Anthropic: agents, datacentres, lobbying

Anthropic is expanding on three fronts at once: data center capacity, always-on AI agents, and Washington lobbying. (cnbc.com, axios.com, testingcatalog.com) On Thursday, April 23, CNBC reported that Anthropic is hiring a London-based “Transaction Principal” to negotiate European data center capacity deals, with pay listed at £225,000 to £270,000. The job posting says the role will source commercial deals, manage developer outreach, and negotiate term sheets across Europe. (cnbc.com) The same CNBC report said Anthropic is also hiring for a similar Australia-based role and wants experience in “FLAP-D” markets: Frankfurt, London, Amsterdam, Paris, and Dublin. A source told CNBC the company is evaluating direct capacity deals with data center developers “across the world.” (cnbc.com) That hiring push follows a burst of infrastructure commitments this month. Anthropic said on April 20 that it will spend more than $100 billion on Amazon Web Services over 10 years for up to 5 gigawatts of new capacity, and on April 6 it expanded a Google-Broadcom arrangement covering multiple gigawatts of next-generation Tensor Processing Unit capacity starting in 2027. (anthropic.com, anthropic.com) Broadcom said that expanded arrangement gives Anthropic access to about 3.5 gigawatts of computing capacity tied to Google’s artificial intelligence processors. CNBC said U.S. hyperscalers’ artificial intelligence infrastructure spending is expected to top $600 billion in 2026. (cnbc.com, cnbc.com) At the same time, Anthropic is spending more heavily to shape policy. Axios reported on April 21, citing federal lobbying disclosures, that Anthropic spent $1.6 million on lobbying in the first quarter of 2026, compared with $1 million for OpenAI. (axios.com) Axios said both companies posted their biggest quarterly lobbying totals yet, and Anthropic’s spending was up from $360,000 a year earlier. That increase comes as Anthropic is also fighting the Pentagon’s designation of the company as a supply-chain risk for classified use, according to CNBC and the Associated Press. (axios.com, cnbc.com, apnews.com) The product side is moving in parallel. TestingCatalog reported on April 1 that Anthropic appears to be testing an unreleased tool called Conway, a standalone Claude environment with webhooks, Chrome integration, notifications, and its own extension system. (testingcatalog.com) Those features point to software that stays running in the background and wakes up when another service pings it, instead of waiting for a user to type a prompt into a chat box. TestingCatalog said Conway references public URLs for wake-up calls, connected clients, and installable add-ons packaged as “.cnw.zip” files. (testingcatalog.com) Anthropic declined to comment to CNBC on its European data center plans, and Conway has not been formally announced by the company. But the April hiring, spending, and leaked product details all point in the same direction: Anthropic is building the physical capacity, software hooks, and political footing needed to run larger AI systems at scale. (cnbc.com, testingcatalog.com, axios.com)