NVIDIA DGX Cloud
NVIDIA’s AI factory in the cloud.
Overview
DGX Cloud—How NVIDIA Builds AI
DGX™ Cloud is NVIDIA’s internal cloud environment for building and operating AI at scale to support NVIDIA’s most demanding internal AI use cases. It is used to develop open source frontier and foundational models, validate new system architectures, and run production AI workloads.
DGX Cloud is built on NVIDIA-accelerated infrastructure and runs across CSPs and NVIDIA Cloud Partners (NCPs). This ensures that the software and operational patterns proven in DGX Cloud are directly applicable across the ecosystem.
How It Works
What Is NVIDIA DGX Cloud?
DGX Cloud serves as NVIDIA’s AI proving ground. Operational challenges encountered at scale are solved inside DGX Cloud first, then transformed into repeatable software, architectures, and reference implementations.
The software, operational intelligence, and infrastructure patterns developed inside DGX Cloud are externalized as modular, open infrastructure components through NVIDIA DSX OS.
NVIDIA Nemotron™ is a family of open models trained and optimized on NVIDIA DGX Cloud, demonstrating the scale and reliability of NVIDIA’s AI factory, and supporting multi-node training across tens of thousands of GPUs in production.
Build and Operate Leading AI Cloud Factories
NVIDIA DSX OS is a portfolio of open source, modular, and composable-by-design software that helps partners build and operate AI factories at scale reliably, efficiently, and securely.
NVIDIA on DGX Cloud
NVIDIA AI Is Accelerated by NVIDIA DGX Cloud
NVIDIA’s mission-critical research and next-gen open models are built and accelerated by NVIDIA DGX Cloud.
News and Blogs
Explore the Latest NVIDIA AI Developments Driven by DGX Cloud Innovations
Next Steps
Measure What Matters With Performance Benchmarking
Optimize AI workload performance on any NVIDIA infrastructure with NVIDIA performance benchmarking.
Become An Exemplar Cloud
Achieve optimal AI workload performance per TCO in partnership with NVIDIA with data-driven validated benchmarks.
