Zuckerberg Says Meta's Enterprise AI Opportunity Extends Beyond Agents

Meta CEO Mark Zuckerberg used the company's second-quarter earnings call to outline a broader enterprise AI strategy than the AI agents most of the industry has focused on, telling investors Meta intends to sell APIs, compute capacity and its own internal software tools directly to businesses.

The core pitch to investors

Zuckerberg laid out the opportunity directly: "We see a large enterprise opportunity to sell to businesses, including APIs, business agents, potentially selling compute directly." That framing positions Meta not just as a company using AI to improve its own advertising and consumer products, but as a company aiming to become an AI infrastructure and tooling vendor in its own right, competing more directly with cloud providers and API-focused AI companies than Meta traditionally has.

A results-based pricing philosophy

On how Meta intends to charge for these new enterprise offerings, Zuckerberg emphasized outcomes over simple usage-based billing: "We will get paid when we deliver results for those businesses," a pricing philosophy that ties Meta's enterprise AI revenue directly to demonstrated business value delivered to customers, rather than a flat per-query or per-seat fee model that doesn't account for whether the AI tools actually worked.

Restraint on the compute side

Despite discussing plans to sell compute capacity, Zuckerberg was careful to signal this wouldn't come at the expense of Meta's own AI ambitions, saying it "would be foolish" to "sell all of the compute and take a short-term profit." That comment suggests Meta is thinking carefully about how much of its infrastructure capacity to monetize externally versus reserve for its own model training and product development, wary of the kind of short-term revenue trade-off that could constrain its own long-term AI roadmap.

A strategy built on six distinct fronts

Zuckerberg outlined a genuinely broad set of priorities rather than a single focused bet: AI agents for customer service and business operations, leveraging Meta's existing advertiser relationships to cross-sell AI tools, selling its own internal productivity and coding tools externally, offering premium-priced compute infrastructure, developing personal AI agents and smart glasses aimed at consumers, and using AI internally to rapidly build and ship new social apps. That is a wide spread of initiatives for a single company to pursue simultaneously, spanning enterprise infrastructure sales, consumer hardware, and internal tooling all at once.

Recent product launches as proof of the "build fast" strategy

Zuckerberg pointed to a string of recent launches as evidence the AI-accelerated app-building strategy is already working in practice: a Marketplace seller app, a standalone Facebook Groups app, a gaming app called Pocket, and an AI-powered bedtime story app, a genuinely varied product slate that suggests Meta is using AI internally to prototype and ship niche consumer apps faster than its traditional product development cycles would normally allow.

Why this matters for how Meta gets valued

Meta formally entered the enterprise AI market in June, meaning this earnings call represents one of the company's clearest public articulations yet of how far it intends to push beyond advertising, its dominant and long-standing revenue source, and subscription products. If Meta's enterprise AI ambitions materialize at the scale Zuckerberg described, spanning APIs, compute, internal tools and agents simultaneously, it would mark a meaningful diversification of Meta's revenue base, reducing its heavy reliance on advertising even as it continues investing aggressively in consumer-facing AI products like agents and smart glasses at the same time.