AI is changing the power conversation. Are your customers ready?
Sep 23, 2026
By Joe Skorjanec, Product Line Manager, DC Power
The conversation around AI infrastructure is moving fast.
Organizations are exploring new ways to support AI workloads, increase compute capacity and prepare for higher-density deployments. For IT channel partners, this creates significant opportunities to expand conversations beyond servers and GPUs to power architecture, rack density and future-ready infrastructure.
Eaton’s OCP solutions portfolio includes the 50V DC power shelf that provides up to 33 kW of DC power in 1RU and scales up to 198 kW per rack, depending on busbar capacity.
Why power is part of every AI conversation
For years, data centers operated comfortably within traditional rack power requirements. AI changes that equation. Today, AI and high-performance computing (HPC) deployments are driving demand for:
- Higher rack densities
- Increased power capacity
- Liquid cooling readiness
- OCP-aligned infrastructure
- More flexible deployment models
As customers plan for these workloads, power delivery is becoming just as important as compute performance. Partners can help customers develop a broader infrastructure strategy that supports continued AI growth.
DC power is generating new interest
While traditional AC infrastructure continues to play a critical role across enterprise environments, many organizations are evaluating rack-level DC solutions for higher-density compute deployments. The key point to understand is this: DC isn't a replacement strategy. It's a scalability strategy.
Modern DC power solutions can help organizations support increasing rack densities while creating an infrastructure path that can evolve alongside future AI requirements. When discussing DC power with customers, focus on business outcomes rather than technical debate.
The most useful conversation isn't: "Should we move to DC?" It is: "How do we support the next generation of AI workloads?"
Advanced cooling for next-generation performance
As your customers evaluate higher-density architectures, it becomes apparent power and cooling are inseparable conversations.
For partners, liquid cooling presents another opportunity to help customers build complete and integrated infrastructure. A coolant distribution unit (CDU), together with supporting components such as manifolds and cold plates, helps deliver the thermal performance required for high-density compute environments. Have the conversation about these advanced cooling strategies alongside discussions about rack design, power delivery and future capacity planning.
Start with one rack, not a whole data center
One of the biggest misconceptions surrounding DC and liquid cooling adoption is that a complete facility redesign is required. Most organizations start much smaller with:
- A dedicated AI rack
- A proof-of-concept environment
- A high-density pod
- A new application cluster
This incremental approach allows customers to introduce DC where it provides value while continuing to operate AC infrastructure elsewhere.
Key takeaway: the future isn't AC or DC. It's AC and DC working together.
Partners who can explain how both architectures can coexist are often better positioned to guide customers through long-term infrastructure planning discussions.
Understanding ORV3 and Open Compute trends
As AI projects become more power-intensive, customers are becoming familiar with Open Compute Project (OCP) standards and Open Rack Version 3 (ORV3). For partners, ORV3 represents another opportunity to help customers navigate industry changes.
In typical rack-level deployments, power shelves distribute 50V DC across a common busbar, enabling power delivery to multiple servers without individual AC power supplies. This simplifies the power architecture, reduces cabling requirements, streamlines service, improves efficiency and enables higher rack densities.
Two practical points can help simplify the conversation during planning and evaluation discussions:
- Converting to 21-inch racks doesn't mean existing equipment becomes obsolete. Many customers assume ORV3 requires replacing everything they own. Not necessarily. The appropriate mounting rail kits allow traditional 19-inch EIA equipment to operate within ORV3 racks, simplifying migration and protecting existing investments.
- OU vs. RU isn't as complicated as It sounds. Customers comparing rack capacities often encounter open units (OU) for the first time. Here’s a simple rule of thumb:
- 39OU is roughly equivalent in height to a traditional 42RU rack
- 44OU is roughly equivalent in height to a traditional 48RU rack
Preparing customers for what's next
The industry is discussing emerging architectures, including 800V DC, to support increasingly dense AI environments. Most customers don't need to solve for 800V today, but they can begin building familiarity with rack-level DC power, high-density deployments, OCP standards, modular power scaling and AI-ready infrastructure design.
Questions every partner should be asking
When discussing AI initiatives with customers, consider asking:
- Are rack power densities increasing?
- Are AI or HPC workloads being planned?
- Are dedicated AI zones or clusters under consideration?
- Is the organization evaluating OCP architectures?
- Are future power requirements becoming difficult to forecast?
Several “yes” answers may signal an opportunity to discuss future-ready power infrastructure.
Final takeaway
AI growth is creating new infrastructure challenges and new opportunities for channel partners.
The organizations seeing the most success are taking practical steps today that create flexibility for tomorrow. Partners can help their customers plan to support higher rack densities and explore DC-ready architectures and OCP standards, while building a flexible roadmap for the next generation of compute.
Want to learn more about the market forces driving this growth?
Read how Eaton is helping partners address expanding AI and white space opportunities.