🧠 Food for thought

Implications, context, and why it matters.

The startup is entering a crowded power architecture race

• C2i wants to rethink power delivery from grid to GPU, joining an industry push to move past 12-volt systems toward higher-voltage setups. These include 48-volt distribution, plus higher-voltage direct current (DC) options such as 800-volt high-voltage DC (HVDC) and ±400-volt proposals for future data centers 1, 2.
• The shift tracks the steep power demand of upcoming AI server racks. Some forecasts put racks at one megawatt (MW) or more before the decade ends 1.
• On chips and boards, vertical power delivery (VPD) places power modules closer to the accelerator processor. The goal is to cut power loss while opening board space 3, 4.
• C2i will compete with Infineon, Monolithic Power Systems (MPS), and Delta. Infineon is working with NVIDIA on 800V HVDC concepts for future AI server racks. Infineon and Delta also work together on high-density power modules meant to support VPD 1, 3.

Power infrastructure is becoming a limit on AI buildouts

• C2i’s mission matches a wider view that electricity supply, delivery, and grid infrastructure can slow AI growth. The constraint can rival chip availability 5.
• The topic has moved beyond budget talk into policy debates about grid limits and economic competitiveness. Some writers also link power shortages to national security risk 5, 6.
• Separately, ETAP (a power system design and simulation software company) and NVIDIA back a “Chip-to-Grid” planning strategy. It treats the electrical ecosystem as one network from transmission to silicon 7.
• Some analysts tie power-delivery advances to “sovereign AI” buildouts, which are national efforts to build and run AI infrastructure at home. Public commentary here does not connect C2i directly to sovereign AI programs 8.