Delta highlights collaboration with NVIDIA on 800V DC power and cooling designs for AI data centers at OCP Global Summit 2026

Delta Electronics is using the OCP Global Summit 2026 to draw attention to how it is working with NVIDIA to develop energy-efficient 800V DC power and cooling solutions for AI data centers. In a report carried by Yahoo Finance, Delta said the effort is tied to NVIDIA’s 800V DC direction for power delivery and thermal management, where the goal is to reduce inefficiency in how electricity is converted, distributed, and cooled across high-density computing racks.

The collaboration centers on what the report describes as NVIDIA 800 VDC power and cooling solutions. While details of the underlying architecture were not laid out in the announcement, the emphasis reflects a broader industry shift toward higher-voltage data center power systems designed to improve efficiency at scale. Higher-voltage DC approaches can reduce current for a given power level, which can lower electrical losses in wiring and power conversion and, in turn, support more power-dense server deployments.

Delta also pointed to a specific reference in the report to “800 VDC” along with “TOU2VN,” suggesting Delta is aligning its own power and cooling offerings to NVIDIA-oriented system requirements. The company did not describe the meaning of TOU2VN in the Yahoo Finance write-up, nor did it provide performance figures or engineering specifications that would let readers compare the approach to alternative power and cooling designs.

OCP Global Summit is a major meeting point for operators and suppliers who work on open and interoperable infrastructure for data centers, including power delivery, cooling, and rack-level hardware. By choosing the summit, Delta is indicating that it wants its NVIDIA-aligned work to be evaluated in the same ecosystem where hyperscale buyers and component makers exchange requirements and reference designs.

Industry observers generally view power and cooling as a central constraint for AI data centers because the economics of deployment depend not just on compute performance, but on the ability to deliver electricity efficiently and remove heat reliably. Cooling systems, in particular, can become a dominant source of facility overhead when racks run at high utilization. The report’s framing of “energy-efficient” 800V DC power and cooling positions the effort as an attempt to tackle both electrical and thermal efficiency together, rather than treating power delivery and cooling as separate problems.

For NVIDIA, working with power and cooling vendors like Delta is part of how it influences the broader data center stack around its compute platforms. NVIDIA’s push for standardized, higher-voltage power delivery patterns is intended to help system designers build more efficient, tightly integrated racks for AI workloads. For Delta, aligning its designs with NVIDIA’s direction can also make it easier for customers to procure compatible infrastructure components and reduce integration effort.

The report did not disclose when specific deployments are expected, which customers would receive the 800V DC power and cooling solutions first, or whether Delta’s work is limited to reference designs or includes production supply commitments. It also did not provide any quantified efficiency results, such as power-usage effectiveness impacts, conversion-loss reductions, or thermal benchmarks. Those omissions matter because efficiency and reliability outcomes are the key decision points for data center buyers deciding how to refresh rack and power infrastructure.

What to watch next is whether Delta and NVIDIA provide more concrete technical material as the OCP Summit discussions progress, such as design documentation, compatibility guidance, or measured results from prototypes or early deployments. Buyers will likely want to see the specific performance characteristics behind the “energy-efficient 800 VDC” claim, along with clarity on how solutions like the referenced TOU2VN fit into complete AI rack architectures.