CoreWeave (CRWV) Q2 2026 Earnings Call: Q&A Summary

Q: Samik Chatterjee with JPMorgan asked about customer intent for contract periods and the magnitude of the renewal opportunity.

A: Mike Intrator noted that older infrastructure remains valuable for many use cases, and Nitin Agrawal added that only a limited part of the fleet is up for renewal, with managed inference providing a fast-growing alternative.

Q: Samik Chatterjee with JPMorgan asked how the company is securing supply chain capacity beyond the Solidigm agreement.

A: Mike Intrator explained that the company aggressively manages a complex supply chain through long-standing relationships with ODMs, OEMs, and NVIDIA, while Nitin Agrawal noted that margin expansion is outpacing input cost increases.

Q: Brad Zelnick with Deutsche Bank asked about initial learnings from the managed inference offering and the strategy for allocating capacity to it.

A: Mike Intrator stated that managed inference grew from $1 million to $100 million in one quarter, allowing the company to monetize older GPUs, while Nitin Agrawal noted that capital markets are supportive of underwriting shorter-duration contracts.

Q: Brad Zelnick with Deutsche Bank asked to unpack the drivers behind the 5% to 10% margin improvement on recent deals.

A: Mike Intrator attributed the improvement to a combination of platform value, security, TCO, and the fact that customers are increasingly willing to pay a premium as they monetize their own products.

Q: Irvin Liu with Evercore ISI asked if the $18 billion to $19 billion ARR exit target for 2027 remains valid given the pricing environment.

A: Nitin Agrawal clarified that the company raised its 2026 exit ARR guidance to $18.5 billion to $19.5 billion, which is already reflected in the current outlook.

Q: Irvin Liu with Evercore ISI asked about the impact of data center regulatory opposition on the goal of reaching 8 gigawatts of active power by 2030.

A: Mike Intrator stated that moratoriums impact location rather than demand and emphasized the need for collaboration with local communities through grid upgrades and job creation. Nitin Agrawal added that the company has already contracted 4.2 gigawatts of power.

Q: Raimo Lenschow with Barclays asked if the fleet needs to evolve to support the increased CPU and storage requirements of inference.

A: Mike Intrator explained that the company builds AI infrastructure designed to handle the full life cycle, allowing it to move seamlessly between training and inference workloads.

Q: Raimo Lenschow with Barclays asked how the company views the evolution of the market between edge, smaller clouds, and hyperscalers.

A: Mike Intrator stated that the cloud is built to serve both edge and non-latency-sensitive workloads, and Nitin Agrawal noted that demand, pricing, and margins are expanding despite increased competition.

Q: Michael Turrin with Wells Fargo Securities asked about the linearity of capacity additions and early market signals for Vera Rubin monetization.

A: Nitin Agrawal noted that 300 megawatts of power were added in June, which is back-end loaded for Q3 and Q4. Mike Intrator added that the Vera Rubin platform is seeing margin expansion from the start.

Q: Brett Knoblauch with Cantor Fitzgerald asked about the cadence of recontracting older GPU fleets versus placing them into spot or inference products.

A: Mike Intrator explained that the company evaluates whether to place infrastructure back into long-term contracts or shorter-term agreements to extract additional margin based on market dynamics.

Q: Brett Knoblauch with Cantor Fitzgerald asked how the success of the DDTL 5.5 financing changes the view on target contract durations.

A: Mike Intrator stated that the financing allows the company to diversify terms and enter new markets by offering 2-year or 3-year contracts that are more attractive to enterprise clients.