Jensen Huang just bypassed every earnings call caveat and posted two words on X that the AI industry has been arguing over for years. Whether a 100,000-GPU training run actually proves his point is a much messier question.
NVIDIA (NASDAQ:NVDA | NVDA Price Prediction) CEO Jensen Huang made his most provocative statement of the AI cycle over the weekend, and he made it on X rather than an earnings stage. Responding to news that OpenAI had trained its new GPT-6 Astra model on his company’s flagship silicon, Huang wrote: “GPT-6 Astra, trained on ~100K+ NVIDIA Grace Blackwell NVLink72. From ChatGPT to o1 to Astra in 4 years. AGI has arrived. Congratulations @OpenAI team. 400K GPUs coming online next.”
The post, published September 6, 2026, ties artificial general intelligence directly to NVIDIA hardware. It has since drawn 37,442 likes and 4,196 retweets, and it reframes a debate Huang has been navigating carefully on recent earnings calls.
From Earnings Call Caution to Public Declaration #
On the August 26, 2026 Q2 FY2027 call, Huang was more measured. Asked directly about AGI and recursive self-improvement, he said “for many tasks, we could say that we’ve already achieved AGI” before pivoting to economics: “I think all of those milestones are kind of senseless at this point. The most important thing that matters for the industry is that one, AI is now doing productive and useful work. Two, AI is generating profitable tokens. And three, if we had more compute, we could generate more profitable tokens.”
That framing was captured in his opening remarks: “AI has reached its inflection point. It’s doing useful work. Its tokens are productive and profitable. Now, compute is revenue.”
Compute Is Revenue, and the Numbers Show It #
Q2 FY2027 revenue reached $96.22 billion, up 105.8% year over year and beating consensus of $92.07 billion. Data Center revenue hit $89.02 billion, growing 117%. Non-GAAP gross margin expanded to 75.0% from 72.5%, and net income rose 125.9% to $59.69B.
Guidance calls for $108.0 billion in Q3 revenue, and management said it expects fiscal 2028 revenue to grow approximately 70%. Huang described the outlook as supply-constrained, saying “at this moment, we have supply for 70%. Our demand is much higher than that.”
Inside the 100,000 GPU Cluster That Prompted the Claim #
The Astra cluster follows other six-figure Grace Blackwell deployments NVIDIA has flagged. In Q3 FY26, the company announced the Oracle-built Solstice supercomputer, 100,000 Blackwell GPUs for the U.S. Department of Energy. On the Q2 call, Huang framed the broader buildout in unit economics: Grace Blackwell generates $25 billion per gigawatt in revenue opportunity, and successor Vera Rubin lifts that to $40 billion per gigawatt. Top five hyperscaler capex is expected to reach nearly $800 billion in 2026 and $1.3 trillion in 2027 (all of that spend has to be powered, cooled, and networked by somebody, and we pulled together seven suppliers doing exactly that in a free report here).
Market Response #
NVDA closed at $230.36 on September 4, 2026, up 5.89% over the past week, 23.67% year to date, and 34.37% over the past year. Market cap sits near $5.56 trillion. Watch whether Huang’s AGI language shows up in Q3 FY27 prepared remarks, and whether the 400,000 GPUs he says are “coming online next” get quantified in future guidance.
Contact [email protected] for any questions or corrections.