Anthropic’s $895 billion compute commitments test AI financing
Anthropic’s latest filing shows just how expensive this race is: $17 billion in third-quarter revenue, a $23 billion quarterly loss and nearly $895 billion in compute commitments. Those long-term compute deals lock in capacity for Claude. But with an IPO approaching, the question is whether investors will keep footing the bill, and credit agencies are reportedly eyeing a “junk” rating.
It’s not just Anthropic. Four Big Tech companies are on pace to spend more than $720 billion this year, and investors are getting spooked. Alphabet shed half a trillion dollars in its worst day ever, lenders are starting to call the shots, and OpenAI raised less than half of its $30 billion target. So much for the money being endless.
- NY Times: Big Tech’s Data Center Blitz Is Burning Through Cash and Scaring Wall Street
- NY Times: The Tech Tumult Rocks Wall Street
- TechCrunch: Cracks are emerging in the AI spending boom
- TechCrunch: AWS will be a $1T business, Jassy says. But first, it will spend big.
- NY Times Opinion: What Anthropic’s I.P.O. Filing Says About the Future of A.I.
- Transformer: Was this the AI election?
OpenAI pledges local protections as data-center backlash grows
After voters backed data-center limits in 40 of 41 local measures, OpenAI is promising to play nicer with host communities: consulting locals, publishing annual reports, paying for electricity and grid upgrades, and dropping most NDAs with public officials. The catch? It only covers campuses OpenAI owns. Partner sites are on their own.
Voters cited power bills, water use and pollution, and a separate tally tells the same story. Democratic gains could also mean tighter AI rules and construction pauses in major states. Meanwhile, OpenAI’s Georgia campus will test up to 1,000 megawatts of demand response under its new Grid Protectors program, Meta expects tougher scrutiny, and stocks erased their post-election rally.
The industry spent the year telling us AI is inevitable. Voters just argued back.
- WSJ: The Election Day Revolt That Threatens America’s AI Boom
- Ars Technica: Big Tech claims voters’ AI backlash is “anti-Trump” not “anti-tech”
- Stratechery: AI on the Ballot
- Transformer: What the blue tsunami means for AI policy
- TechCrunch: OpenAI is already trying to shape the 2028 election
- WSJ: How the Data Center Backlash Could Play Into the Hands of Big Tech
- The Texas Tribune: Under Hinojosa, Texas data center approvals could stay paused
- WSJ: What AI Leaders Can Learn from the Midterms
- OpenAI: Setting a new standard for responsible AI infrastructure
Google’s school AI opt-out also cuts off Search
Google will turn off AI tools by default for K–12 accounts early next year. Sounds great, until you read the fine print: schools that reject AI entirely also lose Google Search, since the AI features are baked in. Critics say that’s not much of a choice.
For schools that opt in, Google is pushing Guided Learning, which coaches students through problems instead of handing over answers. It’s in Classroom globally and being piloted in Search at select U.S. schools. Nice idea. But there’s still no AI-free version of Search.
OpenAI’s GPT-6 Cyber reports major Firefox and Linux findings
OpenAI says its new GPT-6 Cyber found 81 previously unknown high-severity Firefox flaws and built working exploits for 948 Linux kernel vulnerabilities, with fixes and disclosures underway. Great news for defenders. Also a little terrifying: TechCrunch asks whether discovery this fast will outpace patching.
On Google’s side, Gemini Omni 1.2 Flash now generates clips up to 15 seconds at 30% lower prices, and its October roundup included Gemini 4 Argon and a pre-release cybersecurity model, Gemini 4 Carbon.
Amazon and DoorDash report few orders from outside AI agents
Remember when AI agents were going to do all our shopping? Not yet. Amazon CEO Andy Jassy says third-party agents aren’t meaningfully denting its ad business, even as Alexa for Shopping reached more than 400 million customers this year. DoorDash, which handles about 3 million deliveries a day, has received only “several hundred” orders from outside agents.
Both are building their own agents anyway. The real fight isn’t whether AI shops for you. It’s whose AI does.
Quote of the Week
[With AI] the benefits will only be achieved if we build the technology in the right way, and — so long as we use the time we gain well — it is worth taking unusually deliberate care to get it right.
Dario Amodei, Anthropic CEO, NY Times
More links this week
- Nikkei Asia: US tech giants' hidden AI debt swells 90% to $3.2tn in 6 months
- WSJ: AI Investors Are Racing to Get Insurance on the Biggest Market Bet in History
- TechCrunch: Data centers are turning to old jet engines as the wait for power grows
- NY Times: OpenAI and Anthropic Set Up ‘War Rooms’ to Fight Disinfo About Midterm Elections
- TechCrunch: Google faces backlash for plan that would turn off Search for schools that restrict AI
- NY Times: Parents Sue Google Over A.I. Safety Risks on School Chromebooks
- WSJ: AMD Beats Wall Street Expectations as it Pivots to Data-Center First Status
- NY Times: Driven by A.I. Boom, Marvell and Lumentum Announce $115 Billion Merger
- TechCrunch: Almost no one uses AI agents to book their Uber, company says
- WSJ: Silicon Valley Promises AI Will Cure Disease. Drugmaker CEOs Aren’t Buying It.