Overview: The Shift from Copper to Control

Powering a server was once a solved problem. For decades, a typical server rack drew 10 to 15 kilowatts from a 48-volt distribution bus, and most racks outside AI still do. Artificial intelligence has ended that. Today’s AI racks already draw around 130 kW, and the next generation is designed for a megawatt. Compute density keeps climbing, distribution voltage is jumping by more than an order of magnitude, wide-bandgap devices switch faster than silicon, accelerator loads swing in fractions of a second, and efficiency, safety, and security expectations keep tightening. Every force lands on the digital controller that runs each conversion stage between the grid and the processor.

Eric Chou, Director, Head of Power & Energy Competence Center (PECC), PDSA-China Region, STMicroelectronics, frames the shift: “Everyone is watching the power devices get bigger and faster. The quieter story is what is leaving the rack: copper, passive bulk and margin. What replaces them is control, and that is where the next decade of AI power will be decided.”

The Rack That Changed Everything

NVIDIA’s 800 VDC architecture is designed to scale from 100 kW to more than 1 MW per rack, with the first megawatt-class racks expected between 2027 and 2028. The 54-volt busbars inside today’s AI racks cannot follow. By NVIDIA’s calculation, a single 1 MW rack at 54 V would need up to 200 kilograms of copper busbar, and a one gigawatt data center up to 200,000 kilograms. Distributing at 800 volts DC removes that copper and, by the same estimate, improves end-to-end efficiency by up to 5%.

The architecture splits power conversion from IT equipment: a dedicated power rack feeds IT racks with high-voltage DC, an approach hyperscalers are standardizing through the Open Compute Project’s Diablo specification (OCP, October 2025). Heat follows the same curve, and current rack-scale AI systems are liquid-cooled by design. Each step removes copper and losses and concentrates responsibility in the remaining conversion stages that step 800 volts down to processor voltages.

Why 800 Volts Changes the Control Problem

Raising the voltage is right for efficiency and harder for control. Every switching transition now happens across 800 volts, so it must be faster and cleaner. A fault at 800 volts is far less forgiving than one at 54, so protection must be tighter, including when a technician swaps a live module. The conversion chain is also getting shorter: to save losses, the industry is collapsing stages, leaving fewer, denser stages with more responsibility and less margin. A converter can now step from 800 volts directly to a low-voltage bus beside the processor, a jump already demonstrated for 800 VDC racks.

The load adds difficulty. An AI accelerator’s power draw swings with computation in fractions of a second, which is why the 800 VDC architecture includes energy storage to absorb those swings. Every stage that stores, converts, or delivers power is a controlled stage with its own real-time loop, and all must be coordinated so the rack behaves as one clean source.

What Is Now Asked of the Controller

The switches are increasingly wide-bandgap devices—gallium nitride (GaN) and silicon carbide (SiC)—which switch faster and waste less energy than legacy silicon. GaN can switch near one megahertz, and GaN converters running directly from 800 volts have demonstrated efficiency above 98%. Faster switching shrinks the magnetics and capacitors around each switch, which is how passive bulk leaves the power stage. That speed only pays off if control keeps up: at one megahertz the whole switching cycle lasts a millionth of a second, and every transition must land inside it with nanosecond accuracy.

The controller’s job has changed. It must be fast and precise; think harder, because control strategies for wide-bandgap devices demand far more computation per cycle; protect in hardware, because a fault at 800 volts leaves no time to consult software; endure heat beside the hottest components for years; and be trusted. Power infrastructure is critical infrastructure, so the controller must verify its own software, accept only authenticated updates, and remain trustworthy over a service life that will outlast today’s cryptography—which is why readiness for the post-quantum standards NIST finalized in 2024 is entering the requirement now.

The Power Stage as a System Decision

The digital power stage is no longer a component choice. It is a system decision. Switch, gate driver, magnetics, and controller must be specified together, because the limit of one sets the ceiling for all the others: a faster transistor is wasted behind a slow loop, and a precise loop cannot rescue a poorly driven switch. Operators and their system partners increasingly evaluate the stage as a whole. That favors suppliers who can co-design wide-bandgap switches, drivers, and digital controllers. The OCP Diablo specification signals that hyperscalers want standardized high-voltage DC power racks, pushing the industry toward common interfaces while leaving differentiation in control and conversion performance.

Implications for AI Data Centers

For operators, rack power delivery becomes a first-class design problem. A 1 MW rack cannot be served by scaling up yesterday’s 54-volt busbar; it requires a dedicated power rack, high-voltage DC distribution, liquid cooling, and conversion stages that step down to processor voltages with minimal margin for error. The controller sits at the center of that chain, coordinating energy storage, managing load swings, enforcing hardware-level protection, and remaining trustworthy over a service life that may span cryptographic transitions.

Future Outlook

Between now and the first megawatt-class racks expected in 2027–2028, the industry will rebuild the power path around 800 VDC. Copper will leave the rack; control will take its place. The winners will be those who treat the power stage as an integrated system—switch, driver, magnetics, and controller—and who can deliver nanosecond-scale precision, robust protection, thermal endurance, and authenticated, future-proof firmware. The quiet story of AI power is not only bigger and faster devices. It is the rise of control as the decisive factor in how AI factories are powered.