Michael IntratorCo-Founder, Chairman, CEO & President, CoreWeave, Inc. (CRWV)
10 recorded appearances
Intrator argued CoreWeave's near-term capacity is effectively sold out and that older GPU generations will stay economically useful far longer than the market assumes, pushing back on depreciation-based bear cases.
Speaking to Caroline Hyde after Q1, Intrator called the quarter transformational and defended the heavy capex cycle as demand-led rather than speculative.
He reaffirmed 2026 revenue and operating-margin guidance and framed compute demand as the binding constraint, not customer appetite.
Recorded live at Nvidia GTC; Intrator's segment (0:37-32:58) covers how CoreWeave contracts capacity, why it builds rather than rents, and how it underwrites counterparty risk on multi-year AI deals.
A long-form sit-down on the Nvidia relationship, how CoreWeave differs from hyperscaler clouds, and the financing structure behind its data-center buildout.
At Davos he argued companies should change internally to insulate themselves from geopolitical turbulence, and described power and supply chain as the real limits on AI infrastructure.
Intrator made the long-duration case for AI infrastructure returns and discussed what AI ROI actually looks like for enterprise buyers.
Intrator and CSO Brian Venturo took the bear case head on: what happens to CoreWeave if AI demand flattens, why the company uses debt financing, and how it builds data centers so quickly.
He addressed a third-party data-center delivery delay that hit the quarter and explained why contracted backlog, not quarterly revenue timing, is the right lens on the business.
On stage with Bloomberg's Tom Giles, Intrator laid out how purpose-built AI cloud infrastructure differs architecturally from general-purpose cloud and why that gap is widening.