Inflection AI has announced a major update to its enterprise platform, now opting to use Intel’s Gaudi 3 accelerators over Nvidia GPUs. Initially deploying its Pi application on Nvidia, Inflection AI intends to transition its operations to the Gaudi 3 architecture, leveraging Intel’s Tiber AI Cloud for both cloud and on-premises services.

Founded in 2022 as a personal assistant developer, Inflection AI shifted its focus to providing specialized AI models for enterprises. Following the exit of co-founders Mustafa Suleyman and Karén Simonyan to Microsoft earlier this year, the startup reoriented its strategy towards creating bespoke enterprise solutions.

The updated platform, Inflection 3.0, will enable enterprises to harness proprietary datasets for refined AI applications. Intel, a partner in this venture, plans to be among the first adopters, although the financial terms of the accelerator usage remain unspecified.

Despite utilizing Gaudi 3 for its own services, Inflection AI assures flexibility for clients, allowing them to choose alternative solutions for their end products. With Gaudi 3 poised as a cost-effective competitor, CEO Sean White highlights potential for improved cost-performance metrics. This assertion includes a doubling of cost efficiency compared to prevailing market options.

Intel’s Gaudi 3, boasting 128 GB of memory and extensive bandwidth capabilities, has been introduced amidst fierce competition from Nvidia and AMD’s latest GPUs. With competitive pricing strategies, Intel aims to bolster its presence in the AI sector.

This strategic alliance marks another win for Intel, which also has plans to introduce Gaudi 3 accelerators in IBM Cloud by early 2025. To facilitate seamless integration, Intel committed to providing guidance for easy transition to future Falcon Shores GPUs.