⚠ The AI Companion ShiftLow threat

Meta's moat assumes that social life means connecting people to other people, and a strange new possibility threatens that assumption: that some of the time and emotional need people once brought to human networks migrates to artificial ones. As AI companions and assistants grow more capable and lifelike, a share of the conversation, entertainment, and even companionship that flowed through the social graph could flow instead to a machine — which owns no graph at all.

The danger is subtle but strikes at the foundation. If the unit of social life shifts from person-to-person to person-to-AI, then the map of who knows whom — Meta's crown jewel — protects less, because the value is no longer in the connections between people. A rival with a compelling AI companion would not need to overcome the empty room; it would sidestep the very idea of a populated network.

The counterweight: Meta is racing to be a leader in exactly this, building AI companions and assistants directly into its apps, where its three-billion-person distribution gives it an enormous head start — and that human social needs, the desire to be seen and connected by real others, are unlikely to be wholly replaced by machines.

The risk stays low-to-moderate and speculative. A wholesale shift of social time from people to AI would genuinely erode the value of the graph — but it is far from certain, and Meta's distribution puts it among the best placed to own the AI companion rather than be displaced by one — Meta AI ships inside the same apps, built on its own Llama models¹.