**[NVIDIA Launches "AI Compute Partnership Program": Offering Compute Power Guarantees in Exchange for Cloud Providers' Revenue Shares and Equity]** BlockBeats News, July 2 – NVIDIA is advancing a new initiative called the "AI Compute Partnership Program." According to participating cloud service providers Firmus and Sharon AI, NVIDIA has committed to providing compute capacity guarantees for these startup cloud providers: if the GPUs they purchase cannot be rented out, NVIDIA will pay to lease back the unused compute power. In return, NVIDIA will take a portion of these startups' cloud service revenue (with the revenue share percentage gradually decreasing over the contract term) or receive stock subscription rights. Due to the high costs of building data centers and purchasing GPUs, startup cloud providers with lower credit ratings often struggle to secure financing. NVIDIA's capacity swap guarantee effectively serves as a critical credit endorsement, helping these companies obtain loans more easily. Currently, major players like Amazon, Microsoft, SpaceX, Oracle, Meta, and Google have purchased most of NVIDIA's chips, but these giants are also developing their own chips. NVIDIA aims to reduce its reliance on large clients by supporting startup cloud providers. As early as September 2025, NVIDIA had pledged to guarantee $6.3 billion worth of unsold compute capacity for CoreWeave through 2032. Additionally, NVIDIA is reportedly in talks to provide similar financial guarantees for a $500 billion large-scale data center proposed by OpenAI. [Original Article Link]