Cisco expands Secure AI Factory with Supermicro deal
Fri, 4th Sep 2026 (Today)
Cisco has expanded its Secure AI Factory with Nvidia through a partnership with Supermicro, adding high-density, liquid-cooled compute to the companies' AI infrastructure offering.
The announcement brings Supermicro into an existing Cisco-Nvidia collaboration focused on integrated AI systems for large-scale deployments. The expanded arrangement is intended for the rack-scale phase of AI infrastructure, where operators manage larger clusters, higher power demands, and more complex networking requirements.
The companies are combining Cisco's networking and security products with Nvidia computing and software, alongside Supermicro server and system designs. The offering targets organisations building AI infrastructure across data centres, with an emphasis on bringing compute, networking, storage and security together in one stack.
Rack-scale AI systems have become a focal point for infrastructure suppliers as demand grows for training and inference on larger models. These deployments typically require denser server configurations, faster interconnects and more advanced cooling systems than conventional enterprise data centre workloads.
Liquid cooling is central to the latest expansion. Higher-density AI hardware consumes more power and generates more heat, prompting suppliers to redesign racks and data centre layouts around cooling efficiency as well as raw compute capacity.
The addition of Supermicro broadens the hardware options available within the Secure AI Factory framework. Supermicro is known for building servers and integrated systems used in AI and high-performance computing environments, giving the partnership a route into more tightly packed infrastructure designs.
The broader effort reflects a shift in the AI market from individual hardware components to bundled systems. Customers increasingly want validated combinations of servers, networking, accelerators, storage and management software rather than sourcing and integrating each layer separately.
Integrated stack
Cisco has been positioning its AI infrastructure strategy around demand for integrated systems. By tying networking and security more closely to AI compute platforms, the company is seeking a larger role in data centre projects often led by chipmakers, server makers and specialist cloud providers.
Nvidia, meanwhile, has expanded beyond accelerators into full system architectures, networking and software tools for AI model development and deployment. Its partnerships with server manufacturers and networking suppliers have become a core route to market as organisations seek pre-integrated designs that can be deployed at scale.
Supermicro's addition underlines how server makers are trying to capture a larger share of AI infrastructure spending by offering complete rack-level systems rather than standalone hardware. The company has been active in liquid-cooled and high-density server configurations, which are becoming increasingly relevant as AI clusters grow larger and more power-intensive.
Market pressure
The partnership also comes as enterprises and service providers face growing pressure over the economics of AI buildouts. The cost of power, floor space and cooling has become as significant as processor performance, especially for operators planning multi-rack or hall-scale installations.
Security remains another area of focus. AI deployments create additional concerns around data protection, model integrity and access control, particularly when infrastructure is shared across teams or used for sensitive workloads. Cisco's role in the partnership centres in part on addressing those issues alongside connectivity and systems integration.
The companies did not disclose financial terms. They described the expansion as part of a broader push to give customers a packaged route to deploying AI infrastructure in data centres that need both high-density compute and tighter operational control.
"Together, Cisco, NVIDIA and Supermicro are providing customers with a secure, scalable path for building AI factories in the rack-scale era," said Chuck Robbins, Chair and Chief Executive Officer, Cisco.
"By integrating liquid-cooled, high-density compute into Cisco's Secure AI Factory with NVIDIA, we are helping organisations simplify deployment while addressing the operational demands of next-generation AI infrastructure," said Robbins.
"As AI factories evolve to support ever-larger models and more demanding workloads, rack-scale systems become essential," said Jensen Huang, Founder and Chief Executive Officer, Nvidia.
"With Cisco and Supermicro, we can deliver an integrated infrastructure foundation that brings together computing, networking and software for this new class of AI deployment," said Huang.
"Customers are looking for proven, efficient designs as they build out AI infrastructure at increasing scale," said Charles Liang, President and Chief Executive Officer, Supermicro.
"Our liquid-cooled systems and rack-scale building blocks, combined with Cisco and NVIDIA technology, will help organisations deploy dense AI environments with greater efficiency," said Liang.