Microsoft: Can Superconductors Reshape Data Centre Power?

As AI workloads intensify and rack densities climb, power delivery has become one of the defining constraints in data centre design. 

Microsoft is now examining whether high-temperature superconductors – materials capable of transmitting electricity with zero resistance – could transform how power is delivered across its global cloud estate.

Alistair Speirs, General Manager of Azure Infrastructure at Microsoft, argues that the industry must rethink conventional electrical architectures to meet rising demand.

“As the demand for AI and data-intensive computing is on the rise, the need for efficient and reliable power delivery is critical,” says Alistair, writing on the company's website.

Microsoft is investigating high-temperature superconductor (HTS) technology to understand how its data centres can meet growing power requirements while improving operational sustainability. 

Superconductors allow electricity to flow without resistance, reducing transmission losses and eliminating the heat build-up associated with copper and aluminium conductors.

Writing on LinkedIn, Noelle Walsh, President of Cloud Operations and Innovations at Microsoft, says: “As we unlock greater electric power to support cloud and AI, we have an even greater responsibility to use that power well. 

“That’s why I’m excited about our work exploring breakthrough research in high‑temperature superconductors. 

By moving power more efficiently and compactly, this technology carries the potential to minimise energy waste and reduce land use in the communities where our data centres operate.”

Rethinking electrical design inside facilities

Traditional conductors encounter resistance at every stage of transmission, generating heat and limiting the amount of current that can be moved within a fixed footprint. Superconducting materials behave differently when cooled to cryogenic temperatures, creating a pathway for current to move with zero resistance.

At the core of the approach are high-availability cooling systems that maintain the required cryogenic environment. These systems are intended to support the same levels of operational resilience expected within hyperscale facilities.

Data centres can benefit from HTS because they concentrate electrical loads in increasingly compact footprints as AI demands and high-performance compute clusters push beyond previous density limits.

Data centre operators often face compromises between expanding substations, adding feeders, reducing rack density or delaying growth.

According to Alistair, superconductors offer a way to “break this trade-off” by increasing electrical density without expanding the physical footprint. Inside facilities, more power delivered directly to racks supports higher-density workloads with improved efficiency.

He notes that high-temperature superconductor cables are lighter than copper and capable of carrying current over longer distances, enabling optimisation of distribution across racks and pods while reducing potential bottlenecks.

Microsoft has shared elements of this architecture at the OCP 2025 Summit and has tested a 3MW superconducting cable connected to a rack prototype, demonstrating the feasibility of direct-to-rack delivery. 

In practice, HTS systems have shown the potential to reduce the size of power cables by an order of magnitude when delivering power directly to a server rack.

Scaling capacity for AI growth

Power availability is increasingly cited as the primary constraint on data centre expansion. As AI systems grow, electrical infrastructure must scale in parallel.

Alistair suggests that updating power systems with superconductors could allow facilities to increase capacity without requiring entirely new transmission corridors or large-scale substation expansion.

Next-generation superconducting transmission lines can deliver substantially higher capacity than conventional lines at the same voltage level. This could accelerate site expansion and interconnection, allowing operators to deploy compute more rapidly in response to demand.

Superconductors also have implications beyond individual campuses. They could enable new facility designs by supporting higher-density electrical backbones within smaller physical constraints.

However, realising this potential requires re-examining long-established assumptions about voltage levels, distribution topologies and redundancy models.

Tim Heidel, CEO at VEIR, a Microsoft Climate Innovation Fund portfolio company, says: “Superconductors are a category-defining technology poised to transform how power is moved across the electricity value chain, stretching from generation to data centre chips. 

“At VEIR, we build complete power delivery solutions that take advantage of these remarkable materials, enabling customers to overcome critical bottlenecks in energy infrastructure, unlock new data centre capacity faster, and achieve higher power and compute density.”

Grid impact and community footprint

Beyond the data centre perimeter, superconducting transmission lines may reduce strain on surrounding grid infrastructure. Because they minimise voltage drop and can incorporate fault-current limiting capabilities, they offer the potential to enhance stability for high-demand facilities and neighbouring communities.

High-temperature superconductor systems require smaller trenches and reduce the need for large overhead lines, lowering the physical footprint of new connections. They can transfer comparable power at lower voltage, reducing right-of-way requirements and visible infrastructure.

Daniel McGahn, CEO at American Superconductor Corporation, notes: “Superconductors enabled ComEd to interconnect electrical grid substations in Chicago without disrupting local businesses or communities. Our proprietary solution uniquely increases grid resilience.”

For Microsoft, superconductors form part of a broader push to modernise data centre infrastructure alongside innovations in networking and cooling. 

If commercialised at scale, high-temperature superconductors could reshape how power is transmitted from generation through to the rack, addressing one of the most pressing constraints in AI era data centre development.