Google Reveals Their Most Potent Cloud TPU, The Trillium, Announces NVIDIA's Blackwell Integration As Well
Google has it. exhibition It also announced the availability of its new “sixth generation” Trillium TPUs, and the NVIDIA Blackwell AI server in early 2025.
Google wants to expand its AI capabilities by adding cutting-edge hardware like The Trillium TPU, and NVIDIA Blackwell-powered servers to its arsenal by early 2025.
I/O 2024 was once again peppered with the word “AI” throughout the showcase, and we can't argue with that, considering how big artificial intelligence is for Google and its partners. We saw the integration of AI features into mainstream applications such as Gmail And Google Photosand as well, the firm told us that AI models like Gemini and The newly revealed Veo. To back everything up on the hardware level, Google also announced its new “Trillium” sixth-generation TPUs.
Expanding on Trillium, Google revealed in a blog post that the new TPUs can achieve a 4.7x increase in peak compute performance. The previous generation, doubling both HBM capabilities and interconnect bandwidth. This allows Google to run its AI models and activities more efficiently. Additionally, sixth-generation TPUs are known to be 67% more power efficient, allowing for widespread adoption in data centers.
Trillium TPUs have already been announced to be integrated into Google's AI hypercomputers, and the company has partnered with tech firm Hugging Face for “smooth model training and serving.”
Our partnership with Google Cloud makes it easy for Hugging Face users to debug and run open models on Google Cloud's AI infrastructure, including TPUs.
We're excited to further accelerate open source AI with the upcoming sixth-generation Trillium TPUs, and we expect open models to continue to deliver superior performance thanks to a 4.7X increase in performance per chip over the previous generation. . We will make Trillium's performance readily available to all AI builders through our new Optimum-TPU library.
– Jeff Baudier, Head of Product, Hug Face
In addition to developing TPUs, Google also announced that it has added NVIDIA's Blackwell architecture to its arsenal of “AI weapons” and claims that it will bring computing performance to a whole new level. will be able to carry on. It's important to note that few companies have access to Blackwell products, which shows how “rare” they are in the markets, at least for now. NVIDIA is actually adopting a special relationship with clients, by prioritizing one. Blackwell GPUs will power Google Cloud Server solutions and will be available by early 2025.

Combined with their most powerful TPUs integration With NVIDIA's Blackwell architecture, Google is sure to witness an increase in AI computing performance. They certainly need it, considering the scale of progress the firm has made in integrating generative AI into mainstream consumer applications and rapidly evolving with new and existing AI models.