On stage on Monday CEO Tim Cook dinner’s Apple introduced a splashy cope with OpenAI to incorporate its highly effective synthetic intelligence mannequin as part of its voice assistant, Siri.

However within the superb print of a technical doc Apple printed after the occasion, the corporate makes clear that Alphabet’s Google has emerged as one other winner within the Cupertino, California, firm’s quest to catch up in AI.

To construct Apple’s basis AI fashions, the corporate’s engineers used its personal framework software program with a variety of {hardware}, particularly its personal on-premise graphics processing models (GPUs) and chips accessible solely on Google’s cloud known as tensor processing models (TPUs).

Google has been constructing TPUs for roughly 10 years, and has publicly mentioned two flavors of its fifth-generation chips that can be utilized for AI coaching; the efficiency model of the fifth technology provides efficiency aggressive with Nvidia H100 AI chips, Google mentioned.

Google introduced at its annual developer convention {that a} sixth technology will launch this 12 months.

The processors are designed particularly to run AI purposes and prepare fashions, and Google has constructed cloud computing {hardware} and software program platform round them.

Apple and Google didn’t instantly return requests for remark.

Apple didn’t focus on the extent to which it relied on Google’s chips and software program in contrast with {hardware} from Nvidia or different AI distributors.

However utilizing Google’s chips usually requires a shopper to buy entry to them by way of its cloud division, a lot in the identical approach prospects purchase computing time from Amazon.com’s AWS or Microsoft’s Azure.

© Thomson Reuters 2024