Meta CEO Mark Zuckerberg has officially acknowledged that the pace of development of AI agents in the company has fallen short of expectations. This statement comes amid growing pressure from investors expecting breakthroughs in artificial intelligence from Meta.

Why was the slowdown unexpected?

According to Zuckerberg, over the past four months, the AI agent direction has not accelerated as Meta had anticipated. The company was betting on autonomous algorithms capable of performing complex chains of actions—from managing advertising campaigns to interacting with users in virtual reality. However, practical implementation has encountered technical limitations and the need for fine-tuning models for specific business scenarios.

According to data from the analytical platform Artemis, the volume of transactions through the x402 protocol, which supports Meta's AI agents, amounted to only about $2 million over the past 30 days. This is a modest figure for a company with a market capitalization of over $1 trillion. For comparison, similar solutions from competitors, such as Google DeepMind, demonstrate many times higher activity.

What does this mean for the market?

Meta's slowdown is a signal that even technology giants face fundamental problems when scaling AI agents. The main difficulties lie not in the language models themselves, but in integrating them with real financial and operational systems. The x402 protocol, despite its technological sophistication, has so far failed to attract a sufficient volume of transactions to justify the investment.

From my perspective, this is a temporary phenomenon. Meta has unique data and infrastructure to catch up within 6–12 months. However, the current dynamics are a reminder that AI agents are still in the experimental stage, not a mature product. Investors should be patient: the path from concept to a scalable solution is rarely straightforward.