomes. Anthropic's Q2 revenue was $11.5 billion, doubling year-on-year, and it was profitable. Palantir, CoreWeave, and Micron all have growth rates higher than OpenAI's. ChatGPT's user growth is slowing, while Claude Code is dominating among programmers. OpenAI has gone from the one "defining the sector" to the one being "defined" by it.

The executive departures say even more. Denise Dresser, the Chief Revenue Officer, left after less than a year in the role. Brad Lightcap and Fidji Simo have also left. If a company's business were booming, would executives be lining up to resign? No. A wave of departures suggests internal disagreements over the direction of the future, or a lack of confidence in the IPO prospects. After all, OpenAI secretly filed in June, planning to go public in 2027—but with these results, how would it pitch itself on the roadshow? "We lost $12.3 billion this quarter, but please believe that we'll make money in the future"?

Even more surreal is Nvidia's $600 billion. OpenAI has pledged to deploy 12 gigawatts of Nvidia AI infrastructure by 2030, creating a $600 billion commercial opportunity in computing power. The figure is absurdly large, but the premise is that OpenAI can survive until 2030 and deliver on its growth promises. If it continues burning money at this rate, investors will eventually ask: What moat has this $12.3 billion loss actually built? GPT-5.6? But Anthropic's Claude is also iterating, and it isn't losing nearly as much money.

Put plainly, OpenAI's current situation looks a lot like Uber's back then—rapid growth, massive losses, executive turmoil, and competitors hot on its heels. Uber eventually made it through, but it relied on global dominance and economies of scale. Does OpenAI have those? The foundation-model sector has no network effects: users can use ChatGPT today and switch to Claude tomorrow, with switching costs close to zero.

Sam Altman says the growth rate will rebound in Q3. Hopefully. But in the capital markets, "hopefully" is the least valuable thing. When Anthropic could go public as early as October this year, while OpenAI will have to wait until 2027, those two years in between will be enough to change a lot. #Gate股票观点挑战 $NVDA