Zuckerberg Outlines Meta's Grand AI Ambitions, Massive Infrastructure Investments

Zuckerberg stated that Meta's "long term vision is to build general intelligence, open source it responsibly, and make it widely available so everyone can benefit."

Zuckerberg Outlines Meta's Grand AI Ambitions, Massive Infrastructure Investments

In a video message posted to social media on Thursday, Meta CEO Mark Zuckerberg articulated the company's expansive vision for artificial intelligence along with sizable investments aimed at bringing this future to fruition.

Zuckerberg stated that Meta's "long term vision is to build general intelligence, open source it responsibly, and make it widely available so everyone can benefit." To realize this goal, the company is consolidating its AI research division, FAIR, with the teams working on building generative AI products across Meta’s apps.

Going forward, Joelle Pineau, the VP who leads FAIR, and Yann LeCun, chief AI scientist, will report to Meta’s chief product officer, Chris Cox. They previously reported to CTO Andrew ‘Boz’ Bosworth, who leads the Reality Labs division. (source: The Verge)

The video also highlighted Meta's substantial computing infrastructure investments to power next-generation AI systems. Zuckerberg revealed that by year's end, Meta will possess "around 350,000 NVIDIA H100s or around 600,000 H100 equivalents of compute if you include other GPUs."

To put that in perspective, 350,000 H100 GPUs carry an approximate price tag of $9 billion. Thus Meta is allocating tremendous capital towards constructing what Zuckerberg described as "absolutely massive" AI computing infrastructure that it says likely exceeds any other individual company.

In addition to bulking up infrastructure, Meta aims to leverage these resources to train ever-more sophisticated AI models. The company is currently developing Llama 3, the latest iteration of its widely popular, open-source large language model. Zuckerberg is setting loftier goals for Llama 3 and subsequent versions, expressing ambitions to achieve "state of the art" performance and eventually surpass rival models.

Zuckerberg also highlighted the company's progress in integrating AI with devices like the Ray-Ban Meta smart glasses. He described a future where AI isn't just a background technology but an integral part of our everyday experiences, readily assisting us through devices we use daily.

"Over time, I think a lot of us are going to talk to AIs frequently throughout the day. And I think a lot of us are going to do that using glasses because glasses are the ideal form factor for letting an AI see what you see and hear what you hear."

While Meta's vision is clear, the path to achieving AGI remains complex and undefined. The company faces intense competition in attracting and retaining AI talent, with Zuckerberg acknowledging that rivalries to recruit top researchers are reaching new heights, with multiple companies going for the same profile individuals, and VCs and other investors pouring large amounts of money into various projects.

To stand out, Meta is emphasizing its ambitious vision and unparalleled scale in resources. Zuckerberg believes this will motivate talent, noting "a lot of the best researchers want to work on the more ambitious problems."

A unique position that Meta is taking, in contrast to some of its competitors, is its commitment to open source AI. Zuckerberg is arguing for the democratization of AI technology as long as it can be done responsibly and safely. That's an important caveat, but ultimately, he believes that making powerful AI tools widely available will help to prevent the concentration of power, and foster equal access to technological advancements.

Meta's foray into developing general AI now puts it in direct competition with other tech giants like Google and Microsoft. Between the vast data from Meta's social platforms, billions in infrastructure investment, elite talent, and early consumer devices, the company seems well-positioned.

Let’s stay in touch. Get the latest AI news from Maginative in your inbox.

Subscribe
Mastodon