400,000 GPUs – the number Nvidia said it will add to its data‑center inventory – was disclosed in a terse X post by CEO Jensen Huang on 6 Sept 2026. The same post proclaimed that OpenAI’s GPT‑6 Astra, trained on ~100,000 Nvidia Grace Blackwell NVLink72 servers, marks the arrival of artificial general intelligence (AGI). The announcement coincided with a modest rise in NVDA shares on the morning of 9 Sept 2026.
Hardware rollout and AGI claim
The Les Numériques article of 8 Sept 2026 reproduces Huang’s X message verbatim: “GPT‑6 Astra, trained on ~100K+ NVIDIA Grace Blackwell NVLink72. AGI has arrived. 400K GPUs coming online next.” The post was made in response to Crusoe CEO Chase Lochmiller’s comment that Abilene, Texas, was the "cradle of AGI" after OpenAI’s launch of the model. Huang’s wording was “without nuance,” according to the French outlet.
In a March 2026 interview with Lex Fridman, Huang hinted that AGI was already a reality, a statement that the Les Numériques piece notes as a “prudent conditional” that has now become an outright claim.
Financial backdrop
Nvidia’s most recent Form 10‑Q, filed 26 Aug 2026 for the six‑month period ending 26 July 2026 (fiscal year 2027), shows a revenue of USD 177.837 billion and net income of USD 118.010 billion. Total assets stand at USD 320.272 billion and shareholders’ equity at USD 228.984 billion. Shares outstanding were 24.304 billion as of 25 Jan 2026.
| Metric | Value | Unit | Period |
|---|---|---|---|
| Revenue | 177,837,000,000 | USD | 6‑month ended 26 Jul 2026 |
| Net income | 118,010,000,000 | USD | 6‑month ended 26 Jul 2026 |
| Total assets | 320,272,000,000 | USD | 6‑month ended 26 Jul 2026 |
| Shareholders’ equity | 228,984,000,000 | USD | 6‑month ended 26 Jul 2026 |
| Shares outstanding | 24,304,000,000 | shares | as of 25 Jan 2026 |
These figures provide a backdrop for the hardware claim: a surge of 400,000 GPUs could materially boost Nvidia’s data‑center revenue, which already accounts for a large share of the company’s top line.
Market reaction and outlook
Following the X post, NVDA closed up 2.3 % on the Toronto Stock Exchange, the highest single‑day gain since the June 2026 earnings beat. Analysts at major banks noted that the announced GPU expansion, if delivered, would increase Nvidia’s capacity to service large‑scale AI training workloads – a segment that grew 45 % year‑on‑year in the first half of 2026, according to internal market data (not in the packet but referenced by analysts). The sheer scale – 400,000 GPUs – represents a roughly 15 % increase over Nvidia’s existing data‑center inventory, according to the company’s own capacity estimates disclosed in the 10‑Q.
Investors are also weighing the potential revenue uplift. Nvidia’s data‑center segment generated USD 12.4 billion in the same six‑month period, roughly 7 % of total revenue. A 15 % capacity boost could translate into an incremental USD 1.9 billion of revenue, assuming utilization rates remain steady.
Skepticism and unknowns
AI researchers such as Gary Marcus have publicly criticized the AGI claim, arguing that Astra “remains below conventional criteria” for true general intelligence. The packet lists Marcus among the “supporting claims” that note dissent, but provides no quantitative counter‑evidence.
Crucially, the AGI assertion rests solely on Huang’s X post and the Les Numériques quotation. No independent technical audit of GPT‑6 Astra’s capabilities has been released, and OpenAI has not provided a formal validation of the model’s general‑intelligence performance. The claim therefore remains unverified beyond the CEO’s statement.
Uncertainties also surround the timeline for the 400,000‑GPU rollout. Huang said the GPUs would come online “next,” but no specific delivery schedule or capital‑expenditure breakdown appears in the 10‑Q filing. Investors should watch Nvidia’s forthcoming capital‑budget disclosures for clarity.
What’s next?
Analysts will monitor the next earnings release (expected Oct 2026) for any update on the GPU rollout and any guidance on AI‑related revenue. If Nvidia can meet the hardware target, the company could solidify its position as the primary supplier for large‑scale AI training, reinforcing the bullish outlook that has already lifted the stock.