Groq expands low-latency operations via Equinix

Groq has expanded its AI inference infrastructure to the Asia-Pacific region through the deployment at Equinix’s International Business Exchange data center in Sydney, bringing its low-latency LPU-based system closer to Philippine and regional users.

The move follows earlier rollouts in the United States and Europe and is positioned to support organizations in the Philippines that require faster inference, data compliance and lower operating costs.

Groq said the Sydney facility will operate with 4.5 megawatts of capacity and is expected to become one of the largest AI inference sites in the region.

The company claims its Langue Processing Unit, or LPU, can deliver up to five times faster performance at lower cost compared with GPU-based or hyperscaler cloud systems.

Through Equinix’s Fabric, a software-defined interconnection service, enterprises in the region will be able to access GroqCloud with secured, low-latency links for AI workloads.

The setup is aimed at companies seeking to run production AI systems without sending data offshore, a key requirement in many government and regulated sectors.