Meta just stopped renting nvidia for training. its third-gen mtia accelerator, codenamed iris and…
meta just stopped renting nvidia for training. its third-gen mtia accelerator, codenamed iris and now in production this september, is a broadcom-designed 3nm chip aimed at doubling meta's compute footprint to 14 gigawatts.
the nvidia monopoly inside meta is over.
Context
Reuters, republished by CNBC on 9 July 2026 from an internal memo it reviewed, reports that Meta planned to start manufacturing its Iris AI chip from September as part of a plan to raise computing power to 14 gigawatts next year, with Broadcom on design and TSMC on manufacturing. The chip is described as aimed at augmenting the GPUs Meta buys from Nvidia and AMD. Meta's own post of 11 March 2026 lists four MTIA generations in two years, with MTIA 300 on ranking and recommendation training and already in production, and MTIA 400, 450 and 500 mainly for generative AI inference. Meta Engineering's post of 24 August 2026 describes MTIA 300 as the first of a family of training accelerators for ranking and recommendation models.
Production in September is a plan reported from a memo as of 9 July, and no first-party or later confirmation was found. No source read maps Iris to a third generation, since Meta's pages use MTIA 300, 400, 450 and 500, and 3nm was not in the text read. Stopped renting Nvidia for training does not match the sources, which describe the chip as adding to purchased Nvidia and AMD GPUs and put MTIA 300 on ranking and recommendation training. The 14 gigawatts is the memo's company-wide plan (7 in 2026, 14 the next year), not an effect of one chip. The monopoly inside Meta is over is the author's take.
Watch next
- A Meta or Broadcom confirmation of the production start and how Iris maps to the MTIA numbering.
Sources
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 21 September 2026 at 13:04 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →