Meta recently revealed plans to deploy 600k Nvidia H100 GPUs by end of 2024. For context, that’s about 8x the size of Tesla’s targeted AI supercomputer by that timeframe. It represents a staggering investment in AI infrastructure, more than doubling Meta’s GPU power since the H100 launched, why is Meta massively scaling up AI capabilities amid slowing revenue growth and big layoffs?
The ~600K GPU target implies Meta is spending upwards of $30 billion on H100s alone over the next two years. That’s a huge chunk of change for a company with ~$100 billion in annual revenue. Net income fell nearly 20% in 2021-2022 as the social media business struggled.
Yet even after laying off 11,000 employees, Meta is going pedal to the metal on AI investment. This has some investors scratching their heads. Does Meta know something we don’t about monetizing AI? Or will this go down as another wasteful spending spree if the metaverse doesn’t take off?
On one hand, Meta needs to make big bets to stay relevant in the future. AI powers everything from targeted advertising to content recommendations to hate speech detection. Doubling down on AI could help Meta defend its turf against TikTok and other competitors.
And who knows — AI breakthroughs using that massive GPU fleet could unlock new revenue streams down the road. If Meta cracked general artificial intelligence, it would be incredibly valuable. The same goes for progress in areas like natural language processing and computer vision.
But it’s not guaranteed those 600K H100s will pay off. Meta’s track record with long-term moonshot investments is mixed at best, the staggering price tag here, potentially $30 billion or more, comes with massive opportunity cost.
That’s tens of billions that could’ve gone to hiring, new products, or even dividends and buybacks when the stock is depressed. If Meta’s core business continues to struggle, shareholders may view this AI spending spree as a waste.
For now, Meta is making a big bet that advanced AI capabilities will drive future success. The company better hope these GPUs power more than just improvements to News Feed. With growth slowing, Meta needs major new revenue streams before investors start asking what the point of all this AI investment was in the first place.