AI's Surging Power Demands Spur Data Center Efficiency Innovations

Urgent need for decarbonization and transmission solutions as AI models grow exponentially more power-hungry

Mar. 28, 2026 at 3:04pm

Got story updates? Submit your updates here. ›

As AI models become increasingly powerful and versatile, their voracious energy consumption is emerging as a critical constraint on the technology's growth. Data center operators, technology companies, and clean energy startups are racing to develop new solutions to improve efficiency, transmission, and decarbonization of the power-hungry infrastructure required to run advanced AI.

Why it matters

The rapid scaling of AI is putting immense strain on the electrical grid, with data centers housing the powerful servers and GPUs needed to train and run AI models now consuming hundreds of kilowatts per rack - a massive increase from just a few years ago. This energy crunch threatens to slow the pace of AI innovation unless new breakthroughs in power transmission, carbon capture, and renewable energy can keep up.

The details

Companies like Veir and Ardent are working to address the AI energy crisis from multiple angles. Veir is developing superconducting cables and bus bars that can transmit 10 times more power to dense server racks, while Ardent is focused on carbon capture solutions to decarbonize the natural gas power plants feeding data centers. Experts say the data center industry must collaborate closely with electrical equipment suppliers, architects, and power providers to implement these new technologies at scale.

  • In the past three years, AI models have seen exponential growth in power demands.
  • At the Imagination in Action event at Davos in January 2026, industry leaders discussed the urgent need for AI energy innovations.
  • Data centers are shrinking their construction timelines from 48 months to just 18 months to keep up with surging AI power needs.

The players

Veir

A company building new superconducting cable and bus bar designs to dramatically increase the power transmission capacity of data center infrastructure.

Ardent

A startup focused on carbon capture solutions to decarbonize the natural gas power plants that feed energy to data centers.

Tim Heidel

CEO of Veir, a company developing superconducting transmission technologies for data centers.

Erica Nemser

CEO of Ardent, a startup working on carbon capture solutions for data center power.

Got photos? Submit your photos here. ›

What they’re saying

“The data center industry is a full ecosystem with a lot of different stakeholders that all have a role to play in the adoption and scaling of new technologies.”

— Tim Heidel, CEO, Veir

“We capture the carbon that comes out of the natural gas combined cycle power plants that feed the data centers. And so we end up working a lot with the companies that are providing that power and sequestering the CO2 as they deliver those power solutions.”

— Erica Nemser, CEO, Ardent

“Energy access is the gating factor in who's going to win the AI race. And increasingly, we're going to continue to see bottlenecks show up in the electrical labor for deploying these systems, the supply chains for conventional conductors. And ultimately, the highest performance data centers that can train models the fastest will win.”

— Tim Heidel, CEO, Veir

What’s next

As data centers race to keep up with AI's surging power demands, companies like Veir and Ardent will continue working closely with industry partners to rapidly scale their innovative transmission and decarbonization technologies.

The takeaway

The AI revolution is running up against a physical constraint - the energy required to power the exponentially growing models and data centers. Solving this energy crisis through new transmission, efficiency, and decarbonization solutions will be critical to sustaining the pace of AI innovation.