Intel announced that Inflection AI, a prominent AI startup (which recently shifted its business focus towards enterprise, following the departure of the former CEO to manage AI operations for Microsoft), has switched from using NVIDIA GPUs to leveraging Intel Gaudi accelerators for enhanced performance.
Previously, Inflection utilized NVIDIA GPUs, like many others in the market, with their chatbot Pi targeting the consumer market. However, the latest model, Inflection 3.0, catering to the enterprise market with ‘Inflection for Enterprise,’ will run on Gaudi 3 instead.
Of particular interest is the initial phase where Inflection 3.0 will run via Intel’s Tiber AI Cloud service. This move is likely due to both financial constraints at Inflection, preventing them from purchasing Gaudi devices for in-house usage, and Intel’s push to attract new customers to utilize Gaudi chips to establish their market presence and compete with NVIDIA.
Looking ahead, for organizations seeking to run Inflection 3.0 server systems, the company will offer AI appliances with Gaudi 3 chips embedded, set to be available for sale in Q1 2025.
Source: Intel, Inflection, The Register
TLDR: Intel’s Inflection AI switches from NVIDIA GPUs to Intel Gaudi accelerators for improved performance, with a focus on enterprise markets and plans to offer AI appliances with Gaudi 3 chips in early 2025.
Leave a Comment