Anthropic is preparing to go public, and its potential valuation highlights just how unusual the rules of the game have become when it comes to the largest AI companies. According to Reuters, the company is forecasting revenues of between $190bn and $200bn in 2028. Bankers and investors are said to be using these forecasts, rather than current results, as one of the main benchmarks for valuing the IPO.
The scale of the expected growth is substantial. In May, Anthropic reported an annualised revenue rate exceeding $47 billion, compared with around $9 billion by the end of 2025. In June, the company confidentially filed documents relating to an IPO in the US. A few days ago, the *Financial Times* reported that some investors see the potential for the company to be valued at at least $2 trillion.
However, such a valuation is based primarily on the assumption that revenue will grow faster than the costs of infrastructure, computing power and model development. This is crucial, as the technology sector continues to allocate huge sums to AI infrastructure. Recent results from Microsoft and Amazon have improved sentiment surrounding these investments, but the market is increasingly scrutinising which companies will actually translate expenditure into sustainable growth in profits and cash flow.
Anthropic’s IPO could therefore set a new benchmark for AI companies. A successful debut would reinforce the argument that investors are prepared to pay for the future scale of the business, even with high current costs. Weaker growth or a slower improvement in margins, on the other hand, could undermine not only Anthropic’s valuation but also the high valuations across the entire AI sector.
