Meta Platforms, the corporate entity overseeing Facebook and Instagram, announced that it is developing a custom chip “family” to improve the efficiency of its data centers.
According to reports, the company said that the chips will be designed specifically for artificial intelligence (AI) workloads, and that will help Meta to improve the performance of its AI-powered products and services.
Meta also discussed their future plans to redesign its data centers, emphasizing the implementation of advanced AI focused networking and cooling systems.
The cost-effectiveness of integrating AI into data centers, particularly in regard to generative AI (genAI), has sparked debates. Projections indicate that the implementation of genAI is expected to result in a substantial rise in both data center server infrastructure and operating expenditures.
A recent forecast by Tirias Research estimates that by 2028, these costs will exceed $76 billion, challenging the business models and profitability of services that incorporate genAI. The report added that this projection incorporates an aggressive 4X improvement in hardware compute performance, but this gain is outweighed by a 50X increase in processing workloads, despite the rapid rate of innovation around inference algorithms and their efficiency.
There has been a boom in the Asia-Pacific data center industry as the market continue their focus on cost management strategies. A recent report by Research and Markets stated that the Asia-Pacific data center market will reach US$ 53.58 Billion in 2028 and expand at a CAGR of 12% from 2023 to 2028. This growth is fueled by the soaring demand for data center operations and a surge in investment.