Cisco introduced substantial updates to its AI infrastructure portfolio at the Cisco Partner Summit in Los Angeles, unveiling a dedicated family of AI servers and AI PODs designed to accelerate and simplify AI adoption for enterprise customers. The newly launched UCS C885A M8 servers are built specifically for GPU-intensive AI workloads and harness NVIDIA’s HGX supercomputing platform, integrating H100 and H200 Tensor Core GPUs to deliver high-performance AI computing. These eight-way accelerated servers represent Cisco’s first dedicated foray into the AI server market and are equipped with NVIDIA NICs or SuperNICs to optimize AI networking and bolster security through NVIDIA BlueField-3 DPUs. Meanwhile, Cisco’s AI PODs provide modular, use-case-specific stacks of compute, networking, storage, and cloud management, allowing businesses to deploy AI solutions with greater speed and reduced complexity.
The increasing volume of AI workloads across industries has amplified demand for secure, scalable, and programmable infrastructure capable of supporting advanced AI applications. Cisco’s latest offerings are designed to bridge this gap, providing modular, adaptable options for enterprises whether they’re beginning AI integration or expanding existing capabilities. Managed through Cisco Intersight, both the UCS servers and AI PODs bring centralized control and automation, simplifying everything from initial configuration to daily operations, making AI adoption smoother. According to Cisco’s AI Readiness Index, 89% of enterprises plan to implement AI workloads within the next two years, though only 14% currently have infrastructure that’s fully capable of handling these advanced tasks. Cisco aims to provide the technological foundation to support this rapid AI growth.
Cisco’s new infrastructure offerings come as part of a broader ecosystem strategy, collaborating closely with technology partners to facilitate innovation in the enterprise AI space. The new systems integrate seamlessly with Cisco’s Nexus switching platforms, including the 800G-capable models powered by the Cisco Silicon One G200 chip, which ensure robust data transfer speeds and reliability for demanding AI applications. Both the UCS C885A M8 servers and the AI PODs will be available for order in late 2024, with Cisco offering financing solutions to help customers manage payment risks and cash flow. Cisco also highlights partner solutions to enhance AI deployment and build out capabilities efficiently.
- AI Server Launch: Cisco has launched the UCS C885A M8 servers to meet growing demand for dedicated AI computing infrastructure. These servers are built specifically for GPU-intensive tasks, using NVIDIA’s HGX supercomputing platform, which incorporates powerful H100 and H200 Tensor Core GPUs. With support for up to eight GPUs, these servers handle even the most intensive AI training and inference workloads. Integrated NVIDIA NICs or SuperNICs enhance network performance, while NVIDIA BlueField-3 DPUs provide zero-trust security, ensuring data protection at every level. This addition marks Cisco’s first dedicated AI server portfolio, providing enterprises with the robust processing power required for modern AI applications.
- New AI PODs: Cisco’s AI PODs offer plug-and-play infrastructure stacks for specific AI use cases and industries, combining compute, networking, storage, and cloud management to create a complete AI solution. Built on Cisco Validated Designs (CVDs), these PODs allow organizations to begin with an established, industry-proven foundation. With customizable options, they are adaptable to a wide range of AI applications, enabling faster deployment and reduced complexity. The AI PODs integrate NVIDIA AI Enterprise, a comprehensive, cloud-native software platform designed to accelerate AI and streamline data science pipelines, resulting in consistent performance and faster time-to-value for enterprises.
- Centralized Management: Cisco Intersight manages the entire suite of new AI infrastructure products, delivering centralized control and automation that simplifies complex operations. From initial setup to ongoing management, Intersight provides the flexibility needed for high-performance environments, making it easier for IT teams to keep AI workloads running smoothly. The platform’s built-in automation and scalability features reduce operational complexity, enabling teams to focus more on strategic initiatives rather than troubleshooting infrastructure.
- Scalable Networking: Cisco’s new AI solutions integrate with its 800G Nexus switching platforms, which use the high-performance Cisco Silicon One G200 chip to ensure fast, reliable data transfers that can support demanding AI applications. These switches are designed for data-intensive environments and provide the scalable networking needed to support the increased data flow associated with enterprise AI. This new layer of networking infrastructure ensures that AI applications perform reliably, even as workloads grow in scale and complexity.
- Availability: Cisco’s UCS C885A M8 servers and AI PODs will be available for order in late 2024, with the AI PODs expected to ship in November. Cisco also offers financing options to make these solutions accessible, helping partners manage balance sheet risks and improve cash flow. Cisco’s financing approach is designed to make the transition to AI smoother for enterprises, supporting them in deploying advanced AI infrastructure without the financial strain of up-front costs.
“A new era of intelligent generative AI applications is driving transformation for enterprises worldwide. With the debut of its dedicated AI server portfolio, AI PODs, and expanding systems with full-stack NVIDIA acceleration, Cisco is advancing the infrastructure enterprises need to grow in the AI age,” stated Bob Pette, vice president, Enterprise Platforms at NVIDIA.