What is rack-scale compute in this context?

Rack-scale compute refers to full server racks that include computing, networking, and cooling in an integrated system. Cisco's expansion brings liquid-cooled systems starting with Blackwell and moving to Vera Rubin, partnered with Supermicro.

What does Cisco Validated Designs do?

Cisco Validated Designs comply with the Nvidia Cloud Partner reference architecture to reduce late-stage integration problems when customers bring complex AI systems into production.

How does Nvidia Spectrum-X allow flexibility?

The open interfaces in Nvidia Spectrum-X allow customers to run their own technologies on top of the platform, and Spectrum-X licensing lets Cisco Silicon One switches connect to the access network.

AI Business & IndustrySiliconANGLE AI1h agoAug 26, 2026, 01:01 JST2 min read

Cisco and Nvidia expand AI factory to rack scale

Send on LINE

Key takeaway

Cisco and Nvidia are expanding their AI factory offering to full rack-scale compute with liquid-cooled systems.
The goal is to help customers get AI systems into production faster.
The partnership includes Cisco Validated Designs and Nvidia Spectrum-X networking to ensure performance and flexibility.

3 Key Points

What happened
Cisco announced an expansion of its Secure AI Factory to include full rack-scale compute, partnering with Supermicro. The systems are liquid-cooled, starting with Blackwell and moving to Vera Rubin.
Why it matters
Enterprises, neoclouds, and sovereign programs are racing to get AI systems into production. Cisco and Nvidia aim to reduce late-stage integration problems with Cisco Validated Designs that comply with the Nvidia Cloud Partner reference architecture.
What to watch
The partnership emphasizes day-two operations—monitoring, health, availability, and software upgrades—plus open interfaces in Nvidia Spectrum-X that let customers run their own technologies while using Cisco Silicon One switches.

Ask the AI about this article →

Summaries like this, in your inbox every morning.

Get news like this every morning — free

Context & Analysis

The announcement reflects a broader shift from buying GPUs to deploying usable AI factories. Cisco and Nvidia are addressing the challenge of making AI systems work in production, not just lighting up clusters. The Cisco Secure AI Factory now includes rack-scale compute with Supermicro, and it emphasizes day-two operations like monitoring and upgrades, which many customers overlook.

The partnership builds on Cisco's networking heritage and Nvidia's reference architecture to create a unified deployment model. By integrating compute, networking, and lifecycle management, they aim to reduce the friction that occurs when separate teams handle servers and networking. The open interfaces in Spectrum-X give customers flexibility without forcing them to abandon familiar tools.

For enterprises and neoclouds, this could signal a move toward more turnkey AI infrastructure. The emphasis on speed and performance suggests that competitive advantage will increasingly depend on how quickly AI systems can be brought online and maintained, rather than just the raw hardware itself.

FAQ

What is rack-scale compute in this context?: Rack-scale compute refers to full server racks that include computing, networking, and cooling in an integrated system. Cisco's expansion brings liquid-cooled systems starting with Blackwell and moving to Vera Rubin, partnered with Supermicro.
What does Cisco Validated Designs do?: Cisco Validated Designs comply with the Nvidia Cloud Partner reference architecture to reduce late-stage integration problems when customers bring complex AI systems into production.
How does Nvidia Spectrum-X allow flexibility?: The open interfaces in Nvidia Spectrum-X allow customers to run their own technologies on top of the platform, and Spectrum-X licensing lets Cisco Silicon One switches connect to the access network.

SiliconANGLE AIRead Original Article