TLDRocket
Sign in

Zuckerberg says Meta’s enterprise AI opportunity extends beyond agents

TechCrunch Sarah Perez Covered by 6 sources

Zuckerberg says Meta's enterprise AI push goes way beyond the customer-service agent it launched in June. Think APIs, selling compute directly, and business agents Meta gets paid for when they deliver results.

Based on reporting by TechCrunch, Sarah Perez — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

Meta's enterprise AI plans are bigger than the agent it rolled out to businesses back in June, according to CEO Mark Zuckerberg. Speaking on Wednesday's second-quarter earnings call, he told investors the company sees room to sell APIs, business agents, compute, and other services to large customers — a lineup that could eventually open new revenue streams beyond the advertising business that still drives most of Meta's income, and beyond subscriptions, which contribute a smaller slice.

The first move is close to home. Meta plans to lean on its existing advertiser relationships, offering AI agents that work across its messaging apps so businesses can talk to their own customers through an AI layer. Zuckerberg framed the pricing the same way Meta prices ads: get paid when the agent actually delivers results. He called it a natural extension of the sales relationships Meta already has with millions of advertisers and hundreds of millions of small businesses.

Beyond that small-business base, Zuckerberg pointed to a second, less obvious opportunity — taking the coding, development, and internal productivity tools Meta built for its own engineers and offering them to outside companies, large and small. He was candid that this won't be effortless; selling to enterprise customers, he said, calls for a "different muscle" than the one Meta has spent years building around advertising.

Then there's compute. Meta says it can currently sell compute at a real premium over what it paid, which sounds like easy money, but Zuckerberg pushed back on the idea of cashing that in wholesale. He called that path "foolish" and described Meta's actual approach as a portfolio mixing short-term and long-term plans, because the company still needs that hardware for its own ambitions — including, as he put it, the interfaces people will eventually use to interact with something like personal superintelligence.

Agentic AI threads through the rest of the call too, and it isn't just a business story. Meta is promising consumers their own personal AI agents, plus AI-equipped smart glasses that can respond to the world around the wearer. Internally, Meta is also using large language models to speed up how fast it ships new social apps — recent examples include tools for Marketplace sellers, a Facebook Groups app, and an app for vibe-coded games. Zuckerberg said shipping new apps should keep getting easier, with Meta's recommendation systems doing the work of finding the right audience for whatever comes next.

My take — AI-written commentary, not fact-checked reporting

Meta trying to sell enterprise software and compute is a strange pivot for a company built on ad auctions, and Zuckerberg's own admission that this requires a 'different muscle' is the tell. Selling compute at a markup is tempting easy revenue, but the smarter play — and the one Zuckerberg seems to actually favor — is keeping that hardware for Meta's own AI ambitions rather than flipping it for a quick win. The real story here isn't the agents businesses will pay for, it's that Meta is quietly betting its future on infrastructure it doesn't want to give away.

Read more about this at: TechCrunch

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.