On March 5, Bloomberg reported that although Meta has recently reached major agreements with top chip manufacturers, its CFO said that the company is still committed to independent research and development of chips, including plans to develop processors that can train its future AI models.
Meta CFO Susan Lee
"Some of our workloads are actually tailored to our needs," Meta CFO Susan Li said Wednesday at a tech conference hosted by Morgan Stanley. "We started with sorting and recommending those types of workloads, and that's where we deploy custom chips at scale. But we anticipate and hope to expand the scope of applications over time to eventually include training AI models."
Although Meta is not a cloud computing service provider, it is one of the largest operators of data centers used to train and run AI models. In the past few weeks, the company has reached major agreements with chip leaders Nvidia and AMD to purchase chips and equipment to support AI workloads. Meta is also continuing to advance the research and development of its own AI processor.
Susan Lee said the company is purchasing different types of chips to handle tasks of different natures. "Based on the information we have now and the current needs, we think about, which chip is the best choice for each application scenario? Custom chips play a big role in this."