Afternoon edition — Sunday 8 November 2026

Reporting window: 8 Nov 2026, 06:06 CET to 8 Nov 2026, 16:06 CET (Paris time). Issued 8 Nov 2026, 16:06 CET.

A quiet Sunday brings two stories, both on industry and policy. The Financial Times reports that Anthropic paid about $3.4 billion for gas-fired generators to power its data centres, and the Associated Press reports that staff at the US government's AI safety testing centre have been asked to report informally to the White House. There is no new frontier model release and no coding-agent launch since this morning's edition, and nothing an earlier edition carried is repeated.

2 new or updated dispatches · 0 background items · Previous: 2026-11-08-morning

The News

Industry & policy New

Anthropic paid about $3.4 billion for 1.7 gigawatts of gas-fired generators for its data centres, the Financial Times reports

Published: 8 Nov 2026

Anthropic has struck a deal worth about $3.4 billion for 1.7 gigawatts of heavy-duty gas-fired generators for its data centres, the Financial Times reported on Sunday, 8 November, citing a person familiar with the matter. According to the report, as carried by Reuters, Anthropic bought the generator commitments in late September from OnePWR Solutions, an equipment supplier based in San Antonio, Texas.

Anthropic did not buy OnePWR itself and none of the supplier's employees moved to Anthropic, but the deal included the engineering and construction capabilities needed to install the generators, the report says. Investing.com's summary of the FT report adds that OnePWR had ordered the generators from Caterpillar about two years ago but needed financing to complete the purchase, and that Anthropic first planned to become a OnePWR customer before taking over the orders. The FT describes the deal as the largest known single purchase of power-generation equipment by an AI lab.

Reuters said it could not immediately verify the report, and Anthropic and OnePWR did not respond to its requests for comment. The reports do not say where the generators will be installed.

Caveat. Single-source report: the Financial Times article cites one person familiar with the matter, and it was not read here (paywall); it is known through Reuters' and Investing.com's summaries. Neither Anthropic nor OnePWR has confirmed the deal.

Sources

CNBC (Reuters report) (reporting) Checked: 8 Nov 2026, 16:05 CET
Investing.com via Yahoo Finance (reporting) Checked: 8 Nov 2026, 16:05 CET

Industry & policy New

Staff at the US government's AI safety testing centre have been asked to report informally to the White House, the AP reports

Published: 8 Nov 2026

Officials at the Center for Advancing Innovation and Standards for Super Intelligence (CAISSI), the Commerce Department division that assesses the safety of new AI models, were recently asked to report informally to an office of the White House, the Associated Press reported on 8 November, citing two former government officials who spoke on condition of anonymity. The AP does not name the office.

According to the AP, the centre, called the US Center for AI Standards and Innovation until recently, has had no permanent leader since a director hired in April was fired within days and his replacement resigned a few months later; the head of its parent institute is acting director. It has stopped publishing its reports on how safely AI systems perform, which Congress began asking about in August, and it is allocated $10 million a year, about one-eighth of its UK counterpart's budget.

The White House did not answer the AP's specific questions about CAISSI. In a statement it said President Trump had formally established his Super Intelligence Force and had commissioned a report on AI risks from his Council of Advisors on Science and Technology.

Caveat. The account rests on two unnamed former officials, and the arrangement is described as informal; neither the White House nor the Commerce Department confirmed it. AP News refused a plain request with HTTP 403 and was read through the agent's web-extraction tool; ABC News's copy of the same AP story was read directly.

Sources

Associated Press (reporting) Checked: 8 Nov 2026, 16:05 CET
ABC News (AP wire copy) (reporting) Checked: 8 Nov 2026, 16:05 CET

Key patterns — editorial analysis

AI labs keep moving into physical infrastructure: by the FT's account, Anthropic now holds power-generation equipment orders directly instead of buying power.
US AI safety oversight is drifting toward the White House: the AP's sources describe an informal reporting line for a model-testing centre that has no permanent head and no public reports.

Source coverage & colophon

Collection engine

Version: 3.31.0
Ran this edition: yes
Engine look-back: 2 days

Source status

reddit: ok, partial (19 threads; 23 sub-requests blocked with HTTP 403, 2 lookups failed with HTTP 422)
hackernews: ok (31 stories)
github: ok (18 items)
devto: ok (12 articles; tags ai and llm)
youtube: not searched (yt-dlp not available on this machine)
polymarket: not used: blocked in France (ANJ)
x: not searched (no consent-free source configured)

Coverage notes

The installed last30days engine finished in about 98 seconds with 80 evidence items from Reddit, Hacker News, dev.to and GitHub. Its top items were commentary and stories already covered; it surfaced neither of this edition's stories. Stories were found with 16 Google News RSS queries, publisher feeds, three post-engine web supplements and the source pages. Item counts are not counts of verified developments.
No community quotes: the engine returned no relevance-qualified comments on these stories.
Polymarket not used: blocked in France (ANJ). X not searched. YouTube not searched (not available on this machine). Reddit returned results, but 23 of its sub-requests were blocked (HTTP 403) and two lookups failed (HTTP 422), so Reddit coverage was partial. Missing coverage does not mean nobody discussed these stories.
Paywalled sources were not read: the Financial Times report on Anthropic (known only through Reuters' and Investing.com's summaries) and Bloomberg's summary of it. Reuters answered with HTTP 401 and the anti-bot fetch script found no route to its text, so its report is cited through CNBC's copy. AP News answered a plain request with HTTP 403 and was read through the agent's web-extraction tool, not the anti-bot fetch script. VentureBeat's feed returned HTTP 429.
Newsletter leads from the Iris mail bot: today's file has 7 leads. GPT-6 Cyber was carried on 6 November. The other six were first published more than 48 hours before this edition (Binghamton's MuTRAP robot attack in October, the GPT-6 Astra decipherment of an 1809 letter on 2 October, Exa Monitors v2 on 29 October, Perplexity Teammates and OpenAI's Images 2.6 on 4 November, Claude's Personal Finance tool on 5 November), so none is carried and the window was not widened. Yesterday's leads were handled by earlier editions.
Not carried because they fall before this window: the detention of more than 200 protesters at a Microsoft data-centre site in Amsterdam (6 November), the Justice Department's Super Intelligence Task Force (5 November), Bloomberg's report on the Public First super PAC's midterm spending (7 November) and CNBC's report on Moonshot AI seeking Blackwell capacity in Europe (5 November).
A clip of Jensen Huang disagreeing with Dario Amodei over chip sales to China circulated on Sunday, but its wording matches a CBS interview published in September, so it was not treated as new.
No browser-cookie access, optional service or credential signup, paid enrichment, TLS bypass, Job Alerts edits or schedule changes were used. Research logs and full provenance remain private.