Cisco UCS and NVIDIA RTX PRO 6000 Server Edition: Powering the Next Wave of Enterprise AI

The AI infrastructure challenge

Your organization needs AI capabilities yesterday. Whether it's accelerating drug discovery, optimizing manufacturing processes, enhancing financial modeling or enhancing your workforce by leveraging the latest AI models, the pressure to deploy AI is real. But here's the challenge: Most enterprises struggle to find the blend of system components that can handle training, inference and visualization workloads aligned to their vision of an AI-accelerated business without requiring multiple siloed systems.

Enter the game-changing combination of NVIDIA RTX PRO 6000 Blackwell Server Edition and Cisco UCS compute for a new lineup of NVIDIA RTX PRO Servers from Cisco.

First to market is the Cisco UCS C845A M8, based on the NVIDIA MGX™ platform, which supports a two-by-eight-by-five system configuration with two CPUs, eight GPUs and five NICs in a four-rack unit footprint. This design gives enterprises maximum flexibility for a wide range of accelerated compute use cases, with up to eight NVIDIA RTX PRO 6000 Blackwell GPUs.

The NVIDIA RTX PRO 6000 Blackwell Server Edition will be available in early 2026 on the Cisco UCS C240 M8. The UCS X-Series will also support these GPUs in combination with the UCS X-Fabric and the X580p GPU nodes, extending GPU acceleration across Cisco's modular infrastructure.

The convergence of accelerated compute and visual computing

Modern AI workloads now span a wide range of operations, including model training, real-time inference, visualization and simulation. Each of these tasks places unique demands on infrastructure, requiring balanced CPU performance, high-speed GPU acceleration and unified management to maintain efficiency and scale.

Cisco's UCS portfolio, equipped with NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, is designed precisely for this convergence. These servers bring together powerful multi-core processors, advanced GPU acceleration and full lifecycle management through Cisco Intersight to create a seamless environment for AI development and deployment.

The result is an architecture capable of handling everything from edge inference and digital twin modeling to high-fidelity visualization and generative AI workloads. Cisco Intersight extends visibility across the stack, while Splunk Observability enhances monitoring and analytics to ensure consistent performance and operational insight across compute, network and storage layers.

This combination allows enterprises to consolidate previously separate systems for training, analytics and design into a single, cohesive platform that supports both accelerated compute and advanced visualization at scale.

What makes the NVIDIA RTX PRO 6000 Blackwell Server Edition different?

The NVIDIA RTX PRO 6000 Blackwell Server Edition GPU bridges professional visualization and AI acceleration in a single enterprise platform. It is purpose-built to address a wide range enterprise workloads, including  simulation, design and AI inference.

As the successor to the NVIDIA L40S GPU, the RTX PRO 6000 Blackwell introduces advanced multimodal AI inference capabilities with a fully integrated media pipeline that includes four NVENC encoders, four NVDEC decoders and NVJPEG acceleration. This enables exceptional efficiency in AI workloads involving image and video generation, molecular dynamics and simulation, delivering superior performance per dollar compared to NVIDIA HGX™ B200.

Beyond AI, the RTX PRO 6000 Blackwell excels in graphics-intensive applications such as video curation, rendering and virtual workstation environments. These capabilities make it ideal for enterprise users who need both AI acceleration and professional visualization performance within the same infrastructure, filling a gap that SXM-based NVIDIA HGX™ H200 and NVIDIA HGX™ B200 nodes are not designed to address.

When paired with Cisco's UCS servers, the NVIDIA RTX PRO 6000 Blackwell Server Edition provides a balanced, cost-efficient solution for enterprises. These RTX PRO Servers from Cisco offer a unified platform to consolidate AI inference, visualization and design workloads with Cisco Intersight-managed infrastructure.     

Cisco UCS C845A M8: Scalable AI and visualization power

The Cisco UCS C845A M8 is a 4U air-cooled rack server built for dense GPU computing. Powered by two AMD EPYC 9004 (Genoa) processors, it scales from two to eight double-width GPUs, including the NVIDIA RTX PRO 6000 Blackwell Server Edition, NVIDIA L40S and H-series accelerators.

The platforms also support five NICs typically deployed in a fashion to support GPU Clustered Networking at 400GB/s with NVIDIA ConnectX® and NVIDIA BlueField® DPU support. Hence, the reference as a 2-8-5 system. This modular design enables flexible GPU scaling while maintaining efficiency through advanced airflow and redundant power configurations.

Cisco UCS C240 M8: Edge and enterprise flexibility

The Cisco UCS C240 M8 is a 2U rack server powered by 6th-generation Intel XEON Scalable (Emerald Rapids) processors. It supports up to two double-width NVIDIA GPUs, such as the RTX PRO 6000 Blackwell (in Q1 2026) or H200 NVL (today), delivering a balance of compute and efficiency for distributed AI workloads, analytics and real-time visualization. With support for up to 28 NVMe, SAS or SATA drives, it enables high-speed data access for edge AI and vision-based workloads where GPU coordination is not required.

Cisco UCS X580p: Your future-proofed modular compute architecture

The Cisco UCS X580p PCIe Node is the newest expansion node in the Cisco UCS X-Series Modular System. It delivers advanced GPU acceleration for AI, visualization and compute-intensive workloads. Supporting up to four NVIDIA GPUs, including the NVIDIA RTX PRO 6000 Blackwell, NVIDIA H200 NVL and NVIDIA L40S, the X580p provides exceptional flexibility and performance for modern enterprise environments.

Cisco Secure AI Factory with NVIDIA and a validated design approach

The Cisco Secure AI Factory with NVIDIA infrastructure creates a standardized, secure framework for AI deployment, integrating Cisco UCS compute (including these new RTX PRO Servers) with NVIDIA AI Enterprise, NVIDIA BlueField, Cisco Nexus networking, and Cisco Secure Workload segmentation, management and observability through Cisco Intersight and Splunk.

These architectures are supported by Cisco Validated Designs (CVDs), developed in collaboration with ecosystem partners including NetApp, Pure Storage and VAST Data. Each CVD provides a prescriptive, tested architecture that shortens deployment time and reduces risk.

World Wide Technology (WWT) extends this value by allowing our clients to test CVD-aligned configurations directly in WWT's AI Proving Ground. Powered by our Advanced Technology Center (ATC), clients use the AI Proving Ground to validate designs like these against actual workloads, compliance requirements and integration needs before production deployment. We touch on these capabilities in the final section of this Research Note.

Industry impact and cross-domain innovation

The combination of Cisco UCS servers and NVIDIA RTX PRO 6000 Server Edition GPUs is transforming how enterprises deploy AI across industries while enabling horizontal innovation that cuts across business functions. For example:

• Healthcare: Researchers accelerate genomic sequencing and AI-assisted diagnostics, reducing analysis time from weeks to hours. NVIDIA RTX Servers from Cisco provide simultaneous compute for both research and clinical visualization.
• Manufacturing: Digital twins and predictive maintenance models accelerated by RTX PRO Servers from Cisco enhance production efficiency and reduce downtime. Real-time rendering allows engineers to visualize outcomes instantly.
• Financial services: Financial institutions leverage GPU-accelerated analytics to run complex simulations and fraud detection at scale, with unified observability through Intersight and Splunk.
• Energy and utilities: Energy providers accelerate seismic analysis and infrastructure monitoring while optimizing power distribution with real-time AI modeling.
• Higher education and research: Universities and research institutions deploy RTX PRO Servers from Cisco for robotics, machine learning and simulation, expanding access to AI innovation while simplifying operations through WWT's validated frameworks.
• Cross-industry innovation: NVIDIA RTX PRO Servers from Cisco also support horizontal applications like:
  • Agentic and physical AI for workforce and customer outcomes
  • High-fidelity 3D graphics and video, driving content creation, post-production and immersive VR/AR experiences
  • Excels at real-time computer vision, video analytics, image and video generation

Inside WWT's AI Proving Ground, these scenarios are validated under real workloads, letting clients see outcomes before committing to full-scale deployment.

Deployment considerations and use cases

Adopting GPU-rich infrastructure requires careful design across facilities, software and operations — areas where WWT provides hands-on assistance through our ATC and AI Proving Ground lab ecosystem.

• Power and cooling: A fully configured Cisco UCS C845A M8 can draw between six and seven kilowatts under load, requiring robust airflow, redundant power and optimized rack density. In WWT's ATC, engineers model real-world thermals, airflow and power draw to validate readiness before deployment, reducing data center integration risk.
• Network and topology: PCIe Gen5 delivers high throughput; but balanced placement of NICs, GPUs and DPUs is critical to achieving full performance. WWT network architects design and validate GPU fabrics that align with Cisco UCS and Nexus architectures to prevent data path bottlenecks.
• Software validation: Frameworks and drivers must be certified for new GPU architectures, such as NVIDIA Hopper™ and NVIDIA Blackwell. In the AI Proving Ground, WWT pre-validates NVIDIA AI Enterprise, CUDA frameworks, and container orchestration stacks to ensure optimized, production-ready performance.
• Observability: Visibility across compute, storage and networking is essential for managing large-scale AI operations. WWT integrates Cisco Intersight with Splunk Observability to deliver unified dashboards for GPU utilization, power efficiency and performance telemetry, giving IT and data science teams actionable insight from day one.

Aligning NVIDIA GPUs with Cisco Compute Platforms

The Cisco UCS portfolio is designed to align each NVIDIA GPU with the right compute platform for the workload. While the NVIDIA RTX PRO 6000 Blackwell Server Edition and NVIDIA L40S GPUs deliver exceptional PCIe-based performance in the C240 M8 and C845A M8 platforms, Cisco also provides SXM-based options for higher-density, ultra-scale AI workloads. 

The UCS C880A M8 and C885A M8, built on NVIDIA HGX™ architecture, are designed to support current-generation NVIDIA H100 GPUs as well as future NVIDIA H200 and NVIDIA Blackwell GPUs. This alignment ensures clients can scale seamlessly from visualization and inference to full-scale AI training and simulation within a consistent, Intersight-managed infrastructure.

Building the foundation for secure, scalable AI

The Cisco UCS, combined with NVIDIA RTX PRO 6000 Blackwell Server Edition, forms the compute foundation of Cisco AI PODs and the Cisco Secure AI Factory with NVIDIA.

Together, they deliver performance, scalability and security validated through WWT's ATC and AI Proving Ground. Clients can test configurations, optimize workloads and measure performance before purchase — a true "try before you buy" experience that accelerates AI adoption while reducing risk.

Cisco UCS continues to define the future of intelligent infrastructure, uniting compute, networking, security and observability with WWT's proven enablement expertise. This partnership empowers organizations to deploy AI confidently, with validated storage integration from NetApp, Pure Storage and VAST Data, and offers a clear path toward the AI-driven enterprise.

Experience the difference yourself

Every AI initiative begins with an idea, but true value comes from turning that idea into measurable outcomes. WWT bridges that gap through hands-on design, testing and deployment expertise across Cisco's AI infrastructure portfolio.

In our ATC and AI Proving Ground, clients can move beyond theory and test their real workloads under realistic conditions. Within these labs, you can:

• Test your workload patterns on fully configured Cisco UCS C845A systems
• Compare performance across different NVIDIA GPU configurations, including RTX PRO 6000 Blackwell and data center-class accelerators
• Validate architecture and system integrations with both existing and emerging tools, frameworks and observability platforms

WWT's approach extends far beyond lab validation. With a global team of more than 4,000 engineers, we help clients across industries at every stage of their AI journey. We can help you:

• Design the optimal configuration and architecture for your specific workloads and operational goals
• Validate performance using your actual applications in the ATC or through remote access
• Deploy at scale through WWT's Global Integration Centers, ensuring seamless implementation and logistics worldwide
• Learn to manage infrastructure efficiently with provisioning and lifecycle management through Cisco Intersight
• Optimize and secure ongoing operations using built-in observability and security integrations aligned to Cisco Secure AI Factory with NVIDIA

From idea to outcome, we help organizations like yours accelerate AI adoption with confidence. Explore, validate and deploy your next-generation AI infrastructure in partnership with Cisco and NVIDIA, powered by real engineering, real infrastructure and real results.