During Meta‘s Q2 2024 earnings call, CEO Mark Zuckerberg outlined an expansive enterprise AI strategy that reaches well beyond the AI agents for customer service and support that the company unveiled in June. Zuckerberg described a ‘large enterprise opportunity’ comprising multiple pillars: selling APIs, business agents, direct compute capacity, and other services built for large customers. This diversification represents a potential new revenue stream for Meta, which currently relies primarily on advertising and, to a lesser extent, subscriptions. Meta intends to start by serving its existing base of millions of advertisers and hundreds of millions of small businesses that already use its platforms. The initial offering will be AI agents integrated across messaging apps, allowing businesses to interact with customers through an AI interface. Zuckerberg explained that Meta will charge based on results delivered for those businesses, analogous to its ad system.

Looking further ahead, Meta plans to extend its enterprise AI offerings to larger customers by externalizing tools that the company builds for its own internal use. ‘We’re building coding and developing and internal productivity tools partially because we need to build them ourselves, and we need to make sure that we have tools that are tuned for ourselves,’ Zuckerberg said. ‘Now that we have those, we feel like there’s a large opportunity to serve — whether that’s small businesses or larger businesses.’ He acknowledged that this shift requires a ‘different muscle’ than Meta has historically exercised in its enterprise sales approach.

On the topic of selling compute to enterprises, Zuckerberg emphasized a balanced strategy. Meta currently has the ability to sell compute at a ‘significant premium over what we paid for it,’ but he cautioned against selling all available capacity for short-term profit. Instead, Meta views its compute infrastructure as a portfolio with both long-term and short-term plans. This is tied to the company’s vision of personal superintelligence, which Zuckerberg noted will require hardware that enables seamless interaction.

The earnings call also touched on Meta‘s broader agentic AI ambitions for consumers, including personal AI agents and AI-powered smart glasses that can interact with the user’s environment. Additionally, Meta is leveraging large language models to accelerate development of its social app suite. Recent experiments include an app for Marketplace sellers, a Facebook Groups tool, and games built with vibe coding. Zuckerberg teased that more apps are on the way, facilitated by easier shipping and recommendation systems to scale usage.

Overall, Meta‘s enterprise AI strategy is evolving from a narrow agent play to a multi-faceted approach encompassing APIs, compute, and productivity tools, while maintaining a long-term view on infrastructure investment and balancing immediate revenue opportunities with future compute needs.