Nvidia unveils x86 server for AI and enterprise workloads
Nvidia introduces a server design combining x86 processors and GPUs to run agentic AI alongside traditional enterprise applications.
Computex 2024 – Nvidia has announced a new server design, the RTX PRO, aimed at bridging the gap between traditional enterprise workloads and cutting-edge agentic AI. The server combines x86 processors with eight GPUs, connected by a high-speed internal switch, to handle both AI and mainstream applications seamlessly.
Key Features of the RTX PRO
- GPU Powerhouse: Equipped with eight NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, offering 24,064 CUDA cores, 752 fifth-gen Tensor cores, and 188 fourth-gen RT cores, delivering 120 TFLOPS at FP32 precision.
- High-Speed Interconnect: An internal switch enables 800 Gbps GPU-to-GPU communication, though Nvidia has yet to disclose full details.
- Compatibility: Supports Kubernetes, hypervisors from Broadcom, Nutanix, and Red Hat, and is designed to run ERP and desktop virtualization workloads without requiring new management tools.
Nvidia’s Vision for AI and Enterprise
CEO Jen-Hsun Huang emphasized that while general-purpose IT infrastructure is ill-suited for the AI era, many enterprises still rely on familiar systems. The RTX PRO aims to address this by integrating AI capabilities into existing workflows. Huang also highlighted Nvidia’s AI data platform, which optimizes data flow from external storage to GPUs, claiming it revolutionizes compute, storage, and networking for AI.
Agentic AI and Robotics
Huang expressed enthusiasm for agentic AI, revealing that Nvidia’s software engineers already use such tools. He likened Nvidia’s role to an "HR department for AI agents," ensuring their responsible deployment. Additionally, he teased the upcoming Blackwell 300 series accelerators, set to debut in Q3, which will power 20-petaFLOPS desktops.
On the robotics front, Huang argued that humanoid robots could unlock a multi-trillion-dollar market by being versatile enough for both industrial and domestic use. Nvidia is actively supporting this development.
Availability
Nvidia’s partners plan to release the RTX PRO servers starting in July 2024.
Related Links
Related News
AWS extends Bedrock AgentCore Gateway to unify MCP servers for AI agents
AWS announces expanded Amazon Bedrock AgentCore Gateway support for MCP servers, enabling centralized management of AI agent tools across organizations.
CEOs Must Prioritize AI Investment Amid Rapid Change
Forward-thinking CEOs are focusing on AI investment, agile operations, and strategic growth to navigate disruption and lead competitively.
About the Author
Michael Rodriguez
AI Technology Journalist
Veteran technology journalist with 12 years of focus on AI industry reporting. Former AI section editor at TechCrunch, now freelance writer contributing in-depth AI industry analysis to renowned media outlets like Wired and The Verge. Has keen insights into AI startups and emerging technology trends.