1、1 800 VDC Architecture for Next-Generation AI Infrastructure 800 VDC Architecture for Next-Generation AI Infrastructure Jared Huntington&Mike Tu The Architectural Imperative of 800 VDC and Integrated Energy Storage 2 800 VDC Architecture for Next-Generation AI Infrastructure Contents Introduction.3

2、Increasing Power Demand for Higher Efficiency.3 Load Swings and Energy Storage.5 Grid Interconnect Requirements.7 Power Distribution Options.9 800 VDC MGX Architecture.13 Datacenter Reference design.16 Industry Collaboration and Path Forward.17 Conclusion.18 3 800 VDC Architecture for Next-Generatio

3、n AI Infrastructure Introduction Only a few years ago,data centers were built around the compute space,vast data halls of servers,with power and cooling systems taking up a smaller share of the footprint.Then came the GPU revolution,transforming the data center into“AI Factories”.As shown in Figure

4、1,GPU racks are approaching 100 times greater power density compared to web servers and are increasing in power at a near-exponential pace,flipping the balance.Power infrastructure,once secondary,now rivals or even exceeds the space dedicated to compute.Figure 1:Increasing Power Generation over Gene

5、ration As CPUs or GPUs are improved,there is typically an incremental increase in the GPU thermal design point(TDP)for power of 20%generation over generation.This leads to an increase in power required per server over time.Nvidias NVLink allows multiple GPUs to be networked together to effectively a

6、ct as one large synchronous GPU and provide a significant improvement in performance when compared to operating over Ethernet connections.This networking of GPUs is most effective when connected over copper from both a power and cost standpoint,but at the expense of a limited reach due to signal int