Nvidia CEO Jensen Huang sent shockwaves through X on Sunday by declaring that artificial general intelligence had finally arrived, four years after ChatGPT. The post praised OpenAI for GPT-6 Astra, trained on 100,000 Grace Blackwell NVLink72 chips. Yet behind the enthusiasm, Wall Street answered with caution: Nvidia shares slipped 2.01%.
🔑 Key takeaways
- Jensen Huang declares AGI « has arrived » after OpenAI unveils GPT-6 Astra
- Nvidia drops 2.01% to $225.73; CoreWeave jumps 15%
- OpenAI hits $40B ARR but posts a $12.3B loss in Q2 2026
- Gary Marcus demands a clear definition, publishes 10 benchmarks for AGI
- Economist Halperin says real interest rates are the true macro signal of AGI
The post that shook X
Huang published his message on Sunday, claiming AGI « is here » after OpenAI launched GPT-6 Astra the previous Thursday. The model is said to have been trained on roughly 100,000 Nvidia Grace Blackwell NVLink72 chips, with 400,000 GPUs set to be « operational next. » The original post cited 300,000 chips, which has since been deleted and corrected.
Reactions were sharply divided. Jim Cramer argued Nvidia was the winner. Hunter Horsley, a crypto CEO, called Huang « the most credible referee. » Gary Marcus, a critic of frontier AI, fired back: « Huang offered no evidence and no definition — this looks like an attempt to seize a scientific question by corporate decree. » He published ten benchmarks co-authored with Miles Brundage, formerly of OpenAI: an educated adult can write an Oscar-worthy screenplay, laugh at the right moment in a film, master a new video game in hours.
Markets do not follow the boss
Wall Street had its first window to react on Tuesday, after the long Labor Day weekend. CoreWeave exploded by 15%; SoftBank added another 2%. But Nvidia, the sector’s colossus, dropped 2.01% to $225.73, down $4.63. Back in March, on the Lex Fridman podcast, Huang had also declared AGI: the stock lost 0.3% that day.
Gil Luria, head of technology research at D.A. Davidson, decoded the move: « If you look slightly elsewhere, you’ll see the market is responding. » In his view, SoftBank, Oracle and CoreWeave — the names most exposed to OpenAI — are climbing, while Nvidia, dominant since 2024, is « too big to keep growing. »
| Stock | Move | Price / exposure |
|---|---|---|
| CoreWeave | +15% | OpenAI exposed |
| SoftBank | +2% | OpenAI exposed |
| Oracle | — | OpenAI exposed |
| Nvidia | -2.01% | $225.73 |
Two weeks ago, Nvidia reported the most profitable quarter in capitalism’s history: $96 billion in revenue, +106% year-over-year, with guidance of $108 billion for the current quarter. The data-center unit, which includes AI chips, generated $89 billion. Luria sums it up: « They’re so good that nobody believes it can continue. » For traders, the stock is now more a liquidity source to fund CoreWeave than a buy/sell call on the AI race.
OpenAI: soaring revenue, abysmal losses
Threads on Hacker News cast a harsh light on OpenAI’s finances. Some users cite a 2026 turnaround: exploding revenue and ARR climbing to $40 billion annually. Others point to a $12.3 billion loss in Q2 2026, a postponed IPO, and the recent departure of two senior executives.
Sam Altman, on the Sources podcast, called AGI « a very poorly defined marketing term. » Greg Brockman, OpenAI’s president, had nonetheless declared: « Welcome to the era of AGI. » OpenAI touts Astra’s ability to fill in online forms, update customer records, organize calendars, run research, and create documents and spreadsheets, calling it « the best model for software engineering to date. »
« Huang offered no evidence and no definition — this looks like an attempt to seize a scientific question by corporate decree. »
Gary Marcus, AI researcher
A moving target
Huang’s own definition of AGI has shifted. In March, on the Lex Fridman podcast, he defined it as the ability to build a billion-dollar company. Sunday’s post on X dropped that benchmark entirely. Asked whether Astra had met it, Luria replied that, absent a billion-dollar firm built by one person using an AI model, « it’s hypothetical. »
Basil Halperin, an economist at the University of Virginia and co-author of a widely circulated analysis arguing that markets do not price in transformative AI for decades, ran his own test last weekend. He asked GPT-5.6 to migrate his old iTunes playlists to Spotify. « The task was done, but I had to babysit it for three hours. » By a strict definition — « the cognitive versatility and competence of a well-educated adult » — « we have absolutely not reached AGI, even if the models are stunning. »
The real signal sits in rates
Halperin argues the true thermometer of AGI is not Nvidia’s share price but the real interest rate. If AI boosts growth, it also pushes rates higher, since they reflect the discounted value of future earnings. Higher rates, in turn, shrink the present value of those earnings — mechanically dragging Nvidia’s stock down.
His empirical rule: each extra point of growth should add roughly one point to real rates. When Sam Altman or Dario Amodei talk about 5–10% growth, they describe a world where real rates hover near 10%. « That would annihilate everything else, if it actually happens. »
Yet real rates have only risen three to four points since 2021, and Halperin attributes part of that move to hyperscaler capex (≈2% of GDP per year). A study by MIT economists Isaiah Andrews and Maryam Farboodi found that long-dated Treasury yields drop on average by more than a tenth of a point on the day of major model launches — « which would suggest the market has been broadly disappointed. »
For the next five years, Halperin forecasts « the dot-com boom, but twice as fast and twice as strong. Not the singularity. Not yet. »
Conclusion
Between Huang’s enthusiasm for GPT-6 Astra and skeptics like Halperin, Wall Street is charting its own course: betting on the names exposed to OpenAI rather than the incumbent champion. As long as real rates do not rip higher, the market will likely treat Nvidia as a source of liquidity rather than as the ultimate AI winner.
But if a true cognitive breakthrough ever materializes, it will be the macroeconomy — not the micro — that rings the alarm. The next few quarters will tell whether Huang’s declaration was a scientific turning point or a corporate PR coup.
Sources
- Fortune — Markets react to AGI declaration
- Yahoo Finance — AGI: Nvidia CEO Jensen Huang
- Hacker News — Discussion on OpenAI financials
- Fox Business — Huang declares AGI
- Business Insider — Nvidia, Jensen Huang, AGI, OpenAI
- Mashable — Huang on the Lex Fridman podcast
This article is published for informational and educational purposes only. It does not constitute investment advice. Do your own research (DYOR) before making any decisions.
