The Setup
Nvidia CEO Jensen Huang declared AGI achieved over the Labor Day weekend, and the stock dropped 2%. That's the third time he's made the call, and the third time markets basically shrugged. CoreWeave jumped 15%, SoftBank added 2%, but Nvidia—the company that just posted the most profitable quarter in capitalism's history—went down.
That tells you something about how markets actually process major tech announcements versus how we think they should. OpenAI's new Astra model trained on 100,000 Nvidia chips (Huang first said 300,000 before deleting that post, which is its own weird signal). The model's supposedly hit artificial general intelligence, meaning it can do cognitive work at the level of a well-educated adult. But if you check where the money actually moved, you'll see the market response was specific and selective, not some broad "AGI is here" rally.
Where the Money Actually Went
The stocks that moved were the ones most leveraged to OpenAI specifically. CoreWeave, which has been getting hammered over debt concerns for months, exploded up 15%. SoftBank, which is deeply tied to OpenAI's funding, added 2%. Oracle, same thing. These are the companies whose business models hinge on OpenAI staying ahead in the model race, and Astra puts them back at the front after about a year of falling behind.
Nvidia went the other direction. Down 2% on the day, which matches the pattern from March when Huang made his last AGI declaration on Lex Fridman's podcast. The stock fell 0.3% that day too. Gil Luria at D.A. Davidson told Fortune that Nvidia's just too big to grow at this point—they're already so dominant that nobody believes the trajectory can continue. Two weeks before Huang's AGI post, Nvidia reported $96 billion in quarterly revenue, up 106% year-over-year, and guided to $108 billion for the next quarter. When you're already that massive, individual AI announcements don't move your stock the way they move smaller, more concentrated plays.
Luria's take is that traders use Nvidia as a source of funds now. You sell Nvidia to buy CoreWeave. That's also why nobody cared when board member Mark Stevens filed to sell up to $1 billion of his shares last week. "Good for him," Luria said.
What AGI Actually Means (And Why Nobody Agrees)
Huang defined AGI to Lex Fridman in March as the ability to create a $1 billion company, but he didn't mention that benchmark in Sunday's post. Can Astra actually build a billion-dollar business on its own? Theoretically, Luria said, but until someone shows a real example of an AI or an agent building that kind of enterprise alone, it's hypothetical.
AI researchers use a stricter definition. The widely cited standard describes AGI as an AI that matches or surpasses the cognitive versatility of a well-educated adult. Gary Marcus and Miles Brundage (who quit OpenAI in 2024 over safety concerns) created a 10-point test list: write an Oscar-caliber screenplay, laugh at the right moments in a movie, master a new video game within hours. By that standard, we're not there yet, even though the models are astounding.
Basil Halperin, an economist at the University of Virginia who co-wrote a widely circulated 2023 analysis on how financial markets price transformative AI, ran his own test. He asked GPT-5.6 to move his old iTunes playlists to Spotify. "It got the job done. But I had to sit there babysitting it for three hours." That's not AGI. That's a powerful tool that still needs human supervision.
Why Stocks Are the Wrong Signal
If you're trying to figure out whether AGI is actually here, watching Nvidia's stock price is probably the wrong approach. Halperin argues that stocks "are a giant pain in the butt because they reflect many different things." The two leading AI labs are private. Existential risk or cybersecurity risk from AGI could push equities down, not up, as investors flee to Treasuries and gold. And competition between OpenAI and Anthropic could leave the industry looking like Uber and Lyft—they make money, but they're not as profitable as Google or Apple.
The cleaner signal is real interest rates. Real rates are the yield on inflation-protected Treasuries with inflation stripped out (TIPS). If AI raises growth, it raises interest rates, since rates represent the expected discounted value of future earnings. Halperin's rule of thumb is that every percentage point of extra growth adds roughly a percentage point to real rates. So when Sam Altman and Dario Amodei talk about growth of 5% or 10% or more from AGI, they're describing a world where real rates sit near 10%. Those kinds of rates would swamp everything else in terms of market impact.
But real rates haven't moved like that. They've climbed three to four percentage points since 2021, and Halperin attributes some of that to hyperscalers' capex, which is now running around 2% of GDP per year. That's massive, but it's nowhere near a singularity. In fact, a paper by MIT economists Isaiah Andrews and Maryam Farboodi found that long-term Treasury yields have on average fallen by more than a tenth of a percentage point around major model releases, and stayed down for weeks. That pattern suggests the market on average has been disappointed by AI announcements, not energized by them.
What the Aggregate Market View Actually Is
The efficient market is aggregating the views of the general public, not just the people on X who are either cheering or dunking on Huang's post. A survey Halperin helped run tested expectations across groups from normal people to AI researchers and found that most expect AI to add about half a percentage point to GDP growth over the next few years.
On one hand, that's enormous. Half a point of extra GDP growth sustained over time is a big deal. On the other hand, it's nothing close to a singularity or the kind of transformative event that would justify the hype around AGI declarations. Halperin's own forecast for the next five years is "like the dot-com boom, but twice as fast and twice as hard. Not the singularity. Not yet."
That seems to be what the markets are saying too. The companies most directly tied to OpenAI's success moved. The broader market didn't. Real rates didn't spike. Long-term Treasury yields didn't react in a way that would signal a fundamental shift in growth expectations.
Where This Leaves Traders
If you're trading around AI announcements, the pattern from this release is pretty clear. Specific exposure to the company making the breakthrough matters. Broad exposure to the sector or to Nvidia as a proxy for AI doesn't react the same way. Nvidia's too big and too established to move on individual model releases anymore. The stocks that jumped were the ones with concentrated, direct leverage to OpenAI's competitive position.
The other thing worth watching is what behavioral patterns traders fall into around major tech news. There's a gap between how we think markets should respond to something like an AGI declaration and how they actually respond. Markets price in expectations, not announcements. If everyone already believed AGI was close, or if they're skeptical of the definition being used, the stock reaction won't match the headline. That's what happened here.
Real interest rates and Treasury yields are the better long-term signals for whether markets believe transformative AI is coming. Until those move in a sustained way, the rest is noise. Huang can declare AGI as many times as he wants. The market will respond to the mechanics, not the declaration.
