➀ NVIDIA collaborates with Texas Instruments, Infineon, STMicroelectronics, and others to advance 800Vdc power architecture for AI data centers;

➁ 800Vdc distribution aims to improve power density by 85% and efficiency via dc-dc conversion, supporting future 1MW AI server racks;

➂ The project involves power component leaders like Delta, Schneider Electric, and Navitas to optimize energy infrastructure.

➀ The dilemma facing Big Tech is the cost of building AI, with estimates ranging from tens of millions to tens of billions;

➁ Some investments in cheap, effective Chinese products have raised concerns about the wisdom of some investments;

➂ Microsoft is withdrawing from data center projects in various countries, and expectations for GenAI are declining due to high failure rates and dissatisfaction with current results;

➃ OpenAI is not yet profitable and expects positive cash flow only in 2029, despite spending $644 billion on AI programs and data centers.

➀ Dell'Oro reported that datacentre capex grew 51% to $455 billion in 2024, driven by hyperscalers deploying accelerated servers optimized for AI training workloads.

➁ The top 10 hyperscalers accounted for over half of global datacentre capex, with NVIDIA's Hopper architecture and custom accelerators from Google, Amazon, and Microsoft contributing significantly.

➂ Dell led all OEMs in server revenue for 2024, with white-box server vendors capturing over 56% of total server revenue.