Meta (ticker symbol: META) plans to deploy a new version of the custom chip to its data center this year to support its artificial intelligence (AI) project.。The move aims to reduce reliance on market-leading Nvidia chips and control costs associated with running AI workloads.。

Meta, one of the world's largest social media companies, has been increasing its computing power to support the introduction of generative AI products into its applications and hardware devices.。To that end, the company has spent billions of dollars accumulating dedicated chip inventories and reconfiguring data centers.。

Deploying its own chips means for Meta that it can cut hundreds of millions of dollars in energy costs and billions of dollars in chip procurement costs.。The move was seen as a positive turn, especially given the company's previous decision to suspend the first version of the chip in 2022.。

Internally called "Artemis," the new chip, like its predecessor, is primarily used to perform the process of inference, i.e., the model generates responses to user prompts based on its algorithm.。Meta is also reportedly working on a more ambitious chip designed to perform both training and inference functions.。

Despite some early missteps, the inference chip may be much more efficient than the energy-intensive Nvidia processor when processing Meta's recommendation model。The move is seen as a learning opportunity that gives new impetus to the company's growth in AI.。

Overall, Meta's initiative shows its continued investment and innovation in the technology sector, which is expected to bring greater efficiency and competitive advantage to the company。