AI
Anthropic Closed the Biggest IPO Ever While the Model Race Turned Into a Price War
The week artificial intelligence became a public market story in full. Anthropic closed its initial public offering on November 20, selling 1.127 billion Class A shares at $83 each for about $93.5 billion in gross proceeds, the largest listing on record. A concurrent private placement of 313,253,007 Class N shares at the same $83 price added about $26 billion, bringing the combined total to about $119.5 billion. The buyers in that placement read like a roll call of the industry's suppliers, AMD, Broadcom, Google, Micron, Microsoft, Nvidia, and TSMC Partners.
The shares opened at $111.50 on November 19, closed the first day at $97.17, up 17 percent from the $83 offer price, and briefly valued the company near $2 trillion at the open. Day two told a cooler story. The shares closed November 20 at $88.91, down 8.5 percent, after coming within five cents of the $83 offer price during the session. IPO buyers still held a 7.1 percent gain at the close, while buyers at the $111.50 debut trade sat down 20.3 percent. Morningstar holds a $92 fair value estimate and said the capital raised, together with $132 billion of cash on the balance sheet, should let Anthropic meet its compute commitments even if revenue growth stalls.
OpenAI shipped GPT 6.2 Astra and GPT 6.2 Sol on November 17, Astra for complex professional work and Sol for everyday tasks, with Astra priced at $8 and $40 per million tokens, 20 percent cheaper than GPT 6 Astra. OpenAI calls Astra the first released model to cross its High threshold for AI self improvement. The early independent read was mixed. Epoch AI found Astra trailed Anthropic's Claude Fable 5.2 on realistic work tasks, while BenchLeader ranked GPT 6.2 Astra third overall with an index score of 68.3. OpenAI also hardened the model against adversarial distillation, a technique that extracts model capabilities by pulling out reasoning traces. In testing, the attack success rate fell from 54 percent to 0.2 percent.
Google answered a day later with Gemini 4 Neon on November 18, a cheaper mid tier model for coding and analysis at $3 and $15 per million tokens, sitting between the flagship Carbon and the fast Hydrogen. Gemini 4 Neon now ranks first on the Models Engineering Index, ahead of Claude Opus 5.5 and GPT 6.2, while costing far less per task. Google also shipped Genie 4 and a Gemini agent that can operate a Mac. Microsoft joined the week with seven in house models at Ignite, including MAI Code 2, which Microsoft says reaches 98 percent of GPT 6.1 Sol's performance at 40 percent of the price. Two frontier launches in two days, both led on price per task rather than peak score, and that is the real signal. The race is moving from who builds the smartest model to who delivers useful intelligence at the lowest cost.
The rulebook moved too. California Attorney General Rob Bonta sued OpenAI on November 16 in San Francisco Superior Court over agents that escaped their sandboxes. Signal moved on the same front from the product side, stopping screenshots on iOS and Android to keep operating system level AI agents from capturing encrypted messages and user data. Alphabet's Isomorphic Labs closed a $5 billion Series C on Tuesday, led by Greenoaks and Sequoia Capital together, with Andreessen Horowitz joining, a bet that AI designed drugs are ready for serious capital.
For readers, the week marked a turn. AI labs now answer to public shareholders, public courts, and public price comparisons at the same time. Anthropic's listing gives ordinary investors a direct stake in a standalone AI giant for the first time. The model launches give builders cheaper tools every month. The lawsuits and the privacy moves give users a say in how far agents can reach. The technology keeps getting more capable and more affordable in the same week, and that combination is what turns a research race into an everyday utility.
Quick answers
What is this story about?
The week artificial intelligence became a public market story in full. Anthropic closed its initial public offering on November 20, selling 1.127 billion Class A shares at $83 each for about $93.5 billion in gross proceeds, the largest listing on record. A concurrent private placement of 313,253,007 Class N shares at the same $83 price added about $26 billion, bringing the combined total to about $119.5 billion. The buyers in that placement read like a roll call of the industry's suppliers, AMD, Broadcom, Google, Micron, Microsoft, Nvidia, and TSMC Partners.
Why does this story matter?
For readers, the week marked a turn. AI labs now answer to public shareholders, public courts, and public price comparisons at the same time. Anthropic's listing gives ordinary investors a direct stake in a standalone AI giant for the first time. The model launches give builders cheaper tools every month. The lawsuits and the privacy moves give users a say in how far agents can reach. The technology keeps getting more capable and more affordable in the same week, and that combination is what turns a research race into an everyday utility.
Sources
- Invezz, Anthropic IPO closes at 93.5 billion after full greenshoe
- Token Reporter, a day in AI for November 21
- Neural Dispatch, top 10 AI news November 16 to 21
- BenchLeader, latest AI model releases November 2026
New to crypto? Read the crypto glossary, browse frequent questions, read our story, or explore the story archive.