Anthropic launches IPO roadshow at up to $1.4tn valuation

Anthropic sets a $70-$78 IPO range and seven founders sell about $250m each, Claude gets Sheets and Whiteboards, and ChatGPT adds verified Business Agents.

GENERAL Newsletter
H
Hugenode

Anthropic set the terms for its Nasdaq listing on Tuesday: $70 to $78 a share, for a valuation of up to $1.4tn. That is short of the $2tn it had been widely reported to be targeting. The same day, Anthropic added spreadsheets, whiteboards and embedded partner apps to Claude, and OpenAI began a beta of verified Business Agents that let brands talk to customers inside ChatGPT. For buyers, that means a major AI vendor heading into public-market scrutiny and a broader Claude pitch to non-engineers. For marketing teams, it means a new customer channel to manage.

Anthropic IPO: $70 to $78 a share, up to $1.4tn valuation

Anthropic launched its roadshow for 825,000,000 Class A shares, to list on the Nasdaq Global Select Market under the symbol “ANT”. The company is selling 801,351,354 of those shares and existing stockholders are selling 23,648,646. Anthropic gets no proceeds from the stockholder sales. Underwriters get a 30-day option on up to 123,750,000 more shares. The registration statement has not yet become effective.

The Guardian reports that Goldman Sachs and Morgan Stanley are leading the float, which could raise up to $64.4bn. Anthropic will raise a further $26bn through a separate private placement with cornerstone investors including Alphabet and Nvidia. A $9.7bn greenshoe, which applies if demand is high, would bring the total to roughly $100bn at the top of the range. That would be more than the $75bn SpaceX raised in the largest listing so far. The $1.4tn figure is fully diluted, which means it counts employee stock options. Trading is expected to begin next week.

The range also comes in below the up to $1.75tn we reported on November 8. Forrester analyst Brian Hopkins told the Guardian that last week’s midterms, in which datacentre opponents won, gave investors an excuse to correct AI valuations they were already uneasy about.

Citing the S-1, TechCrunch reports that each of the seven co-founders will sell 3,378,378 shares. That comes to just under $250m each at the $74 midpoint and just under $264m at $78. The Guardian puts the figure at about $260m each and says each founder’s remaining stake would be worth more than $18bn at the top of the range. By our arithmetic, the seven founders’ sales account for all 23,648,646 shares offered by existing stockholders. Each founder also receives one share of supervoting Class F stock, and together the group controls 51.1% of the votes. Board member Reed Hastings has indicated he wants to buy $300m of shares through his donor-advised fund. TechCrunch reads that as an early sign of strong demand.

According to the Guardian, the filing puts Anthropic’s annual revenue run rate at $81bn as of October 31, up $10bn from the previous month. Against that, it has compute purchase commitments and lease agreements of more than $70bn over the next 12 months, more than $450bn by 2030 and a further $450bn by 2045. The Guardian also cites a Financial Times report that Goldman forecasts $600bn in annual revenue by 2030. CEO Dario Amodei has described what happens if growth falls short. On Dwarkesh Patel’s podcast he said that if he bought $1tn of compute and revenue came in at $800bn, “there’s no force on Earth, there’s no hedge on Earth, that could stop me from going bankrupt.”

What it means: if you are an enterprise customer, the S-1 already shows you more about your vendor’s finances than private labs usually disclose. It shows fast revenue growth against very large fixed compute bills. The Verge reports that Anthropic is under investor pressure to grow revenue and engagement ahead of the listing. Our read is that today’s product launches are part of that push, so expect more selling into teams beyond engineering.

Claude Sheets, Whiteboards and interactive connectors take on Microsoft and Google

At an event in San Francisco, Anthropic unveiled Claude Sheets, Whiteboards and Meetings. They join the existing Claude Docs, Slides, Design and Motion. According to The Verge:

  • Sheets and Whiteboards are in beta on all paid plans.
  • Meetings is in beta for Teams and Enterprise.
  • Claude Data, formerly Dashboards, is now generally available on all paid plans.

Reports on Meetings conflict. The Register says Anthropic promised its “imminent arrival”, and The Decoder says it is launching “soon”. Check your own workspace before you plan around it.

The bigger platform change is interactive connectors. Partners such as Adobe, Atlassian and HubSpot can now embed their apps inside Claude, so users can open an Adobe PDF editor or a HubSpot campaign tracker without switching tabs. Plugins from Salesforce (beta), NetSuite and Workday (beta) come with built-in skills for sales call prep, financial analysis and HR tasks such as time off and expenses. The Verge says Anthropic’s marketplace of tools built on its Model Context Protocol has grown to more than 9,000 plugins. The Decoder reports that enterprise customers get audit logs, isolated virtual machines and a compliance API. Anthropic product lead Nicholas Lin told The Verge that knowledge workers are “the gap we’re hoping to close.”

What it means: little here is new to the market. The Verge notes that Gemini can join meetings and take notes, as ChatGPT can in some cases, and that Gemini already creates spreadsheets. Anthropic’s pitch is that you can do all of it in one Claude window. If you run IT, the governance question is the one that matters. The Register explains that artifacts are meant to be used through Claude via sharing URLs, not stored and moved around as standalone files. If you sell SaaS, interactive connectors are a new way to put your app in front of Claude users, and you should weigh whether to build one.

ChatGPT Business Agents and Business Hub open a new brand channel

OpenAI has begun a beta of verified Business Agents that represent a brand inside ChatGPT. They are available first to eligible US e-commerce and online service businesses. When a business with an agent appears in a ChatGPT answer, users can open a separate, labeled chat with that agent to ask about its products or compare options. Businesses write the agent’s instructions and supply its context.

Companies claim and verify their organization through Business Hub. OpenAI’s VP of monetization, Benji Shomair, declined to give details of verification and said methods vary by business. The hub shows mentions in ChatGPT responses and the top “discovery topics” where a business comes up. The Decoder adds that businesses can upload product feeds with prices and availability and track website clicks. Shopify merchants can connect their stores directly, and agents can serve as targets for ChatGPT ads. OpenAI says user conversations are not shared with businesses. It also says ChatGPT reaches 1.2 billion users a week, and that it plans to expand the offering to more markets and industries.

What it means: The Decoder compares the product to Google My Business, rebuilt for chat. If you are an eligible US merchant, claim your listing early and check how ChatGPT describes you now. Treat your product feed and agent instructions as customer-facing copy that needs an owner and a review process. With AI answers already cutting into search ad revenue, the discovery-topic data may be the most useful part for marketers. Outside US e-commerce and online services, nothing changes yet.

What to watch

  • Anthropic’s final IPO price and first day of trading, expected next week, and whether demand pushes the price above $78.
  • When Claude Meetings actually reaches Teams and Enterprise customers, and which partners add interactive connectors next.
  • How OpenAI verifies businesses, when Business Agents expand to other markets and industries, and how ads tied to those agents work.
  • Whether Anthropic’s revenue keeps pace with more than $70bn in compute commitments over the next 12 months.

Sources