Цукерберг: корпоративная AI-возможность Meta выходит за пределы агентов

Mark Zuckerberg said on July 29, 2026, during Meta's (recognized as extremist and banned in Russia) Q2 earnings call that the company's enterprise AI opportunity spans four areas at once — AI agents, APIs, compute, and internal software — and is not limited to agents alone.

What does Meta consider an enterprise opportunity?

Meta groups four areas under its enterprise AI opportunity, which Mark Zuckerberg listed during the investor call on July 29, 2026: AI agents, APIs for model access, compute, and internal software. According to the company's chief, this is "a big enterprise opportunity" that goes beyond agents alone.

Key areas named by Zuckerberg:

• AI agents — autonomous assistants for business tasks
• API — paid access for developers and companies to Meta's models
• Compute — computing capacity for AI workloads
• Internal software — corporate software built on the company's models
• The statement was made during the Q2 2026 earnings call

AI agents are programs that carry out multi-step tasks without constant human oversight. The API gives third-party developers paid access to Meta's models, while selling compute capacity allows the company to monetize the infrastructure it builds for its own needs. Internal software refers to corporate tools built on top of Meta's models.

Why the statement matters

For the first time, Zuckerberg outlined Meta's enterprise AI strategy this broadly — a strategy that had previously been associated mainly with consumer products. For a long time, the company's public AI focus rested on the Meta AI assistant inside Facebook, Instagram, and WhatsApp, as well as on the open models of the Llama family. Now the company's chief is saying outright that the commercial potential extends beyond agents and the consumer assistant.

"We see a big enterprise opportunity," —

Mark Zuckerberg, CEO of Meta, during the company's Q2 2026 earnings call.

The expansion into API, compute, and internal software means Meta is targeting the same niches where OpenAI, Anthropic, Google, and Microsoft are already earning money: selling access to models and infrastructure for businesses. At the same time, the openness of the Llama models remains Meta's point of difference from its closed-model competitors — the company gives away model weights rather than only selling access through an API.

What this means

Zuckerberg's statement at the Q2 2026 earnings call is a signal that Meta is preparing a commercial AI offering for business, not just consumer features. For now, it is a strategic framework without disclosed revenue figures for the enterprise segment, but it shows where the company intends to put its compute resources and money in the coming quarters.