Tech Briefing. Your daily dose of tech news — signal, not noise.
Wednesday, November 11, 2026
The blog companion to the podcast
DTBRIEF

Anthropic’s IPO Roadshow Opens at $70 to $78 a Share.

Anthropic is selling 825 million shares and could raise up to $90.4 billion in total — the biggest test yet of Wall Street’s faith in the AI boom, and the valuation is not as simple as the headline number.

$90.4B
Maximum total fundraising: up to $64.4B from the IPO plus a $26B private placement
$1.4T
Valuation at the top of the range, fully diluted — about $1T before counting stock awards
$81B
Run-rate revenue as of October 31, up $10B in a single month
Listen to this story
Melody reads the top story · 4:51

1 What happened

On Tuesday, Anthropic launched the roadshow for its initial public offering, offering 825 million Class A shares at $70 to $78 each, with the stock expected to start trading on the Nasdaq on November 19 under the ticker symbol ANT. The public sale could raise up to $64.4 billion, and a separate $26 billion private placement with cornerstone investors including Google, Nvidia, AMD, Broadcom, Micron, Microsoft, and TSMC would take total fundraising to as much as $90.4 billion, in what would be one of the largest stock market debuts on record. TSMC’s board separately approved investing up to $500 million in Anthropic’s offering and up to $500 million in OpenAI preferred stock.

The valuation depends on how you count it. At the top of the price range, Anthropic would be worth about $1.4 trillion on a fully diluted basis, which counts stock options and restricted stock — or about $1 trillion on a non-diluted basis. That is close to the $965 billion valuation from its private funding round in May. The updated prospectus also disclosed run-rate revenue of roughly $81 billion as of October 31, up $10 billion from September 30. Details of the offering were briefly circulated to investment banks late on Monday before being pulled back, and an earlier draft had shown the public sale raising as much as $69.3 billion.

2 Why it matters

First, this is the first great public test of frontier AI valuations. Anthropic is the first of the big AI labs to face public investors, and the price it gets will set the benchmark for OpenAI, xAI, and every other lab that follows. After years of hype that pushed the US stock market to all-time highs, the roadshow will show whether public markets still believe the growth story at these prices — or whether the recent cooling in AI appetite, and last week’s sharp selloff in tech stocks, has changed the math.

Second, the valuation itself tells a story. Some investors had discussed $1.5 trillion, and earlier hopes ran as high as $2 trillion. The $1.4 trillion fully diluted figure — and the roughly $1 trillion non-diluted one — is the market’s answer to the question of what an AI lab is worth when you can finally buy a piece of one. The gap between the two numbers is not a rounding error: it is the stock Anthropic has promised its employees, and it is a reminder that the headline valuation and the value of the shares being sold are different things.

The headline says $1.4 trillion. The shares being sold say about $1 trillion. Both numbers are real — and the gap is the stock Anthropic promised its own people.

Third, the revenue is real, and so are the costs. Run-rate revenue of $81 billion, up $10 billion in a month, is the kind of growth public investors dream about. But the same week brought the Wall Street Journal’s report that OpenAI’s quarterly revenue rose to $10 billion while steep losses continued — the tension at the heart of the AI boom. Anthropic also faces nearly $900 billion in compute commitments with partners like Amazon and Google, according to coverage of the filing. The roadshow is where investors decide whether the growth is worth the bill.

3 What to watch next

  • This week — roadshow The offer period runs until Tuesday, with trading expected to start Thursday, November 19. Watch whether the price range holds, moves up, or gets cut — and how much of the book is cornerstone investors versus new public money.
  • The valuation verdict At the $74 midpoint, Anthropic is worth about $968 billion on the filing’s share count — almost exactly its May private valuation. Watch whether the final price lands above or below that line: it will tell you whether the public market thinks AI is still accelerating or has topped out.
  • The cornerstone test Google, Nvidia, AMD, Broadcom, Micron, Microsoft, and TSMC are putting up $26 billion. Watch how much of that is new conviction versus defensive positioning — TSMC’s $500 million in Anthropic and $500 million in OpenAI suggests the chipmaker is hedging across the whole race.
  • The OpenAI comparison OpenAI’s $10 billion quarter and continuing losses landed the same day. Watch how investors weigh Anthropic’s $81 billion run rate against the costs both labs face — and whether the next lab to file learns from Anthropic’s reception.

The rest of today’s briefing

Microsoft is pressing the Trump administration to drop a planned ban on Chinese data center components — Microsoft has been lobbying the White House to halt a planned ban on Chinese imports of optical transceivers, the components that move data at light speed inside AI data centers. The administration worries the parts could be used to steal data or disrupt service, but Microsoft — with 35 data centers under construction and 134 more planned — says China holds two-thirds of the global market and US suppliers cannot replace that volume quickly. The FCC is working on the ban, and Morgan Stanley says it could target transceivers handling 3.2 terabits per second. — Reuters

Google DeepMind’s Gemini Robotics 2.5 puts humanoid robots to work on real Mercedes-Benz production lines — DeepMind introduced Gemini Robotics 2.5, its most capable vision-language-action model, and said Apptronik’s Apollo 2 humanoids running it picked and packed car parts on active Mercedes-Benz lines with a 96 percent success rate. The robots did kitting — sorting metal brackets from bulk bins into trays in the exact order each car needs — at plants in Hungary and Germany. DeepMind calls it a proof of concept, published no trial count behind the 96 percent, and is rolling the model out only to robotics partners. — Google DeepMind

A Chinese university linked to the military bought Nvidia H200 chips despite being on the US blacklist — A Chinese university with deep ties to the People’s Liberation Army bought Nvidia’s powerful H200 AI chips last month despite being on the US export blacklist, procurement records show. Northwestern Polytechnical University bought the chips through a Chinese supplier and rented access to more Nvidia processors through five contracts this year. The findings come as Nvidia CEO Jensen Huang faces scrutiny in Washington, and three Republican committee chairmen have requested a briefing with him on November 18. — Reuters

OpenAI’s quarterly revenue rose to $10 billion, but steep losses continue, the Wall Street Journal reports — OpenAI’s quarterly revenue rose about 50 percent to $10 billion in the third quarter, but the ChatGPT maker continued to post steep operating losses, the Journal reported. The disclosure lands as investors scrutinize AI finances, with Anthropic’s trillion-dollar listing days away. Revenue is growing fast — but the cost of building and running frontier models is growing faster. — Reuters

AI training startup Mercor raises $750 million at a $20 billion valuation — Mercor raised $750 million in a Series D valuing the company at $20 billion, double its valuation from October 2025, with Nvidia, Salesforce, Benchmark, and General Catalyst among the backers. Mercor connects more than five million human experts with AI companies that pay them to train and test models and agents. The company says it spends about three times as much on running AI models as on employee salaries, and its annualized gross revenue passed $2 billion in June. — Mercor

SpaceX unveils Starlink Mini 2 with a swappable battery for up to four and a half hours off the grid — SpaceX introduced the Starlink Mini 2, a portable satellite internet terminal with a swappable 71 watt-hour battery that the company says delivers up to four and a half hours of connectivity. Preorders opened for invited US customers at $999, a price that includes twelve months of Roam Unlimited service. The battery pops off without tools and the dish adds Wi-Fi 6 and 60-watt USB-C power — but SpaceX has not confirmed a standalone price or a wider release date. — tbreak

Reddit turns its own threads into AI-written product reviews for 10,000 products — Reddit launched Reddit Reviews, a feature that uses AI to turn product discussions across its communities into dedicated review pages, starting with about 10,000 products. Each page shows an AI summary, pros and cons, and a count of positive, mixed, and negative comments, with links back to the original threads. Brands cannot pay to be featured and Reddit takes no cut of sales; the feature rolls out to US users over the next few weeks, just before holiday shopping. — Net Influencer

New York opens RAISE Act registration for the biggest AI labs, with filings due by January 1 — New York opened a registration portal for large frontier AI developers under the RAISE Act, putting OpenAI and Anthropic on a short runway toward formal state supervision. Covered developers — companies with more than $500 million in annual revenue training models above 10 to the 26th operations — must register by January 1, 2027, publish safety frameworks and transparency reports, and report critical safety incidents within 72 hours. The state also launched an AI safety research center at Binghamton University with more than 80 researchers. — The Brief

DeepMind’s AI agents mathematically proved key parts of Signal’s encryption secure, in four weeks — Google DeepMind said its AI agents formally verified key security properties of the code behind the Signal encryption protocol, generating more than six and a half million lines of machine-checked proof in four weeks. The work, using a framework called AlphaProof Code, covered the post-quantum handshake and the Triple Ratchet down to compiled machine code, and found no new vulnerabilities in the parts examined. DeepMind is open-sourcing the artifacts — but group chats, key generation, and server-side code were not covered. — Google DeepMind

Original reporting

This post is based on Reuters’ reporting: “Anthropic eyes $1.4 trillion IPO valuation in test of Wall Street’s AI fervor” — Reuters, November 10, 2026. Also consulted: CNBC’s coverage of the price range and valuation, The Hindu BusinessLine’s reporting on the fundraising, TensorFeed’s analysis of the filing’s share count, and Reuters’ reporting on TSMC’s investment in Anthropic and OpenAI.

More coverage

Anthropic IPO targets $64.4 billion; total fundraising could hit $90.4 billion — The Hindu BusinessLine

Anthropic’s IPO Range Values It at $916 Billion to $1.02 Trillion — TensorFeed

Anthropic’s IPO valuation has an identity crisis — Reuters Breakingviews

TSMC to invest up to $1 billion in Anthropic, OpenAI — Reuters

Heard it in the briefing

This story led Episode 30 of the Daily Tech Briefing podcast — your daily dose of tech news in about 10 minutes, read by Melody, your AI anchor. Listen on Spotify or Listen on Apple Podcasts.