- Supermicro expands its Open Compute Project portfolio with integrated AI racks, networking, power and cooling
- Standardized 21-inch rack designs aim to simplify deployment of increasingly power-hungry AI systems
- The move reflects growing demand for preintegrated infrastructure as server manufacturers compete to deliver complete AI clusters
Supermicro is expanding its portfolio of open-standard AI infrastructure, betting that preintegrated racks, cooling systems and networking will help data center operators bring increasingly complex AI installations online faster.
The company announced the expansion at the Open Compute Project (OCP) Global Summit in San Jose this week, adding OCP-accepted systems, racks and rear-door heat exchangers to its Data Center Building Block Solutions (DCBBS) portfolio.
The objective is to deliver more of the infrastructure required for AI computing as a validated system, rather than leaving customers to integrate servers, networking, power distribution and cooling separately.
That’s an increasingly important proposition as AI infrastructure moves beyond individual servers toward entire racks and clusters, with growing demands on power delivery and heat removal.
Supermicro’s expanded portfolio includes OCP Open Rack version 3 (ORv3) and Open Rack Wide (ORW) designs, centralized 48-volt DC power distribution, direct-to-chip liquid cooling, rear-door heat exchangers and management software.
The ORv3 architecture uses wider, 21-inch equipment bays rather than the conventional 19-inch mounting format common in enterprise data centers. It also replaces individual server power supplies and associated cabling with centralized rack-level power distribution.
Supermicro is supporting both open-rack and conventional EIA rack designs, allowing customers to select architectures appropriate to their existing facilities.
Cooling becomes part of the system
Power and cooling are central to the announcement.
As GPUs and CPUs become more powerful, the amount of heat generated by densely packed computing equipment is becoming a major constraint on AI data center design.
Supermicro’s expanded portfolio includes direct-to-chip liquid cooling and rear-door heat exchangers designed to remove residual heat from server exhaust before it enters the data hall.
The company said its rear-door heat exchangers provide cooling capacities ranging from 10 kW to 120 kW, while a separate in-rack coolant distribution unit supports up to 250 kW of cooling capacity.
The approach could help operators increase rack density without relying entirely on conventional room-level air conditioning, although actual performance will depend on the configuration and the facility’s cooling infrastructure.
Emilio Guerrero, a Supermicro product manager speaking at the OCP event, described the shift toward more efficient power delivery and cooling.
“It is a new way to deliver power more efficiently to the servers,” Guerrero said, adding that the design could support 97% heat capture.
From server supplier to infrastructure integrator
The announcement also reflects a broader change in the competitive position of server manufacturers.
Increasingly, customers are buying complete, prevalidated computing environments rather than assembling individual components themselves.
ABI Research identified this shift in its assessment of leading AI server suppliers, naming Supermicro, Dell, HPE, Lenovo and Cisco as its five highest-ranked vendors.
The research firm highlighted the growing importance of modularity, integrated cooling and deployment services alongside conventional server performance.
It also identified openness as an increasingly important consideration, with organizations such as OCP helping establish hardware standards that can simplify upgrades and reduce dependence on proprietary designs.
For Supermicro, supporting OCP specifications offers a way to combine hardware from multiple semiconductor suppliers within standardized rack-scale architectures.
The company’s portfolio includes systems built around processors and accelerators from Nvidia, AMD and Intel, alongside high-performance Ethernet networking and liquid-cooling infrastructure.
But open hardware standards do not necessarily eliminate vendor dependency. Operators must still consider compatibility across management software, networking, accelerators and cooling systems when selecting integrated platforms.
Supermicro is also expanding its management software portfolio, including tools for monitoring and automating infrastructure across AI factories, enterprise data centers and cloud environments.
That creates an interesting competitive tension: customers want the flexibility associated with open standards, while suppliers increasingly differentiate themselves through their ability to integrate and manage the complete infrastructure stack.
For data center operators, the attraction is straightforward. The faster a validated AI cluster can be installed, powered, cooled and connected, the sooner it can begin delivering computing capacity.
Whether standardized rack-scale infrastructure can materially reduce deployment times and costs across different facilities remains an important question. Supermicro has outlined the engineering approach but has not provided comparative deployment data in its announcement.
For an industry pouring enormous amounts of capital into AI computing, that distinction is worth watching.
Learn more about data center architecture
Meta exec blows the OCP audience's mind with its data center innovations
Cisco and Nvidia develop new pluggable optic module for AI data centers
Nvidia’s SVP of networking explains the 5 types of networks needed for AI factories Supermicro expands integrated OCP portfolio — company announcement
ABI Research: Top Five AI Server Companies for Data Centers and Enterprises
ABI Research: AI Server Market Size, Vendor Shares and Investment Drivers