AMD is employing a financing strategy previously popularized by Nvidia, backing a $300 million loan to cloud startup Crusoe. The loan, provided by Goldman Sachs, will enable Crusoe to purchase AMD’s AI chips for use in a new datacenter located in Ohio. This marks the first known instance of AMD utilizing its chips as collateral for debt, mirroring a practice Nvidia has successfully implemented to bolster sales.
The arrangement effectively positions AMD as a guarantor for the loan. According to sources familiar with the deal, AMD has agreed to rent the chips back from Crusoe should the startup fail to secure alternative customers. This allows AMD to immediately recognize $300 million in AI chip sales, even before the chips are fully deployed and generating revenue for Crusoe. The specific AMD AI chip model being purchased wasn’t disclosed, though it is presumed to be a member of the Instinct MI450 series.
Crusoe distinguishes itself as a cloud provider focused on sustainability, claiming to prioritize data center locations with access to “clean, scalable power.” This emphasis on environmentally responsible infrastructure may have been a factor in securing the loan and partnership with AMD. The company’s commitment to cleaner energy aligns with a growing industry trend towards more sustainable computing practices.
This move by AMD comes as competition in the AI chip market intensifies, with Nvidia currently holding a dominant position. By facilitating financing for customers like Crusoe, AMD aims to lower the barrier to entry for adopting its AI hardware and accelerate deployment. This strategy allows AMD to showcase the capabilities of its chips and gain a foothold in the rapidly expanding AI infrastructure market.
AMD’s broader push into the AI space has been gaining momentum. Last year, OpenAI secured up to 6 gigawatts of AMD GPU compute power, potentially leading to a significant equity stake for the chipmaker. In December, AMD and HPE announced a collaboration to integrate AMD’s Helios rack-scale AI architecture into HPE’s product portfolio, further solidifying AMD’s position as a key player in the AI hardware ecosystem. The Helios architecture is slated for release in the second half of 2026, and AMD recently denied reports of delays to its next-generation MI455X accelerators.
Nvidia pioneered this approach to financing, notably using its H100 GPUs as collateral to support a $2.3 billion loan for CoreWeave, backed by Magnetar Capital and Blackstone. This practice allows Nvidia to effectively create demand for its products and accelerate revenue recognition. AMD’s adoption of a similar strategy signals its intent to compete more aggressively in the AI hardware market and leverage its chip sales to secure financing for its customers.
The use of chips as collateral represents a novel financial instrument in the semiconductor industry. While it benefits both the chipmaker and the customer in the short term, it also introduces a degree of circularity. The arrangement allows companies to report substantial sales figures without necessarily reflecting immediate end-user demand. This practice raises questions about the true extent of market adoption and the underlying health of the AI infrastructure market.
The $300 million loan and AMD’s backing of it underscore the growing importance of AI infrastructure and the intense competition among chipmakers to capture market share. As demand for AI computing power continues to surge, innovative financing strategies like this are likely to become more prevalent, enabling companies to overcome financial hurdles and accelerate the deployment of AI technologies. The success of this model for AMD will likely depend on Crusoe’s ability to attract customers and fully utilize the purchased AI chips.
