Anthropic’s planned initial public offering is drawing attention to its valuation ambitions and the wealth its founders are preparing to realise.
Meanwhile, reports that a Chinese university linked to the People’s Liberation Army obtained restricted Nvidia processors highlight another challenge for artificial intelligence: controlling access to powerful technology as competition intensifies.
According to Anthropic’s updated IPO prospectus, all seven co-founders intend to sell shares worth approximately $1.75 billion combined in next week’s offering. Each plans to sell 3,378,378 shares.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
At the midpoint of the proposed $70–$78 price range, each would receive close to $250 million. The founders include chief executive Dario Amodei, president Daniela Amodei and five former OpenAI colleagues who established Anthropic in 2021.
The planned sales do not mean the founders are abandoning the company. They are expected to retain substantial ownership and, through special voting arrangements, significant influence over corporate decisions.
The transactions illustrate how an IPO can provide insiders with liquidity while bringing a private technology company into public markets. Anthropic could command a valuation around $1.4 trillion, reflecting investor enthusiasm for its AI models and commercial growth.
It also gives investors a clearer view of founder shareholding, voting power and the structure of the proposed sale. Still, the scale of the proposed cash-out may prompt questions about executive incentives.
Investors will want to understand how much capital the company raises, how proceeds support research and infrastructure, and whether leadership remains committed to long-term development.
In a sector requiring enormous spending on computing capacity and data centres, confidence in management will be important. Meanwhile, a Reuters investigation has exposed weaknesses in US restrictions designed to prevent sensitive AI chips from reaching Chinese military-linked institutions.
Northwestern Polytechnical University, a member of China’s “Seven Sons of National Defense,” purchased two Nvidia H200 processors in October for 874,700 yuan, approximately $123,000, according to procurement records.
The university obtained the chips through a domestic supplier, not directly from Nvidia. The institution has been on the US Commerce Department’s Entity List since 2001, with restrictions tightened in 2019.
Export licences for controlled technology destined for the university are generally subject to denial. Yet records show it also rented computing capacity involving H200, A100 and A800 processors through Chinese companies.
Reuters reviewed 26 records indicating other universities and research institutions obtained restricted Nvidia chips or access to computing services powered by them.
The findings raise concerns because advanced processors can support civilian and military applications, including weapons design, intelligence analysis, engineering simulations and AI training.
They expose the difficulty of enforcing export controls when suppliers, servers and intermediaries operate across complex supply chains. The records do not establish where every chip originated or where rented servers were located, complicating verification.
The developments reveal two sides of the AI boom. Anthropic’s IPO demonstrates the financial value investors place on leading AI companies, while the Nvidia case shows how strategic competition shapes access to essential hardware.
Governments must protect national security without unnecessarily restricting innovation. Companies must balance growth, transparency and responsible development.
The next phase of the AI race will be measured not only by valuations and model performance, but also by who controls the infrastructure and how its use is governed.