By Adam Kotrba, Director of Flat Products for Copper Development Association
We attended Data Center World 2026 in Washington, D.C., and left energized by both the pace of innovation and the candor of the conversations. The event blended strong technical content, presentations, panels, and a busy expo floor, with practical perspectives from operators, utilities, and suppliers. Throughout the show, copper again appeared in critical solutions spanning power delivery and cooling, two themes that dominated nearly every discussion.
The concept that stole the headline: Power. AI-focused data centers are exceptionally power-dense and generate heat loads that demand more capable cooling. One statistic, according to S&P, was striking: the global power “queue” tied to planned data center projects now exceeds the power serving today’s operating data centers, which is already just over 100 gigawatts. Growth is uneven by region; Asia Pacific is adding less than half of its current load, while the U.S. and Canada are expanding the most, with growth cited at 144%.
Several speakers highlighted how growing demand is now confronting the realities of interconnection constraints, leading to initial scheduling delays measured in months rather than years. In response, operators are collaborating closely with utilities to accelerate grid capacity. They are also deploying behind-the-meter generation, expecting much of it to eventually be interconnected. Most operators made it clear they prefer utilities to supply power and are willing to invest in the required infrastructure.
Growing electricity demand, driven by both data centers and broader electrification, continues to exert upward pressure on rates. Operators pushed back against the perception that data centers are the primary driver, emphasizing that they finance the infrastructure needed to support their consumption. Nevertheless, negative public sentiment surfaced repeatedly during panels, with misinformation identified as a significant factor.
Protestors gathered outside the venue, highlighting the perception gap between the industry and the public. Panelists stressed the need for clearer public messaging to address common concerns such as water use, noise, and emissions. They even suggested using mainstream storytelling (such as a documentary-style mini-series) to improve public understanding. Many speakers underscored the industry’s role in advancing renewable energy, noting that renewables can often be permitted and constructed more quickly than alternatives such as natural gas turbines or nuclear plants, which frequently face multi-year backlogs. Sustainability was a central theme, exemplified by Google’s commitment to operating on renewable energy to reduce operational emissions. Water use strategies are also evolving: several operators noted that new facilities commonly employ closed-loop systems that significantly reduce consumption, sometimes to levels lower than those of a golf course or laundromat.
Siting is another recurring theme, as many communities remain hesitant to host data center facilities in their own backyards. However, operators enjoy significant flexibility in choosing locations, particularly as hyperscalers increasingly target rural areas with abundant land and scalable infrastructure. According to S&P, regions in Texas, Louisiana, and Iowa now rank among the top ten globally for data center growth. Operators were candid about their decision-making process: they will build where they are welcomed and where power, connectivity, and permits are readily available. Some regions are proactively courting the industry; for instance, Alaska’s governor once again promoted the state’s advantages, such as cooling efficiency, affordable land, natural-gas-powered electricity at competitive rates, high-speed fiber, and plentiful water resources. Speakers also pointed to the positive community impact of data center concentrations, citing Loudoun County, Virginia (“Data Center Alley”), where tax revenues from the sector have helped lower the tax burden on residents.
Supply chain constraints continue to influence project timelines, especially for equipment with extended manufacturing lead times. Transformers (image on the right: three-phase medium-voltage transformer) remain a notable bottleneck, with several suppliers reporting multi-year backlogs. On a positive note, new product offerings focused on improved lead times are emerging, though many still depend on imports. In some cases, natural gas turbines face backlogs of up to four years, fueling increased interest in used and refurbished options. Battery Energy Storage Systems (BESS) are also experiencing rising demand. They can be supply-constrained, as data centers use storage for both backup power and to enhance power quality and buffer load variability typical of AI workloads. Notably, copper and semi-fabricated copper products have not yet been cited as sources of delay, including for domestic supply in the United States.
On the expo floor, liquid cooling was prominent, and for AI-focused deployments, it is rapidly becoming non-optional, as air cooling alone cannot keep pace with GPU heat flux. Copper remains central in these designs, particularly in cold plates, because its thermal conductivity helps minimize hot spots and improve temperature uniformity. The cold plate designs themselves are evolving quickly. Early plates were relatively small and targeted only the chip, while current designs can span a much larger portion of the server footprint to pull heat from adjacent components. (top image: copper cold plates. bottom image: cold plate internals.)
Plumbing architectures are also changing: instead of each plate having dedicated inlet and outlet connections, newer approaches increasingly interconnect plates so that one provides the inlet while another provides the outlet. I observed concepts such as integrating copper cooling bars that conduct heat outward to coolant channels at the edges of the assembly. As liquid becomes available at the rack level, it opens the door to liquid-cooled busbars, using coolant not only for GPUs but also to support substantially higher rack power without proportionally increasing conductor size. This is a compelling example of copper’s combined electrical and thermal performance.
Another major design transformation is underway in electrical architecture, with rack-level voltages advancing from 48V toward 800V and the industry moving from alternating current (AC) to direct current (DC) distribution. The efficiency rationale is clear: minimizing repeated AC-to-DC conversions reduces energy losses. Material and space efficiency are equally important.
With AC, current tends to concentrate near the conductor’s surface (a phenomenon known as the skin effect), which can increase effective resistance and require larger cross-sections. Fundamentally, power is determined by the product of voltage and current. As rack power increases, if the voltage remains low, the current must increase significantly, which in turn drives up the conductor size. Larger conductors add weight and cost, and take up valuable space that could otherwise be used for revenue-generating compute. Increasing the voltage reduces the current at the same power level, allowing for more compact distribution.
The reason for selecting 800V specifically aligns with trends in adjacent markets: electric vehicle platforms are transitioning from 400V to 800V, and some battery energy storage systems and renewables operate at even higher voltages. Safety remains paramount, as higher voltages increase operational risk and necessitate stronger safeguards and procedures. The transition to higher voltages will also depend on the availability of suitable components. While 800V-class equipment is advancing, many anticipate that ongoing supply constraints will result in hybrid architectures for the foreseeable future.
Overall, the industry is in a rare growth cycle driven by step-change advances in AI and accelerated digital demand. Power availability is the primary limiter, stretching utilities, interconnection processes, and key equipment supply chains. Public perception is also becoming a material factor, and operators will need clearer, more consistent communication on impacts and benefits. Meanwhile, enabling technologies such as liquid cooling and higher-voltage, more efficient power architectures are moving from pilots to broader deployment over the next few years. Copper sits at the center of these shifts, enabling compact, efficient power delivery and high-performance thermal management, while supporting reliability and sustainability goals through long service life and recycled content that promotes circularity.