Meta CEO Mark Zuckerberg has publicly acknowledged that the development of AI agents within the company is progressing significantly slower than expected. This statement, made during an internal meeting, confirms that even industry giants face unexpected obstacles on the path to mass adoption of autonomous algorithms.

Reality vs. Expectations: Four Months Without Progress

According to Zuckerberg, over the past four months, the AI agent division has not demonstrated the expected acceleration. The company had anticipated more active growth, but practical implementation has proven more challenging than technical forecasts. This is especially noticeable against the backdrop of the general hype around generative AI, where Meta aims to secure a leading position.

According to data from the analytical platform Artemis, the volume of transactions conducted through the x402 protocol, which supports AI agents, amounted to only about $2 million over the last 30 days. For a company with a market capitalization in the trillions of dollars and ambitious plans for AI monetization, this figure appears modest.

Why Have the Tempos Slowed Down?

The reasons for the slowdown likely lie in the complexity of integrating AI agents into real business processes and user scenarios. It requires not only refining the algorithms but also ensuring their security, scalability, and compliance with regulatory standards. Meta, like other players, faces the "last mile" problem: the path from a laboratory prototype to a working service has proven longer than anticipated.

My expert assessment: Zuckerberg's admission is not a signal for panic but a sober look at reality. The AI agent market is still in its infancy, and $2 million in transactions per month is more a proof of concept viability than a failure. However, for Meta, accustomed to exponential growth, such tempos may indeed seem disappointing. Investors should pay attention to how the company will address the scaling problem—this will determine who wins the race for autonomous AI.