On September 3, OpenAI unveiled its GPT-6 Astra model, with President Greg Brockman proclaiming “welcome to the AGI era” and describing the release as a “generational leap.” Major U.S. stock indexes hit record highs, with AI-related shares driving most of the gains. Sentiment is exuberant, yet a structural risk is being overlooked: AI industry revenue remains heavily concentrated in a small group of companies, and OpenAI’s own messaging on AGI is strikingly inconsistent. Investors would be wise to scrutinize the valuation narrative ahead of a likely IPO.
At the launch event, Brockman suggested that if one fast-forwards a few years and asks when AGI was truly created, the answer may well be this moment and this model. He argued that the new system can solve century-old math problems while accelerating economic growth, and said it is “not unreasonable” to believe we are now in the AGI era.
However, less than a month earlier, on August 11, OpenAI CEO Sam Altman publicly described AGI as “not a super useful term” and largely an “irrelevant marketing term.” The president and the CEO of the same company are offering fundamentally different definitions of the same concept. That contradiction alone deserves attention.
More important, OpenAI is not claiming to have fully solved AGI, nor is it declaring the race over. The company is merely positioning its latest model as placing the industry on a path toward AGI, with experts possibly pointing years later to GPT-6 Astra as the starting point of the era. That carefully hedged language, combined with the timing ahead of an anticipated IPO, makes it difficult to dismiss concerns about valuation management.
The real risk extends beyond messaging. The AI boom is not without fundamental support. Google Cloud posted second-quarter revenue of $24.8 billion, up 82% year over year, with AI services contributing $8.8 billion in operating profit. Microsoft Azure and Amazon AWS are also reporting rapid revenue growth. Hyperscalers are projected to spend more than $625 billion on AI infrastructure in 2026 alone, and cloud revenue is beginning to form a self-funding cycle: infrastructure spending generates cloud revenue, which in turn finances the next round of construction.
But the problem is that the overwhelming majority of that revenue flows to just five companies — Alphabet, Microsoft, Amazon, Meta and Nvidia. Thousands of startups and smaller software firms are building AI features, yet very few are generating meaningful revenue from them. When the S&P 500 hits record highs on AI enthusiasm, the gains concentrate in companies with actual AI earnings, while many stocks riding the sentiment wave lack profit support.
The RAND Corporation has warned that AGI could allow a leader to permanently entrench its position, converting an early advantage into a lasting constraint on competitors and a decisive economic edge. If revenue concentration continues, a winner-take-all market structure will emerge, which is not healthy for the broader innovation ecosystem. Should enterprise AI adoption slow or a major project fail publicly, the correction would not hit only the direct culprits. It would ripple through every stock that traded up on AI sentiment, including many that never produced AI revenue at all.
For investors, the critical question is no longer whether this is a bubble, but who captures the value. In past technology revolutions — the internet, mobile and cloud computing — the biggest winners often emerged at the application layer rather than in infrastructure. The AI application layer has yet to produce a clear leader. That uncertainty is both a risk and an opportunity. But with OpenAI loudly proclaiming the AGI era, the gap between revenue concentration and marketing rhetoric is a warning signal that cannot be ignored.
