After massive demand, NVIDIA's GB200 AI server chips will get into production in September

After announcing the GB200 AI server chips under the Blackwell platform, NVIDIA received a lot of orders from Microsoft, Google, Amazon AWS and Meta. Now, according to UDN, they are gearing up to enter mass production in September 2024. Quanta Computer has bagged its first order of these chips, touted as the most powerful AI chip on Earth. The anticipation is palpable, with mass production set to increase quarter by quarter into 2025.

Why does everyone wants the GB200 AI chips?

Now, why is this such a big deal? Well, for starters, the GB200 AI server chips are not just any chips. They’re expected to revolutionize the way we think about server capabilities, especially in AI. With giants like Microsoft, Google, Amazon AWS, and Meta lining up, it’s clear that these chips are in high demand. The reason? We’re seeing an explosion in the use of large language models (LLMs) and an insatiable demand for AI training servers. It’s a trend that’s only going to grow, with UDN reporting that AI server revenue expected to leap from more than 20% in 2023 to over 50% in 2024 and beyond.

But here’s where it gets even more interesting. The NVIDIA GB200 AI servers aren’t just about raw power. They’re also about efficiency and innovation in cooling technologies. For the first time, these servers will use water cooling, a system developed by Werder Technology. It’s a move expected to push water cooling into the mainstream for data center cooling markets. This is a big step, considering the power these AI GPUs consume—up to 1200W depending on the configuration. This is not necessarily good news since we know that water consumption for cooling AI-related devices is already really high.

Quanta also partnered with Singtel to build a sustainable data center in Singapore, aiming to super-speed AI development for the Singaporean government.

So, what does all this mean for the future of AI and server technology? It’s clear that NVIDIA and Quanta are pushing the boundaries, not just in terms of power but also in sustainability and efficiency. The GB200 AI server chips are set to become a cornerstone in the AI technology landscape, powering everything from cloud services to advanced AI research.

Things are not so bright because at least Microsoft has a big plan to lower its dependence on Asian markets for chips. Recently, they made a deal for AMD server chips and it also has a strong collaboration with Intel for the same goal. After all, Microsoft has plans to acquire 1.8 million AI chips this year alone.

In the meantime, NVIDIA’s GB200 AI server chips are like a breath of fresh air for all the players in the AI market. Let’s see if all or who will grab them first. We’ve learned this from TweakTown.

Leave your thoughts about the new developments in the comments section below.