Zuckerberg Predicts Billions of Personal AI Agents Within Five Years

Mark Zuckerberg has staked out a bold claim about the future of artificial intelligence. Speaking on Meta's investor earnings call for the second quarter of 2026, the company's CEO said that within the next five years, it is essentially unthinkable that billions of people won't be using a personal AI agent that knows their goals.

What Zuckerberg said

It's extremely unlikely that, looking out five years, you won't have billions of people with a personal agent that understands your goals - Mark Zuckerberg, CEO of Meta

Zuckerberg stressed that AI agents need to "just work" for the average user, without requiring any technical knowledge. This is meant to be Meta's next major growth area, alongside advertising and its social platforms. He added that the company is focused on putting "superintelligence directly into people's hands" rather than centralizing the technology in one place, a comment widely read as an indirect jab at rivals such as OpenAI and Google.

Enterprise as a new pillar

Zuckerberg's vision isn't limited to consumers. On the same call, Meta's CEO laid out a broader strategy for selling AI to businesses that goes beyond agents alone. He pointed to APIs, business agents that handle customer interactions on companies' behalf, and direct sales of computing capacity, for which, he said, Meta is already receiving offers "at a meaningful premium over what we ourselves paid for it."

Zuckerberg acknowledged that selling enterprise solutions is "a different muscle" for Meta than the traditional advertising business the company has built over two decades. Still, the company wants to leverage its relationships with millions of advertisers to build new revenue streams alongside advertising and its planned subscriptions.

Market not buying the enthusiasm

Investors responded far more coolly to the results than to Zuckerberg's declarations. Meta shares fell nearly 10 percent in after-hours trading, from $585.61 to $529.15, even though revenue of $60.8 billion beat analyst forecasts of $60.19 billion. The main culprits were earnings per share coming in below expectations and a sharp drop in free cash flow to $784 million.

A second source of investor unease was yet another upward revision to this year's capital expenditure guidance, now at $135-145 billion. Meta attributes this to the need to build data centers and buy chips for training its own models and supporting a growing number of agents. The advertising segment within Family of Apps grew 27 percent to about $59.4 billion, with ad impressions up 14 percent and average price per ad up 12 percent.

What this means for the market

Daily active users across Meta's family of apps (Facebook, Instagram, WhatsApp, Messenger) averaged 3.60 billion in June 2026, up 3 percent year over year. That user base is the intended audience for the personal AI agents Zuckerberg described, tools meant to understand an individual user's goals and act on their behalf, much like the AI glasses Meta has previously touted as being able to interpret their surroundings in real time.

For Polish companies and investors, the key takeaway is the contrast between the declarations and the market's reaction. Meta is showing that the cost of building infrastructure for agentic AI is rising faster than the pace at which the technology is starting to generate revenue, and free cash flow of $784 million against revenue of more than $60 billion signals that even a tech giant is feeling the cost pressure of the AI boom.

The prediction of billions of personal agents within a five-year horizon fits into a broader pattern of big tech companies competing to make ever-bolder forecasts about the scale of future AI adoption, all while explaining rising infrastructure spending to investors. Whether Meta actually delivers on this vision will become clear over the next several quarters of results and the pace at which agents are rolled out across its apps.