Overview of the New Policy Framework

In a significant move to align infrastructure development with sustainability goals, seven Chinese ministries have jointly issued the Implementation Plan for Promoting the Coordinated Transformation and Development of Digitalization and Greening (2026-2030). Released in early September 2026, this comprehensive policy framework explicitly directs the exploration of 800V High Voltage Direct Current (HVDC) power architectures for AI computing facilities. The mandate underscores a strategic pivot toward ultra-high-density computing environments, strongly supporting the deployment of advanced liquid cooling systems and single cabinets exceeding 100kW. This top-level design aims to drive a comprehensive upgrade across the data center power and cooling supply chains.

Technical Details: The Shift to 800V HVDC

The transition to an 800V HVDC architecture represents a fundamental rethinking of data center power distribution. Traditionally, data centers have relied on 48V low-voltage direct current systems. However, as single cabinet power consumption surges past the 100kW threshold, these legacy systems encounter severe limitations, including substantial transmission losses and excessive heat generation. By elevating the voltage to 800V, the new policy addresses these physical constraints. The 800V HVDC architecture drastically reduces current for the same power level, thereby minimizing transmission losses and improving overall electrical efficiency. The government directive to explore and promote this architecture signals an accelerated timeline for industry standardization and the rollout of pilot projects, positioning 800V as the critical power solution for next-generation AI machine rooms.

Cooling Innovations: Mandating Liquid Cooling Systems

Parallel to the power architecture overhaul, the policy directly addresses the thermal management crisis in AI facilities. With the explosive growth in AI large model computing power, the power consumption per server has skyrocketed, pushing traditional air cooling systems to their absolute physical limits. The implementation plan explicitly supports the technological innovation and deployment of liquid cooling and waste heat recovery systems. Consequently, liquid cooling is no longer viewed as an optional premium feature but as a mandatory standard for high-power computing environments. The policy encourages both cold plate and immersion liquid cooling technologies, ensuring that high-density AI server deployments can maintain optimal operating temperatures while significantly improving the energy efficiency of the facilities.

Market Context: AI Compute and Power Bottlenecks

This regulatory intervention comes at a critical juncture for the technology sector. The unprecedented demand for AI model training and inference has led to a continuous spike in single-cabinet power densities. Existing power and cooling infrastructures are fundamentally inadequate to support the next generation of AI accelerators. The bottleneck is no longer just about computational capability; it is intrinsically linked to power delivery and thermal dissipation. By intervening at the top design level, the ministries are proactively eliminating these infrastructural bottlenecks. The policy ensures that the physical layer of data centers can keep pace with the rapid advancements in silicon and AI algorithms, preventing power and cooling constraints from stifling technological progress.

Industry Impact: Upgrading the Supply Chain

The mandate is set to trigger a massive wave of capital expenditure and technological iteration across the data center supply chain. The policy specifically highlights several key areas for industrial upgrade:

  • High-Voltage Power Equipment: The push for 800V HVDC will directly benefit manufacturers of high-voltage power supplies, advanced power distribution units, and high-capacity busbars, driving critical hardware iterations.
  • Liquid Cooling Systems: Explicit policy endorsement will accelerate the commercialization of cold plate and immersion cooling solutions, creating a robust, long-term policy dividend for thermal management companies.
  • Intelligent Operations: The integration of low-power chips and smart operations using digital twins and AI algorithms will foster a new ecosystem of intelligent data center management and real-time energy monitoring.

This targeted support ensures that the entire ecosystem, from power generation to thermal dissipation, is modernized to support ultra-high-density AI workloads.

Future Outlook: Synergizing Compute and Clean Energy

Looking beyond the immediate hardware upgrades, the 2026-2030 plan emphasizes the deep integration of computing power and electricity networks. The policy guides the spatial layout of computing facilities toward regions rich in wind, solar, and hydro clean energy resources. This geographic realignment is supported by breakthroughs in long-duration, large-capacity energy storage and direct green power connection technologies. Additionally, the plan mandates the phasing out of outdated, small, and scattered computing facilities, accelerating energy-saving and carbon-reduction retrofits. Ultimately, this holistic approach ensures that the exponential growth of AI compute is sustainably powered by renewable energy, cementing the foundation for a green, highly efficient, and technologically advanced digital economy over the next five years.