Nvidia and Google Race to Subsidize Compute Power for AI Startups

In a strategic move to solidify its ecosystem, Nvidia is launching a new partnership program designed to provide fast-growing AI startups with easier access to its hardware through a network of cloud service providers. This initiative mirrors aggressive moves by competitors like Google Cloud, which has begun offering startups up to $500,000 in cloud credits and early access to its Gemini models. By lowering the barrier to entry for compute power, these tech giants hope to lock in the next generation of AI companies as long-term customers.

The competition for startup loyalty is increasingly characterized by 'compute-as-currency.' High-performance GPUs have become the most sought-after asset in Silicon Valley, and by facilitating revenue-sharing deals and providing free computing power, established giants are effectively subsidizing the development of the broader AI economy. This strategy aims to create a lasting stream of software and service revenue once these startups scale.

Simultaneously, the physical infrastructure supporting these companies is seeing unprecedented investments. TeraWulf recently signed a landmark $19 billion lease with Anthropic to build a massive AI infrastructure campus in Kentucky. This project highlights the shift toward dedicated, large-scale compute environments designed specifically for generative AI workloads. As startups gain access to these resources through Nvidia and Google's programs, the demand for such specialized campuses is expected to grow, further integrating energy providers and infrastructure firms into the tech ecosystem.