Siemens Bets On Supercapacitors To Make AI Factories Better Grid Citizens

On October 14, 2026, Siemens announced a strategic investment in and exclusive global partnership with Skeleton Technologies, a developer and manufacturer of supercapacitor-based high-power energy storage. Siemens will integrate Skeleton Technologies’ systems into its end-to-end data centre electrical infrastructure portfolio and gain exclusive rights to sell and scale Skeleton Technologies’ products in the data centre market..

The move targets an increasingly common problem for grid operators: AI workloads synchronize thousands of accelerators whose power demand can rise or fall within milliseconds, aggregating into megawatt-scale swings within seconds. Siemens positions Skeleton Technologies’ supercapacitors as an electrical shock absorber that soaks up these swings – protecting both facility equipment and the grid – while allowing operators to raise compute density without costly, time-consuming facility upgrades.

A white space product with a grid-side value proposition

Most coverage frames the deal as a white space power play, and Skeleton Technologies’ stated ambition is to become the world’s largest provider of white space power systems for AI data centres. From a digital grid technologies perspective, however, the more consequential effect is on the utility side of the meter. Regulators are moving to treat large computational loads as grid participants with performance obligations, rather than as passive demand. The North American Electric Reliability Corporation (NERC) January 2025 incident review of a July 2024 event in Virginia – in which roughly 1,500MW of data centre load dropped off the system following a 230kV line fault – put large-load behaviour firmly on the reliability agenda. In 2026, the Federal Energy Regulatory Commission (FERC) ordered new reliability standards for data centres and other computational loads.

Peter Koerte, President of Siemens Smart Infrastructure, summed up the logic: “The race to scale AI is increasingly becoming a race for power”. Verdantix sees three implications for grid technology buyers and vendors:

• Smoothed load profiles become an interconnection asset.

  Utilities sizing connections and running planning studies for AI campuses must assume worst-case ramp and oscillation behaviour. Operators that can demonstrate a buffered, predictable demand profile at the point of connection will have a stronger case in interconnection negotiations – but only if utilities can verify that behaviour through telemetry and validated dynamic models.

• Siemens is assembling a multi-timescale flexibility stack.

  In March 2026, Siemens Smart Infrastructure invested in Emerald AI, integrated Fluence battery energy storage, and partnered with PhysicsX to align compute demand with grid constraints. Supercapacitors fill the millisecond-to-second layer that batteries and workload orchestration software address less well. The open question is orchestration: who coordinates rack-level buffering, facility batteries, on-site generation and utility signals as one controllable resource? This is where microgrid controllers, DERMS and grid edge software – including Siemens’s own Gridscale X and Electrification X portfolios – will need to evolve.

• Fast storage is not a substitute for ride-through compliance.
  The incident in Virginia was driven by customer-side protection and controls reacting to voltage depressions, not by workload swings. Supercapacitors can support ride-through, but only if protection settings, UPS logic and facility controls are designed to use them. Utilities should treat these deployments as one input to large-load interconnection requirements, not as evidence of compliance in themselves.

What to watch

Exclusivity gives Siemens a differentiated component at a time when electrical equipment suppliers, grid edge software vendors and server makers are all competing to own data centre power quality. Over the next 12 to 18 months, Verdantix will be watching for three signals: published field performance data from hyperscale deployments; whether Siemens exposes supercapacitor state and performance to its grid software and to utility operators via open interfaces; and how quickly large-load interconnection rules in North America and Europe translate into procurement requirements.

Utilities and grid operators should start engaging data centre customers now on what load-side flexibility they will be able to see, model and rely upon – before it is written into connection agreements. Digital grid technology vendors, meanwhile, should treat behind-the-meter fast storage as a new class of controllable asset that their platforms will be expected to monitor and coordinate.

Join the upcoming Verdantix webinar, Energy Risk In 2027: How To Protect Your Operations And Energy Budget From The Super El Niño, on December 1, 2026 at 4pm GMT (11am EST).