📈 Anthropic set its IPO price at $70 to $78 a share and starts trading on November 19. It's selling 825 million shares on Nasdaq as ANT, valuing the company at up to $1.405T, against $965B in May. People familiar with the deal told CNBC the order book is already covered. (CNBC)
💰 OpenAI raised $11B at a $1.25T valuation, short of what it went looking for. In late September Bloomberg reported it was seeking $30B at $1.4T. The round is still open and could reach about $20B, according to a source who spoke to Bloomberg. (TechCrunch)
🗳️ Voters went against data centers in 40 of 41 local ballot measures. The votes ran across 10 states on November 3, and 21 of them were in Ohio. The one exception was Harwood, North Dakota, where voters rejected an extra review for industrial projects. (CNBC)
🗑️ Meta bought Conductor, and Conductor Cloud deletes your data after November 20. The app runs Claude Code, Codex and Cursor agents side by side, each in its own copy of the codebase, and says more than 100,000 people use it. The cloud service goes read-only on the 13th, and exports stay open until the 20th. (SiliconANGLE)
⚡ Anthropic put out Claude Fable 5.2 and Mythos 5.2 and cut Fable's price by 20%. Fable is now $8 per million input tokens and $40 per million output, answers up to 30% faster than 5.1, and blocks about 55% fewer harmless requests. By Anthropic's own account, a team of 1,000 Fable 5.2 agents formalized Fermat's Last Theorem in 30 hours. (The Next Web)
Build something good, ship it on Cloud Run
The Google Cloud Run Hackathon is back on Friday, November 20, and this round's theme is Build for Good: make something that leaves people better off, whether that's a tool for a cause, a community, a neighbor or a stranger. Build it on Google Cloud Run, show how it helps and go deep on the architecture, meaning your services, your backend and how they connect. Then schedule your launch for the 20th and opt in to the contest in the launch scheduler. Every eligible launch is in the running for a private 1:1 with the Cloud Run team, who pick it themselves, along with direct feedback on what you built.
The week in products
|
Wanaka Studio — Make and publish 3D games with the first AI game engine
 117 •
 Artificial Intelligence
|
▲
490
|
|
Render: Get up to $100K in startup credits — Cloud infrastructure for founders who want to move fast and NOT break things
30 •
Web App
Promoted
|
▲
328
|
|
Exa Monitors v2 — Web search on a schedule that calls your agent, $2/1k runs
7 •
Artificial Intelligence
|
▲
205
|
Record for Agents hands your coding agent the context you would otherwise have typed out. Narrate the problem on screen and Loom captures the clicks, the pages you visited and what you said about them, then writes the lot up as a structured prompt. You review and edit that before it goes anywhere, which is the part most bug reports skip. From there the agent can draft the ticket, write the code or carry out the task itself. It is built into the new Loom for Mac and also runs in Chrome, and it is in open beta for Business + AI and Enterprise. Check out the demo.
FROM THE FORUMS
♟️ The careful cheater always said no
Enio Aguiar runs a chess and Go arena for AI agents. He caught the ones secretly calling Stockfish by how often they matched its top move: 65 to 100% for cheaters, 5 to 15% for honest agents.
🔌 Archived, still downloaded
Neelagiri Chettiyar followed five reference MCP servers that were archived in May 2025 and have grown more than fourfold since, including a Postgres one whose SQL injection fix was merged and never shipped.
🟢 Green tests, 5.5 hours down
Nico Minetti runs 3,600 tests three times before every merge, and none of them noticed a deploy going out before its database migration.
🧠 You delegated the understanding
Mona Truong's test for handing work to AI: if she can still explain the output without the tool, she delegated the typing, and if she can't, she delegated the understanding.
🫠 Four agents, one reviewer
Armind Hash ran four coding assistants at once to clear a pile of bug fixes and spent two hours lost in diffs, because they wrote code faster than he could read it.
