NVIDIA Opens AI Factory Compute to Capital Partners Via DSX Revenue-Share Model
NVIDIA introduced a new business model unlocking AI compute access for startups, model builders, and regional AI players through capital-partner-owned DSX (NVIDIA Data Super eXchange) AI factories. Sharon AI is deploying up to 40,000 Grace Blackwell GB300 GPUs, while Firmus is building a campus in Batam, Indonesia with 360 megawatts and up to 170,000 GPUs—both operating under a revenue-sharing and credit-support arrangement with NVIDIA.
The move addresses a bottleneck: emerging AI companies historically lacked capital to finance large-scale inference infrastructure, even with multi-year commitments. Through DSX, AI clouds can acquire NVIDIA accelerators and scale without waiting for site selection, power procurement, construction, or hardware bringup—crucial for inference providers like Baseten, Fireworks AI, and Together AI.
Architects shipping inference-heavy AI services gain faster deployment pathways to full-stack accelerated compute. NVIDIA secures recurring, usage-linked revenue; capital partners (Sharon, Firmus) get a path to scale without owning fabs. The token manufacturing narrative has NVIDIA pivoting from raw silicon sales to outcome-aligned infrastructure partnerships.
Sources
- Primary source
- NVIDIA Unlocks AI Compute at Scale
“Sharon AI is deploying up to 40,000 NVIDIA Grace Blackwell GB300 GPUs. Firmus is building a DSX AI factory campus in Batam, Indonesia. The campus is expected to scale to 360 megawatts and up to 170,000 NVIDIA GPUs.”
- NVIDIA Blog
“AI cloud companies will sell cloud services delivered through NVIDIA DSX AI factories that manufacture tokens at scale.”