Dell has updated its enterprise computing lineup by introducing the Pro Precision with GB300, a workstation running on Windows with NVIDIA Grace Blackwell hardware. The system delivers up to 20 petaflops of FP4 processing power and 748GB of coherent memory to run 1 trillion parameter AI models locally. By keeping the machine on Windows, Dell eliminates the need for IT teams to deploy separate Linux setups for enterprise AI tasks.
Dell unified its artificial intelligence hardware by renaming the former Pro Max line to Dell Pro Precision. Both the GB10 and GB300 models now sit under this single brand. Historically, companies maintained Linux systems for machine learning alongside Windows for general engineering and office software. The Pro Precision with GB300 changes this approach by running directly on Windows. Corporate IT teams can deploy the hardware using their existing security policies and device management tools without operating system changes.
The system runs on the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip. This processor configuration produces 20 petaflops of FP4 compute and provides 748GB of coherent memory, letting companies run large models on local hardware rather than paying for cloud compute fees. To handle thermal output, Dell added its proprietary MaxCool system, which increases cooling efficiency by up to 5 times. Software engineers can still work within the CUDA environment through the Windows Subsystem for Linux without requiring dual booting. With pricing still undisclosed.
