Meta might not be quite at the frontier in terms of AI model development, but it’s doing better than most in using AI for its business functions.
That’s the view of NVIDIA CEO Jensen Huang — arguably the most important figure in the AI industry right now, given that his chips power the infrastructure behind virtually every major AI deployment in the world. In a recent interview, Huang singled out Meta as the company that has done more than anyone else to translate AI investment into real business results.
“Nobody uses AI better than Meta,” Huang said. “If you look at the way that they’re using AI — AI went from a classical recommender system running on CPUs to now a generative AI agentic system that is making recommendations.”
The shift Huang is describing is not cosmetic. Meta’s old recommendation systems were rule-based and statistical, optimised to surface relevant content and ads based on historical signals. The new systems are generative and agentic — capable of reasoning, adapting, and creating, rather than just matching and ranking.
“Everything from the way the social media works and the way they recommend ads and help advertisers create content has fundamentally been changed,” Huang continued. “And their earnings show it, and that’s the reason why they’re investing so hard. They see just a much larger future potential for it.”
The earnings do show it. Meta’s full-year 2025 revenue came in at $201 billion, up 22% year-on-year, with ad revenue growing 24% to $58.1 billion in Q4 alone. The company’s AI-powered Advantage+ ad suite — which automates targeting, creative generation, and placement — crossed a $60 billion annual run rate in 2025, tripling in seven months. On Facebook, AI-driven redistribution of ads across users and sessions delivered nearly four times the revenue impact of simply increasing ad load. On Instagram, an expanded AI recommendation model boosted conversions by 5%.
The implications reach beyond ad performance. Huang’s observation points to something that the broader market has been slower to recognise: the companies best positioned to profit from AI are not necessarily those building the most advanced models, but those with the distribution, data, and infrastructure to deploy AI at scale against real consumer behaviour. Meta has 3.6 billion daily active users across its apps. That is an extraordinary data advantage, and it compounds with every interaction.
Meta’s aggressive capital spending reflects this conviction. Having spent $72 billion on AI infrastructure in 2025, the company has guided 2026 capex to between $115 and $135 billion — nearly double — with the bulk directed at what it is now calling Meta Superintelligence Labs. Zuckerberg has framed this as building toward personal superintelligence: AI systems that understand individual users’ goals and tailor experiences accordingly. That vision is, at its core, a monetisation thesis — and Huang appears to think it is already working. Huang himself has argued that the scale of AI capex across the industry is sustainable precisely because companies like Meta are generating the cash flows to justify it.
The endorsement matters because Huang sees the full picture. NVIDIA’s chips power the majority of AI training and inference worldwide, giving him an unparalleled view into which companies are scaling fastest and most efficiently. When he says Meta is using AI better than anyone, he is not speaking abstractly. He is describing what he observes across his customer base — and he is telling investors that Meta’s spending is not a bet, it is a return.