The announcement reflects NVIDIA’s view that AI workloads are increasingly shifting from model development to production inference, where continuously operating AI infrastructure is needed to generate tokens at scale. According to the company, that transition requires large, multi-tenant computing environments that can be deployed quickly, remain highly utilized and support the economics of AI services running in production.
NVIDIA said the new structure is designed to address a longstanding challenge for AI companies that need significant computing resources but have struggled to finance capital-intensive infrastructure, even when backed by long-term customer commitments. By combining revenue sharing with credit support, the company aims to accelerate deployment of AI infrastructure while giving customers faster access to full-stack accelerated computing.
The first deployments under the program are already underway through AI cloud providers Sharon AI and Firmus, which are building NVIDIA DSX AI factories to serve customers across multiple regions.
Sharon AI plans to deploy up to 40,000 NVIDIA Grace Blackwell GB300 GPUs.
“This strategic collaboration with NVIDIA marks a pivotal moment in Sharon AI’s mission to deliver sovereign, large-scale AI compute infrastructure,” said James Manning, cofounder and CEO of Sharon AI.
Firmus is developing a DSX AI factory campus in Batam, Indonesia, with plans to scale the site to 360 megawatts and up to 170,000 NVIDIA GPUs.
“AI-native companies need access to scalable, energy- and cost-efficient compute infrastructure to compete globally,” said Tim Rosenfield, co-CEO of Firmus Technologies. “Firmus AI cloud is building a NVIDIA DSX-aligned AI factory, which will enable our cloud to help more customers access the compute they need to build and scale AI.”
NVIDIA said the program is aimed at organizations building and operating AI services, including model builders, inference providers, agent platforms and enterprises that require large-scale computing for model training, post-training, fine-tuning and high-volume inference. The company cited Baseten, Fireworks AI and Together AI as examples of AI-native businesses whose demand reflects the need for immediate access to cloud-based AI capacity.
Rather than relying solely on conventional GPU sales, the initiative ties part of NVIDIA’s financial return to how heavily deployed infrastructure is used. The company said the model is intended to align its economics with AI cloud operators while expanding access to accelerated computing for organizations moving AI applications from development into production.
This analysis is based on reporting from Nvidia.
Image courtesy of Nvidia.
This article was generated with AI assistance and reviewed for accuracy and quality.
