Receipts compare Google TPU, Gemini and xAI’s 10T training claims
imjustnewatai · x · 2026-07-26
The post gathers receipts for claims about frontier-model training scale and infrastructure, citing Google’s Gemini 4 announcement, xAI’s 10T training claim, Google’s TPU 8T infrastructure, and Google’s 1.2T-parameter GLaM.
It is framed as a response to Elon Musk’s “SpaceXAI Colossus 2” post listing seven models in training, including two 1T variants, two 1.5T variants, a 6T model, and a 10T model.
More from Infra
- China’s AI hardware players face mixed demand as domestic capex rises — teortaxesTex · 2026-07-26
- Paper reports a 460 Gbit/s suspended lithium tantalate Mach–Zehnder modulator — jwt0625 · 2026-07-26
- Fine-tuning framework comparison hides key assumptions about EP, SP and 2M context — StasBekman · 2026-07-26
- BYOK model services are turning into a price war, with middlemen stuck as support — yangyi · 2026-07-26
- Reddit asks whether RTX PRO 4500 makes sense beyond 24/7 low-power use — meikawaii · 2026-07-26
- Enterprises are shifting from token maxxing to budget maxxing — 0xsachi · 2026-07-26