AI, Edge, and HPC infrastructure provider Amax has deployed a 64 Nvidia DGX B200 SuperPod cluster for an unnamed but “prominent” generative AI developer.

No information about where the system will be housed has been released, but the customer will use the cluster to “advance voice synthesis and content creation model training,” Amax said.

The deployment will see a total of 512 Nvidia Blackwell GPUs interconnected via Nvidia’s Quantum-2 InfiniBand networking platform to deliver 4.6 exaflops of AI training and 9.2 exaflops of inference performance.

Amax said it designed and optimized the rack layout and power distribution of the cluster, with its engineering team ensuring the deployment met “stringent data center requirements.”

“By combining Amax’s deployment expertise with the latest Nvidia DGX systems, we deliver a truly turnkey AI solution that maximizes performance, scalability, and efficiency for our customers,” said Don Wang, director of solutions architecture at Amax.

In addition to Nvidia B200 deployments, the company also offers Nvidia DGX H200 and GB200 systems.

Chai AI announces $20m expansion of its in-house compute cluster

Also this month, US-based Chai AI, also known as Chai Research, has announced plans to spend $20 million expanding its in-house AI compute deployment, dubbed kCluster.

According to the company, the expansion will see kCluster offer up to 1.4 exaflops of compute power, although Chai did not specify which precision benchmark this was measured in.

The system utilizes 5,000 GPUs from both AMD and Nvidia, the company added, claiming that the cluster offers nearly 10x the power of Stanford University’s 2,053 teraflops Sherlock HPC deployment, and processes 1.2 trillion tokens a day.

Founded by five former University of Cambridge students in 2020, the company is now headquartered in Palo Alto, California. Chai's platform allows users to create their own AI models and claims to serve more than 51,000 LLMs to users worldwide.

In December 2024, the Washington Post reported that authorities in Belgium launched an investigation into the company after a Dutch father took his life in July of that year, following “extensive chats” with one of the company’s AI companions.

Chai says it has “implemented a range of safety features that allow users to engage in dynamic chats while encouraging them to stay within established guidelines.”