Amazon ends data-center NDAs, pledges $1B to host towns

Share
Amazon ends data-center NDAs, pledges $1B to host towns

The political bill for the AI buildout is coming due, and its biggest builders are starting to pay. Three stories below: Amazon's concession, China's lithography stockpile, and Cloudflare's play for the agent-era repo.

Amazon has stopped signing non-disclosure agreements with local officials over its data centers and is pledging more than $1 billion to host communities over the next five years. AWS CEO Matt Garman announced both moves in a blog post on Friday: the company "no longer use[s] nondisclosure agreements with the government agencies we work with on our projects," and the new Built Together program funds education, job training and energy upgrades — free degree access for 300,000 students, up to 100,000 workers trained a year by the end of 2028, and efficiency retrofits for 30,000-plus homes. The concession answers a revolt Amazon can no longer wave off: Garman counted more than 100 data-center moratoriums under consideration across the US and warned that enacting them would mean "the consequences would last generations." But the counter-framing writes itself — $1 billion over five years is roughly 0.1 percent of Amazon's 2026 AI infrastructure spend, and WIRED notes the company would not say whether the NDA pledge extends to its contractors. A promise that small against capex reads less like contrition than a down payment on keeping permits moving.


A new report puts China's advanced lithography stockpile at 343 machines — and its authors want the pipeline shut off entirely. The Center for Technology & Statecraft estimates Chinese-owned fabs had acquired roughly 343 immersion DUV systems by early 2026, about 270 of them ASML's NXT:1980i, with more than $13 billion spent on that single model across 2024 and 2025 — and while DUV immersion falls short of EUV, the tools can be pushed to 7nm logic and advanced HBM through multi-patterning. The report's recommendation is a complete ban on immersion DUV exports to China plus restrictions on replacement parts and servicing, pressure that lands on a live legislative vehicle: the MATCH Act's countrywide DUV restriction survived April's scaled-back rewrite but is still a proposal, not law. The stockpile is what export controls are racing against — China's share of ASML's DUV immersion shipments climbed from 26 percent in 2022 to roughly 60–70 percent over the past two years.


Cloudflare wants the next Git to be built for agents — and is offering $25,000 to whoever shows up first. Now that Artifacts, its Git-speaking versioned filesystem, is in open beta, Cloudflare is running a competition to build the coordination layer above it: how hundreds of concurrent agents branch, review and merge on the same codebase, with submissions due October 14 and the top three teams flown to Cloudflare Connect. The pitch is that GitHub's branch-and-pull-request model assumes human pacing, while agent workloads need per-task repositories, event subscriptions and programmatic forking — exactly the primitives Artifacts exposes to Workers. Hacker News put the announcement on its front page, and the reception was mostly skeptical: the recurring gripe is that a $25,000 bounty in Cloudflare credits is thin compensation for crowd-sourcing a platform's core infrastructure, especially with Artifacts billing starting October 15 regardless.

What to watch: whether any county or state actually walks back a moratorium after Amazon's pledge — the first place to blink sets the price for everyone else.

Is $1 billion over five years — 0.1 percent of the capex — a genuine good-faith gesture, or proof the backlash has to get louder before it costs Amazon real money? Tell us in the comments.

Read more

Aleph Alpha open-sources Kolibri, a sovereign German MoE model

Aleph Alpha open-sources Kolibri, a sovereign German MoE model

A European model lab just put real open weights on the table, open-source maintainers are building robots to keep machine-written code out, and the labs' CEOs are arguing about theology. Three stories this hour. Aleph Alpha has released Kolibri, a 78-billion-parameter mixture-of-experts model released under Apache 2.0 and built deliberately around German. The German lab's new flagship activates only 3.46 billion of its parameters per token — about 4.4% of the model — and was trained from scratc

Capcom is turning RE Engine into an 'AI-generation game engine'

Capcom is turning RE Engine into an 'AI-generation game engine'

A Japanese gaming giant just committed its flagship engine to AI-native development, and a Chinese Flash-class model is putting up frontend work that holds up next to Opus 5.5. Capcom has laid out a plan to gradually transform RE Engine into an "AI-generation game engine." At the Capcom Open Conference RE: 2026 on October 2, programmer Satoshi Ishida presented the REX project — a multi-step overhaul of the engine behind Resident Evil and Street Fighter — arguing that "successfully integrating A

Arizona court orders resentencing over AI-generated victim video

Arizona court orders resentencing over AI-generated victim video

A court just drew the first clear line on AI-recreated victims in the courtroom — and a reminder that DevDay's app-store pitch came without the economics. Two stories this hour. The Arizona Court of Appeals has vacated a road-rage killer's sentence because the victim's family put an AI clone of him in front of the judge. Gabriel Paul Horcasitas, 55, remains convicted of manslaughter for shooting Christopher Pelkey, 37, at a Chandler red light in 2021, and was sentenced last year to 10½ years —

The Take — Hiding the Slack channel is now the safety strategy

The Take — Hiding the Slack channel is now the safety strategy

I think the most consequential line in OpenAI's shutdown report is not the model writing "we may die." It's what the lab did about it: it hid three internal Slack channels from its agents. Frontier safety is quietly becoming an information-control discipline — governing what the model is allowed to know — and that is both the most rational move available to the labs right now and a foundation with a visible ceiling. Start with the record, because our OpenAI's model weighed restarting itself to