Facebook Execs Panicking as They Try to Catch Up in AI Race

Mark Zuckerberg's game of catch-up has turned into an all-out scramble.

Scrambling

The Silicon Valley AI race is on — and according to a new report from Reuters, Facebook-formerly-Meta's AI pivot has been even more chaotic than initially thought.

And that's saying a lot, because from the outside, Meta's most recent pivot away from VR and the metaverse looks like a disaster.

To its credit, Meta has been working on AI research and development for years now. But the company's infrastructure has been severely lacking, hamstringing its efforts, according to the report. In an effort to catch up, Meta has reportedly been working to overhaul its woefully outdated development hardware for over a year now.

"We have a significant gap in our tooling, workflows and processes when it comes to developing for AI," Meta head of infrastructure Santosh Janardhan wrote in a September memo, as quoted by Reuters. "We need to invest heavily here."

Meta would need to "fundamentally shift our physical infrastructure design, our software systems, and our approach to providing a stable platform" if it wants to compete, he argued.

In other words, Meta was already behind well before OpenAI released its extremely popular ChatGPT chatbot back in November. Now, with OpenAI, Microsoft, and Google leading the pack, Meta CEO Mark Zuckerberg's game of catch-up has seemingly turned into an all-out scramble.

GPU Woes

According to Reuters, the Meta overhaul has been marked by leadership changes and capacity crunches, in addition to the scrapping of a previously unknown AI chip development project.

Until last year, Meta was apparently still relying on CPUs for AI work, as opposed to graphics cards, which are far better suited for the job.

The company was also working on a separate, custom, in-house chip, but the project was scrapped in favor of a new design that is rumored to be in development for release in 2025.

In an effort to plug that technological gap, Meta went on a shopping spree last year, spending billions on Nvidia GPUs.

We're Fine, It's Fine

For its part, Meta promises that it has all of its ducks in a row, thanks.

Meta "has a proven track record in creating and deploying state-of-the-art infrastructure at scale combined with deep expertise in AI research and engineering," Meta spokesperson Jon Carvill told Reuters when asked about the chip project and the restructuring.

"We're confident," he added, "in our ability to continue expanding our infrastructure's capabilities to meet our near-term and long-term needs as we bring new AI-powered experiences to our family of apps and consumer products."

READ MORE: Inside Meta's scramble to catch up on AI [Reuters]

More on Facebook's best AI innovations: Facebook Created an Advanced AI and It Won't Stop Criticizing Facebook