To reduce costs and increase processing power, Meta plans to deploy new in-house AI inference chips this year

Metaplans to deploy the new self-developed model in its data centers this year.AIThe inference chip, Artemis, which is designed to run inference workloads, will be used in platforms such as Facebook, Instagram and WhatsApp, as well as in devices such as Ray-Ban smart glasses.

Meta CEO MarkZuckerberg (name)said earlier this year that the company plans to have about 350,000 of NVIDIA's flagshipAI chip-H100, the current world's most popular NVIDIA-developed product for theartificial intelligence (AI)Server GPUs for workloads.Zuckerberg emphasized that when added up with the in-house version of the new AI chip and AI chips from other potential vendors, Meta will accumulate the equivalent computing power of 600,000 H100 AI chips. Deploying its own homegrown AI inference chip as part of the program will help reduce reliance on Nvidia chips and control the cost spikes associated with running AI loads.

Meta has been working to increase its computing power to meet the growing demand for its AI applications. By developing in-house AI inference chips, Meta aims to increase the processing power of its data centers while reducing power consumption and costs. As AI technology continues to evolve, more and more tech companies are beginning to realize the importance of self-developed AI chips and are investing heavily in related research.

Image Source Web

It's worth noting that Meta doesn't plan to go it alone. It's been reported thatMicrosoft corporation, Google, Intel, AMD and Qualcomm and other U.S. technology giants are accelerating the process of self-developed AI chips. These companies have joined the army of self-developed AI chips to respond to the growing market demand for high-performance AI chips.

However, developing self-developed AI chips is not an easy task. It requires the company to have deep technical accumulation and experience in chip design, manufacturing and optimization. In addition, self-developed AI chips need to undergo rigorous testing and verification to ensure that their performance and stability can meet the needs of practical applications.

Nevertheless, with the continuous development and popularization of AI technology, it is foreseeable that more and more technology companies will join the ranks of self-developed AI chips in the future. These self-developed AI chips will provide strong support for the development of AI technology, and at the same time will also promote the progress and development of the entire technology industry.