Jensen Huang Declares AGI (AGI) Has Arrived as GPT-6 Astra Trains on 100,000 Blackwell GPUs

100,000 Blackwell GPUs Behind GPT-6 Astra

NVIDIA chief executive Jensen Huang declared on Saturday that artificial general intelligence is no longer a distant milestone, announcing to his followers that “AGI has arrived” in a congratulatory post on X directed at OpenAI. The trigger was the release of GPT-6 Astra, OpenAI’s newest flagship model, which Huang says was trained on roughly 100,000 NVIDIA Grace Blackwell GPUs arranged in the NVLink72 architecture. He framed the pace bluntly: the industry traveled from ChatGPT to the o1 reasoning model to Astra in about four years. There is one oddity in the record that COINOTAG flagged on review — Huang’s first version of the post cited 300,000 GPUs; he deleted it and reposted the smaller figure, and NVIDIA has not explained the correction. The hardware detail matters because NVLink72 wires 72 Blackwell GPUs into a single rack that behaves as one giant accelerator, a design that costs billions to build and only pays off at extreme scale. Huang closed the post with a forward commitment: another 400,000 GPUs coming online next. OpenAI’s own release materials describe Astra as its “smartest, most aligned” model to date, citing scores of 98% on FrontierMath Tier 4, 99.9% on ARC-AGI-3 and 100% on ExploitBench, alongside claimed breakthroughs in computer operation, software engineering, cybersecurity, science and professional work. For readers tracking AI-adjacent digital assets, the declaration lands as the loudest narrative signal yet that frontier labs now treat the general-intelligence threshold as a current event rather than a 2029-plus projection.

congratulatory post on Xhttps://x.com/JensenHuang/status/2096700264569090384

Crusoe CEO Crowns Abilene the Birthplace of AGI

A second primary record reinforces the compute story from the infrastructure side. Crusoe co-founder and chief executive Chase Lochmiller reshared OpenAI’s Astra announcement with an aerial photograph of the company’s massive data center campus in Abilene, Texas, adding that the site has now become “the birthplace of AGI” — his response on X points directly at OpenAI’s Stargate buildout. Official OpenAI figures place Abilene as Stargate’s first United States base, already engaged in training and serving frontier models, with a long-range plan to expand American AI infrastructure to 10 gigawatts by 2029. Crusoe itself is a notable actor: the firm started out mining bitcoin with flared natural gas from oil fields before pivoting into AI data center construction, and it is one of the key builders of the Abilene Stargate facility. Skepticism, however, is warranted. OpenAI has not itself declared that AGI arrived, and its president Greg Brockman pointedly stopped short of that claim. The commercial relationship is also thinner than headlines suggest — the widely discussed $100 billion NVIDIA–OpenAI partnership announced in 2025 was never signed, and mid-year reporting indicated OpenAI had been quietly trimming its NVIDIA chip purchases. NVIDIA, for its part, sold $89 billion of AI computers in a single quarter, more than double the year-earlier figure, with Huang telling investors on the earnings call that “compute is revenue.” Markets read the exchange as confirmation that the AI-driven bull market in compute still has fuel, even as the conflict of interest — the chief AGI claimant also sells the machines — draws criticism. Readers tracking the market in real time can follow live spot and futures prices on Bitget.

response on Xhttps://x.com/ChaseLochmiller/status/2096445087505055891

The 400,000-GPU Signal Crypto Should Watch

COINOTAG’s reading is that the two posts together mark the shift from a model race to a compute-factory race, and the verifiable number going forward is not “AGI has arrived” but the promised 400,000 additional GPUs. The primary records here are the founders’ own X statements, not analyst commentary, and neither contains deployment evidence beyond benchmark claims — so treat the announcement as a sentiment event. Assets tied to the AGI narrative, from the AGI token itself to AI crypto wallet infrastructure, tend to reprice on such headlines before any on-chain or deployment data confirms them. Whether the narrative flows into the token’s tokenomics fundamentals is a separate question; traders watching the fear and greed index should weigh Huang’s GPU roadmap as the measurable checkpoint rather than the AGI label itself.