KM4BI KM4BI

Where Brad Gerstner Is Investing Billions

Brad Gerstner lays out his AI, compute, data center and ownership theses.

Brad Gerstner frames the current AI market as one that has already survived several mini-corrections, but whose underlying demand remains strong. In his view, the decisive shift came when inference-time reasoning improved and agents began to suggest a much larger future market for tokens and compute.

A major theme is the importance of Anthropic’s revenue growth. Gerstner argues that Anthropic’s performance has been critical to the broader AI market narrative. Without those revenues, he believes the stock market could have been materially weaker. OpenAI and Google are described as performing well, but Anthropic is presented as the company that most clearly validated the scale of AI revenue demand.

He takes a middle position on the debate over token spending. The bearish view says AI revenue is inflated by waste and experimentation; the most optimistic view says every dollar is already perfectly productive. Gerstner rejects both extremes. Companies will optimize, but adoption is still early enough that usage can keep expanding through that optimization cycle.

That distinction shapes his view of software. He separates companies that are in the “token flow” from those whose products may compete more directly with AI models. Databricks, Snowflake and ClickHouse are discussed as infrastructure beneficiaries, because more AI usage can mean more database queries. Traditional front-end SaaS businesses may face more pressure if models absorb more of their value.

Data centers are treated as both an investment and a national strategy issue. Gerstner calls a data center moratorium disastrous for the United States, arguing it would hurt GDP growth, jobs, productivity and the AI race with China. At the same time, he acknowledges local concerns about water, power costs and direct community benefits, and says a tangible dividend for host communities is needed.

SpaceX and Elon Musk enter the discussion through compute. Gerstner says no one is better at turning electrons into tokens than Elon, and expects more data centers from him, first on Earth and eventually in space. He suggests that this compute angle changes the way investors may think about a future SpaceX IPO.

In portfolio terms, Gerstner says Altimeter has been heavily concentrated in AI and compute across public and private markets. He highlights major bets in OpenAI and Anthropic, as well as exposure to semiconductors, memory, data centers and AI-adjacent defense modernization. He also argues that some public AI winners still trade at surprisingly low multiples relative to their growth.

The conversation closes with a different but related form of long-term investing: Trump accounts. Gerstner describes the program as a way to give American children a funded account invested in the S&P 500, with contributions from government, philanthropy, states, businesses and individuals. His broader argument is that ownership and compounding can move children from having no capital to having a first stake in the economy.

Source