Jensen Huang’s departure from Nvidia has added uncertainty to bitcoin miners’ plans to spend billions of dollars on AI infrastructure. Investors are watching not only whether demand for AI computing will hold up, but also whether large contracts can generate enough cash to cover construction costs and debt payments. The shift also exposes miners to building, financing and customer-credit risks, beyond the volatility of bitcoin prices.
Nvidia’s links to miners’ AI financing
Nvidia plays several roles in the expansion of AI infrastructure: it is a supplier, a customer and, in some cases, a source of financial support. In May, IREN signed a five-year cloud-services contract worth $3.4 billion. In January, Nvidia invested $2 billion in CoreWeave. These ties mean miners’ AI plans depend not only on their own construction schedules, but also on Nvidia’s and cloud customers’ investment and purchasing decisions.
IREN and Applied Digital shares fell in the AI-stock sell-off on Nov. 12, following news of Huang’s departure. For companies that need to raise money by issuing shares, a lower stock price can mean selling more shares to secure the same amount of capital. Erica Spear, a credit analyst at JPMorgan, said new deals could take longer to negotiate or require greater compensation for investors.
If Nvidia slows its investments, guarantees or purchases of computing capacity, miners could face higher financing costs and construction delays. Demand for additional computing capacity could also weaken if cloud-service tenants come under financial pressure.
Project readiness and business models vary
There is no evidence so far that Nvidia has broadly withdrawn its financial support. The company’s board said operations, customer commitments and product road maps are expected to continue without interruption. For now, attention is on the terms of future deals and project financing, rather than any confirmed halt to business.
Miners are pursuing different approaches to AI. Leasing sites with grid connections, operating data-center colocation facilities, and buying GPUs to sell computing services each involve different upfront costs and operating requirements, as well as different exposure to equipment replacement.
Projects with secured funding, available power and binding payment commitments may perform very differently from expansion plans that still depend on future financing and regulatory approvals. For investors, the central questions are whether contract cash flows can support construction costs and debt, and whether projects can be completed on schedule.