NVIDIA's next-gen GB200 AI server chips will go into mass production in September

NVIDIA announced its next-gen GB200 server AI platform at its GTC (GPU Technology Conference) last month. Quanta Computer received its first order for the new GB200 server chips, which will enter mass production in September 2024.

2

UDN reports that Quanta has confirmed it received its first order of NVIDIA GB200 servers, calling it the "most powerful AI chip on Earth" and that mass production begins in September, and will increase quarter by quarter into 2025. UDN reports that according to industry expectations, Quantua is more optimistic about the momentum of server shipments this year than originally expected.

This is because four of the major cloud service providers (CSPs), Microsoft, Google, Amazon AWS, and Meta, are all ordering NVIDIA GB200-powered AI servers, which Quanta will make.

• Read more: Quanta to make NVIDIA GB200-based AI servers for Google, Amazon, and Meta
• Read more: NVIDIA GB200 Grace Blackwell Superchip: 864GB HBM3E, 16TB/sec bandwidth
• Read more: NVIDIA's full-spec B200 AI GPU uses 1200W of power, Hopper H100 uses 700W

We're seeing more and more large language models (LLMs) launching, with client demand for AI training servers increasing daily. UDN reports that the proportion of AI server in overall server revenue will leap from more than 20% in 2023, to over 50% in 2024 and beyond, becoming the "main product line of the server business".

Quanta is expecting volume shipment in September, with Amazon AWS and Meta having their new NVIDIA GB200 AI servers up and running in the second half of the year. Once these US cloud service providers (CSPs) successfully expand, capital expenditure and customer demand is strong -- its insatiable right now, AI GPU demand -- UDN reports that Quanta's cloud business revenue will make up 40% of its yearly revenues in 2024, expected to pass notebook computers and become Quanta's biggest production line.

NVIDIA's new B200 AI GPU can consume up to 1000W to 1200W depending on the configuration, with GB200 packages including 2 x B200 GPUs and a Grace CPU with powerful performance at the ready for AI workloads. The new GB200 and B200 AI GPUs will be the first to use water cooling, which is expected to make water cooling more mainstream in data center cooling markets.

• Read more: NVIDIA spent $10 billion on developing its next-generation Blackwell GPU
• Read more: NVIDIA CEO: our next-gen Blackwell GPU will cost $30,000 to $40,000 each
• Read more: NVIDIA GB200 Grace Blackwell Superchip: 864GB HBM3E, 16TB/sec bandwidth
• Read more: NVIDIA Blackwell AI GPU: multi-chip die, 208B transistors, 192GB HBM3E
• Read more: NVIDIA's new Blackwell-based DGX SuperPOD: ready for trillion-parameter scale for generative AI

Quanta has recently developed liquid cooling technology through its subsidiary Werder Technology, which is said to be a "major leap forward in water cooling technology". Quanta has partnered with Singtel as a technology partner, where the companies will build a "sustainable data center" in Singapore, which will help the Singaporean government to super-speed their AI development.