Archived story — belongs to the Wednesday, September 30, 2026 sweep.
P1 · Tectonic · Tech Industry · sweep Wednesday, September 30, 2026
Compute Escalation in OpenAI Frontier Model Training
OpenAI scales its largest training run to more than 100,000 GPUs at the Stargate site in Texas for the Astra model, while claiming to surpass Anthropic and approaching the AGI threshold. The underlying dynamic is the conversion of capital into concentrated physical compute: competitive advantage in frontier AI now depends on access to energy, chips, and land, not only on algorithmic talent.
Title, thesis and gravity assessment written by an AI model, without prior human review.
Gravity84/100 · Tectonic
Why this gravity
OpenAI releases GPT-6 Astra, trained with more than 100,000 GPUs at its Stargate site in Texas, and claims to have surpassed Anthropic and that the model could be considered AGI. It is a frontier model launch by a Tier 1 lab, but without independent benchmarks or external AGI validation, so it does not reach the P0 threshold.
- Independent sources
- 2
- Cross verticals
- Frontier AI, Tech Industry, Investing, Energy & Materials
- Tier 1 actors
- OpenAI, Anthropic
- Gravity vs. the previous edition
- +4 (80 → 84)
Lead source
OpenAI says Astra was built on its largest-ever training run, using more than 100,000 GPUs at its Stargate site in Texas (Ina Fried/Axios) · TechmemeMore: Financial Times
Timeline · Madrid time
- 20:00OpenAI says it has overtaken Anthropic with its latest AI model · Financial Times
- 20:25
ShareWhatsApp