Google to Introduce Two Variants of In-House AI Chips, Reports Say

Google is preparing to introduce two new variants of its in-house artificial intelligence chips, according to a Bloomberg News report published on Monday.

The move could strengthen the technology company’s position in the increasingly competitive AI chip market, where Nvidia’s graphics processors currently dominate workloads worldwide.

Google’s parent company, Alphabet, saw its shares rise more than 2% in premarket trading. However, the stock had fallen 9% since the beginning of the year through the previous trading session.

New Chips Target Specialised AI Workloads

The two variants, known as TPU 8tx and 8i Max, are not entirely new chips. Instead, they are reconfigured versions designed to address specific workload requirements from customers.

Apple Intelligence Registers with Chinese Regulator for Tablets and Laptops

Google Chief Technologist for AI Infrastructure Amin Vahdat told Bloomberg News that the company had developed the variants in response to customer demand.

The TPU 8tx is designed for secure and confidential computing workloads. These include applications used by government agencies and companies operating in regulated industries.

The 8i Max, meanwhile, targets artificial intelligence applications that require longer context windows.

Google Challenges Nvidia in AI Chip Market

Google’s tensor processing units (TPUs) were initially developed for internal use. However, the company’s homegrown processors have emerged as potential competitors to Nvidia’s expensive graphics processing units.

The growing demand for AI computing has increased competition among technology companies seeking specialised chips to support advanced workloads.

Alphabet did not immediately respond to a Reuters request for comment on the report. Reuters said it could not independently verify the information.