The CoreWeave CEO discusses challenges in the complex supply chain and data center construction, emphasizing the company’s diversification strategy and internal capabilities to mitigate delays and risks. He highlights a long-term Nvidia contract supporting flexible compute capacity for innovation, while noting persistent high demand and supply shortages for GPUs critical to AI development.
In the discussion, the CoreWeave CEO addresses the challenges of not owning the entire supply chain, emphasizing that no company controls every aspect of such a complex ecosystem. From chip design and fabrication to data center construction, many interdependent partners are involved. Despite occasional setbacks with partners, such as delays at the power supply level, these relationships are longstanding and collaborative. The CEO highlights that delays typically result in postponed revenue rather than lost revenue, reflecting the nature of working within a broad ecosystem.
Regarding the impact of recent delays, the CEO refrains from naming specific customers but notes that the affected project is large and involves significant counterparts. The company is actively managing third-party risks by diversifying its client base and suppliers. Over the past year, CoreWeave has reduced reliance on any single client from 85% of revenue to less than 35%, and no single data center provider accounts for more than 20% of their planned 2.9 gigawatts of power delivery over the next two years. This diversification strategy aims to mitigate risks associated with supply chain constraints and ensure business resilience.
To prevent future delays, CoreWeave has developed an internal team dedicated to building and delivering data centers, exemplified by projects in New Jersey and Pennsylvania. This self-build capability allows the company to deploy skilled personnel to assist third-party data center construction when issues arise. Additionally, the company has diversified its data center providers to avoid overdependence on any single entity. The CEO believes that as CoreWeave scales, the relative impact of delays at individual sites diminishes, and that time and growth will help resolve current scaling challenges.
The CEO also discusses the role of government in alleviating supply chain and infrastructure challenges, particularly in expediting permitting and grid connection processes. He advocates for government involvement to facilitate faster infrastructure deployment, which is crucial for the industry’s growth. Furthermore, CoreWeave has a notable contract with Nvidia that spans six years, allowing flexibility to repurpose compute capacity to startups and smaller companies when spare capacity exists. This arrangement supports innovation by providing emerging companies access to critical computing resources.
Finally, the CEO reaffirms Nvidia’s GPUs as the gold standard for CoreWeave’s customers, noting that demand for these chips far exceeds the market’s capacity to supply them. Clients approach CoreWeave with specific infrastructure needs, and the company works closely with them to deliver tailored solutions. The systemic shortage of GPUs and computing infrastructure for AI development remains a significant challenge, and CoreWeave expects this demand to continue outstripping supply for the foreseeable future.