- Launching six new data centers in North America and France
- Partnership with Hugging Face for accelerated AI inference
What happened: Cerebras unveils new AI centers
Cerebras Systems Inc., a frontrunner in AI acceleration technology, has announced the deployment of its wafer-scale AI accelerator chips across six new cloud data centers located in North America and France. This strategic move aims to enhance the speed and efficiency of AI inference, enabling companies to leverage advanced machine learning models more effectively.
The new data centers are set to be established in Texas, Minnesota, Oklahoma, and Georgia, alongside campuses in Montreal, Canada, and France. Notably, Cerebras will maintain full ownership of the facilities in Oklahoma City and Montreal, while the remaining centers will operate in collaboration with G42, a strategic partner. This expansion not only represents a significant investment in infrastructure but also positions Cerebras as a key player in the rapidly evolving AI landscape.
The company’s innovative architecture, which utilizes dinner plate-sized silicon wafers, provides unparalleled performance in high-performance computing (HPC) systems. Cerebras claims its inference service can run models like Meta Platform Inc.’s Llama 3.3 70B at an impressive rate of over 2,000 tokens per second.
This capability is more than 70 times faster than equivalent cloud solutions that rely on Nvidia’s powerful graphics processing units. Such advancements underscore Cerebras’ commitment to driving the future of AI, as emphasized by COO Dhiraj Mallick, who noted that these data centers will serve as the backbone for the next wave of AI innovation.
Also read: E-global Innovative Sdn Bhd’s HostPro2U launches Green Data Centers to power Malaysia’s digital ambitions
Also read: Nokia to upgrade Maxis data centers
Why it’s important
The launch of these new data centers is crucial as the demand for AI inference continues to escalate. Cerebras’ partnership with Hugging Face, known for hosting open-source machine learning models, will make high-speed inference accessible to millions of developers across the globe.
This collaboration allows developers to utilize Cerebras Inference directly through the Hugging Face platform, streamlining the experimentation process with various AI models. By simply selecting “Cerebras” as their provider, developers can access the fastest inference capabilities available today, significantly reducing the time required to develop and deploy AI solutions.
The implications of this partnership are profound, particularly in the context of agentic AI, which can perform tasks and achieve goals autonomously. As models like OpenAI’s o3 and DeepSeek R1 become increasingly prevalent, the need for rapid inference solutions will only grow.
Cerebras’ technology enables complex problem-solving by executing deep reasoning tasks in seconds, a feat that was previously unattainable. This leap in capability not only accelerates the pace of AI development but also empowers industries to innovate at an unprecedented rate, opening new avenues for research and application.
For more information about Cerebras, visit their official website and for updates on Hugging Face, check their news page.
