Krishna Rao

Chief Financial Officer of Anthropic, the AI safety company behind the Claude models, where he runs the compute procurement and capital formation that fund a business growing at exponential rates.

Rao joined Anthropic in early 2024, closing its Series D as the company acquired its first frontier model and FTX was liquidating a block of Anthropic shares. He has since overseen roughly $75bn of fundraising and an over-$100bn multi-year compute commitment across Amazon, Google, Broadcom, and Nvidia. Before Anthropic he helped lead Airbnb’s financing through the pandemic — when revenue fell 70% in seven weeks — and earlier worked in the private equity group at Blackstone.

Core positions

Rao treats compute as the master variable of a frontier lab: bought a year or more ahead against a ‘cone of uncertainty’, used fungibly across chip platforms, and judged by the return on the whole compute envelope rather than as a software-style variable cost. He argues the returns to frontier intelligence are exceptionally high in enterprise, that pricing should stay stable to let customers capture most of the value, and that spending on safety — interpretability and alignment — is not at odds with a large business but reinforces it, because enterprises entrust their most sensitive workflows only to a lab they trust.

In the wiki

• Krishna Rao on Anthropic's Compute Economics, the $100 Billion Commitment, and Financing Frontier AI — compute economics, fungible multi-platform orchestration, the $100bn commitment, and return on compute.