Two semiconductor and energy storage leaders are joining forces to tackle one of AI infrastructure's most pressing constraints: fitting massive amounts of power into increasingly compact data center spaces. Infineon Technologies and Skeleton Technologies announced a collaboration on September 2, 2026, to develop advanced power conversion systems that can deliver more electricity to dense GPU clusters while maintaining efficiency and reliability.

Why Is Power Density Becoming the Limiting Factor for AI Data Centers?

Modern artificial intelligence workloads demand staggering amounts of electricity. Graphics processing units (GPUs), the specialized chips that train and run large language models, consume power at rates that strain traditional data center infrastructure. The challenge isn't just generating enough power; it's delivering that power reliably within the physical constraints of a building. Data center operators face a fundamental problem: they need to squeeze more computing capability into the same square footage, but the power delivery systems themselves take up valuable space and generate heat.

"Power density is becoming one of the defining constraints of AI infrastructure globally. AI data centers need to deliver dramatically more power to increasingly dense GPU infrastructure, without allowing the power system itself to consume more space," stated Taavi Madiberk, CEO at Skeleton Technologies.

Taavi Madiberk, CEO at Skeleton Technologies

This constraint has forced engineers to rethink how electricity flows through data centers, from the grid connection all the way to individual servers. Traditional power conversion systems were designed for different workloads and cannot keep pace with AI's demands.

What Technologies Are Infineon and Skeleton Developing Together?

The collaboration focuses on two complementary technologies that address different aspects of the power delivery chain:

• Solid-State Transformers (SSTs): These devices convert medium-voltage alternating current (AC) from the electrical grid into high-voltage direct current (DC) that data centers use. By combining Infineon's CoolSiC power semiconductors with advanced control systems, SSTs can achieve higher efficiency and faster response times than conventional transformers, reducing energy waste during conversion.
• Peak-Shaving Systems: These use Skeleton's supercapacitor technology paired with Infineon's CoolGaN power semiconductors to smooth out sudden spikes in power demand. When GPUs suddenly require more electricity, supercapacitors can instantly release stored energy, preventing the power grid from becoming overwhelmed and reducing strain on backup systems.
• High-Power Sidecars: These are compact energy storage units that sit alongside existing infrastructure, providing rapid power delivery without requiring extensive retrofitting of data center facilities.

The key innovation is combining these technologies into a unified system. Rather than treating power conversion and storage as separate problems, Infineon and Skeleton are designing them to work together seamlessly throughout the entire power delivery chain.

"AI is fundamentally transforming power infrastructure requirements. The increasing power density of modern data centers calls for new approaches to energy conversion, distribution and storage. By combining Infineon's expertise in power semiconductors with Skeleton's innovation in supercapacitor technology and power conversion systems, we are driving the development of highly efficient and resilient power architectures that meet the demands of next-generation AI workloads," explained Andreas Weisl, Executive Vice President and Chief Sales Officer of Industrial and Infrastructure at Infineon.

Andreas Weisl, Executive Vice President and Chief Sales Officer of Industrial and Infrastructure at Infineon

How Can Data Center Operators Benefit From These Advances?

The collaboration aims to deliver three concrete improvements for data center operators managing AI infrastructure:

• Energy Efficiency Gains: More efficient power conversion means less electricity is wasted as heat, directly reducing operational costs and environmental impact. Every percentage point of efficiency improvement translates to measurable savings across thousands of servers running continuously.
• Increased Power Density: By making power delivery systems more compact and responsive, data centers can install more GPUs in the same physical space, improving return on investment for expensive real estate and infrastructure.
• System Reliability: Supercapacitor-based peak-shaving systems provide instantaneous backup power during grid fluctuations or demand spikes, preventing the cascading failures that can take down entire data centers and interrupt AI services for customers worldwide.

The collaboration also targets reducing total cost of ownership, the full expense of building and operating a data center over its lifetime. More efficient systems require less cooling, consume less electricity, and need fewer redundant backup systems, all of which add up to significant savings.

This partnership reflects a broader industry shift. As AI adoption accelerates globally, the infrastructure supporting these systems has become a critical bottleneck. Companies like Infineon and Skeleton are racing to develop solutions that allow data centers to scale without hitting physical or electrical limits. The technologies emerging from this collaboration could reshape how the next generation of AI infrastructure is built, making it possible to deploy more powerful models in more locations without requiring entirely new power plants or grid upgrades.