AI Infrastructure Spending Expands Into Power Delivery as Amber Semi Raises $30M

Key Takeaways

• Amber Semiconductor announced the initial closing of a $30 million Series C to accelerate power-delivery solutions for AI data centers.
• The company says its PowerTile vertical power-delivery device can reduce board-level power distribution losses to the processor by more than 85%.
• Amber reached silicon tape-out for PowerTile in January 2026, and says shipments to major customers are expected to begin in Q3 2026, with broader production volumes expected in 2027.
• The bigger market message is that AI infrastructure constraints are spreading beyond GPUs and memory into power architecture, efficiency, and electricity availability.

Amber Semiconductor announced on March 10 that it had reached the initial closing of a $30 million Series C financing round from new and existing investors. According to the company, the funding will be used to scale product development, expand customer engagements, and accelerate commercialization of next-generation power-delivery solutions for AI data centers.

As AI processors continue to push higher current requirements and denser compute configurations, power delivery is emerging as a meaningful system-level challenge rather than a background design detail. Amber CEO Thar Casey said in the company’s announcement that “power delivery is becoming a key performance differentiator.”

In January, the company announced the silicon tape-out of the device, describing it as a 1,000-amp vertical power solution that can be mounted on the backside of a server board directly beneath the processor. Amber says the architecture reduces the distance power must travel, lowering resistance and cutting board-level distribution losses by more than 85%. The company also says multiple PowerTile devices can be combined to support future processor platforms requiring more than 10,000 amps

.

The AI buildout is no longer only about access to advanced accelerators but also about the surrounding infrastructure stack: power management, board design, thermal strategy, interconnect architecture, and the broader efficiency of how electricity reaches the processor. Amber’s raise suggests investors see commercial potential in solving those supporting bottlenecks.

The timing matters because the broader data center market is already under power strain. The U.S. Department of Energy said in December 2024 that data centers accounted for about 4.4% of total U.S. electricity consumption in 2023 and could reach roughly 6.7% to 12% by 2028. That rapid demand growth is helping turn power efficiency into a more urgent procurement and design issue across the AI ecosystem.

AI demand is expanding the opportunity set beyond chips and into the components and architectures that make higher-density compute practical. As data centers scale, technologies that reduce energy loss, shrink board-level inefficiencies, and support more efficient power conversion may become increasingly important to roadmap planning and sourcing decisions

.

Amber says it expects to begin shipping to major customers in Q3 2026. At the same time, its earlier tape-out announcement said evaluation and testing with key partners would occur later in 2026, with initial products shipping in material production volumes in 2027.

What Procurement and Engineering Teams Should Watch Next

AI infrastructure growth may increase attention on power-management semiconductors, high-current delivery architectures, thermal design support, and related board-level components. For engineering teams, the story reinforces the need to think about power delivery as part of performance optimization, not merely as a supporting subsystem.

In that sense, Amber Semi’s Series C is less about one company’s funding round and more about the next phase of AI infrastructure. As compute density rises, the competitive edge may increasingly depend on how efficiently power can be delivered to the chip.

Need to source power-management and supporting components for next-generation AI, industrial, or high-density computing designs? Upload your BOM to IBS Electronics to identify supply options, compare alternatives, and move faster from design to procurement.