/AI1h ago

OpenAI Partners Release MRC Protocol For Stargate-Scale AI Training

--0--
#975Lisan al Gaib
Original post
Lisan al Gaib@scaling01#975inAI

I kind of missed this news from May 5th: OpenAI's frontier models are being trained on Stargate Abilene and Microsoft's Fairwater in Atlanta. (https://openai.com/index/mrc-supercomputer-networking/)

and from Nov 12th 2025: Microsoft can do a multi-cluster training with their Fairwater datacenters in Atlanta, Wisconsin and future sites, "which all work together to support OpenAI, the Microsoft AI Superintelligence Team, Copilot capabilities and other leading AI workloads" (https://news.microsoft.com/source/features/ai/from-wisconsin-to-atlanta-microsoft-connects-datacenters-to-build-its-first-ai-superfactory/)

It seems likely that GPT-6 is or was already trained on between ~220k and ~520k GB200s.

That's enough for a 15-30T model.

6:39 AM · Jun 5, 2026 · 357 Views
Sentiment
Sentiment building, check back later.
Cluster Engagement
-
Views
-
Comments
-
Reposts
-
Bookmarks
Expand data
Posts from X
Most Activity
Most Activity
VIEWS7.5KBOOKMARKS19REPLIES6
Lisan al Gaib@scaling01

I kind of missed this news from May 5th: OpenAI's frontier models are being trained on Stargate Abilene and Microsoft's Fairwater sites. (https://openai.com/index/mrc-supercomputer-networking/)

and from Nov 12th 2025: Microsoft can do a multi-cluster training with their Fairwater datacenters in Atlanta, Wisconsin and future sites, "which all work together to support OpenAI, the Microsoft AI Superintelligence Team, Copilot capabilities and other leading AI workloads" (https://news.microsoft.com/source/features/ai/from-wisconsin-to-atlanta-microsoft-connects-datacenters-to-build-its-first-ai-superfactory/)

It seems likely that GPT-6 is or was already trained on between ~220k and ~520k GB200s.

That's enough for a 15-30T model. (larger than my current ~13T estimate for Mythos)

1hViews 7.5KLikes 62Bookmarks 19
LIKES77RETWEETS2
Lisan al Gaib@scaling01

imagine it's december

you are sitting in the cinema watching the epic conclusion of Dune

and in the meantime your GPT-6 or Mythos agent does the work that would take you weeks

you will feel the AGI and the permanent underclass at the same time, as your research agents just cost you 10k in credits

Lisan al Gaib@scaling01

I kind of missed this news from May 5th: OpenAI's frontier models are being trained on Stargate Abilene and Microsoft's Fairwater sites. (https://openai.com/index/mrc-supercomputer-networking/)

and from Nov 12th 2025: Microsoft can do a multi-cluster training with their Fairwater datacenters in Atlanta, Wisconsin and future sites, "which all work together to support OpenAI, the Microsoft AI Superintelligence Team, Copilot capabilities and other leading AI workloads" (https://news.microsoft.com/source/features/ai/from-wisconsin-to-atlanta-microsoft-connects-datacenters-to-build-its-first-ai-superfactory/)

It seems likely that GPT-6 is or was already trained on between ~220k and ~520k GB200s.

That's enough for a 15-30T model. (larger than my current ~13T estimate for Mythos)

1hViews 4.4KLikes 77Bookmarks 4
OpenAI Partners Release MRC Protocol For Stargate-Scale AI Training · Digg