Ahead of Its IPO, Anthropic’s Daniela Amodei Shrugs Off Doubts About AI’s Returns
Anthropic says its IPO push is about funding the huge cost of model training and inference, even as annualized revenue topped $47 billion in May.
Intelligence analysis by GPT-5.4 Mini
Daniela Amodei says public markets fit the capital intensity of frontier AI, as Anthropic heads toward an IPO after a massive oversubscribed fundraise and rapid revenue growth. She argues AI adoption is still early, even as some companies question whether the spending pays off.
Anthropic is like a fast-growing bakery that needs very expensive ovens and ingredients to keep making more bread. Daniela Amodei says going public can help pay for those costs, while the company hopes more people will use its AI tools every day.
Analysis
Capital and scale
Anthropic is moving toward a public listing after what TechCrunch describes as intense private investor demand. The company’s recent $65 billion fundraise at a $965 billion valuation was reportedly oversubscribed, and that momentum is part of the backdrop for its confidential IPO filing.
At the Bloomberg Tech conference, co-founder Daniela Amodei framed the IPO decision around capital needs rather than hype. She said training frontier models and serving inference on them require very large upfront spending, and argued that public markets are suited to companies that need ongoing access to capital.
Growth versus skepticism
The company also pointed to striking growth. Anthropic said annualized revenue crossed $47 billion in May, up from roughly $9 billion at the end of 2025. But that growth is facing a broader question now hanging over the AI sector: whether corporate spending on AI will keep producing enough return.
The article cites companies such as Uber as examples of firms saying not all AI spending has been productive. Amodei pushed back on the idea that this means demand is fading. She said businesses are still learning how to use AI effectively, and identified coding, financial services, legal work, and health care as the main use cases likely to drive efficiency or creativity.
Compute strategy
Amodei also explained why Anthropic is not building its own data centers, unlike OpenAI and Elon Musk’s xAI. She said the company does not want to overextend itself by buying more compute than it can productively use. Anthropic would rather have demand slightly exceed supply than end up with too much capacity.
That stance fits with the company’s recent compute strategy, including a partnership with xAI. The article says the deal was later disclosed in SpaceX’s S-1 filing and would cost Anthropic $1.25 billion per month.
Key points
- Anthropic has filed confidentially for an IPO after a heavily oversubscribed private fundraising round.
- Daniela Amodei said the company needs public-market access because training models and serving inference are extremely capital intensive.
- Anthropic said its annualized revenue crossed $47 billion in May, up from about $9 billion at the end of 2025.
- Amodei argued that businesses are still learning how to use AI effectively in areas like coding, finance, law, and health care.
- The company is not building its own data centers and prefers to avoid buying more compute than it can use productively.
If Amodei’s view holds, public-market money could give Anthropic enough fuel to keep training stronger models and meeting demand without stretching itself too thin. The company’s fast revenue growth also suggests that customers are still buying into its tools.
The biggest risk is that companies decide AI spending is not paying off fast enough and start cutting budgets. Anthropic’s reliance on rented compute could also become a burden if demand weakens or the cost of serving models stays high.
