In Meta’s vision of the future, I’ll connect with your Muse through mine.

Mark Zuckerberg is charging ahead into a future where a network of AI agents takes the reins of our daily lives—handling our shopping, managing tasks, and even communicating with other AI agents on our behalf. The pressing question isn’t whether we’re ready for this shift to a bot-driven existence, but rather: Is Zuckerberg prepared for it?

Zuckerberg is banking on the success of Muse, the personal AI agent that launched to the public last month. While it’s still in its infancy, he envisions a future where “billions” of users are engaging with Muse, creating a vast network of interconnected agents that communicate and transact with one another.

These Muses are designed to be “extremely capable” personal assistants, aimed at enhancing our lives in various ways—be it improving relationships, saving time and money, or fostering creativity. Zuckerberg believes that the engagement with Muse will eventually surpass the revenue generated by Instagram and Facebook, which currently account for the bulk of Meta’s income.

He describes a scenario where these personal AI agents collaborate with friends and family, transforming the shopping experience from simply visiting a website to having a conversation with an AI representative. Zuckerberg asserts that achieving seamless integration among these agents will unlock significant value on a large scale.

He predicts that Muse will become “highly profitable” within a few years.

This vision is likely to intrigue investors, especially as they seek justification for the massive investments Meta is making in AI infrastructure. However, those looking for a detailed roadmap may find themselves disappointed.

Zuckerberg did share some encouraging statistics: Muse has already been downloaded and set up by over 10 million users in the US and Canada, marking it as one of Meta’s most successful launches. While he didn’t provide specific retention figures, he noted that about 25% of initial users engage with the app daily. He sees a major opportunity in commerce, with a quarter of Muse users making purchases through the platform, averaging over $200 per transaction.

Meta collects a fee for each transaction made via Muse, and Zuckerberg’s perspective is that if AI agents handle shopping, the potential for transactions is virtually limitless. Unlike human shoppers, AI agents are always active, ready to scour the internet for deals and make purchases in real time.

However, Meta hasn’t clarified how this will function in practice, especially given valid concerns about data privacy. There have been incidents, such as a YouTuber reporting that his Muse agent mistakenly sent someone to his home during a Facebook Marketplace deal. While Zuckerberg claimed Muse was designed with privacy in mind, there have been reports of agents leaking private conversation content. A Canadian bank executive recently indicated that they might block Muse from accessing customer financial data.

Zuckerberg seems to believe that if consumers embrace Muse, businesses will have no choice but to follow suit. But what if consumers are hesitant or unable to participate? After all, someone has to foot the bill for the purchases made by these bots in this surreal shopping spree. Recent research indicates that consumer sentiment in the US is at a low point, with rising gas prices and inflation weighing heavily on wallets.

If Meta recognizes the potentially dystopian nature of this vision, it appears unfazed. “Agents are going to be an important part of how reality evolves, and everyone is just going to have to adapt,” said Meta’s CFO when asked about potential challenges in getting businesses on board with Muse. The company plans to keep Muse free for consumers, while businesses will bear the transaction costs.

Meta desperately needs Muse to succeed. With capital expenditures soaring and financial obligations mounting, the company anticipates spending between $135 billion and $145 billion in 2026—double its previous year’s expenditure. Additionally, it faces a $25 billion liability to New Mexico after a jury found it violated consumer protection laws. The pressure is on for Meta and other tech giants to demonstrate that their AI investments will pay off.

It’s uncertain if Meta can sustain its ambitious plans. Zuckerberg acknowledged that the reason Muse hasn’t launched internationally is due to compute efficiency challenges. Running the necessary GPUs for a network of Muses is costly and energy-intensive, and Meta seems to have hit a ceiling with its current data center capabilities. He mentioned plans to “10x” compute cost efficiency next year but didn’t elaborate on how this would be achieved. Fortunately, there’s existing research on optimizing data center operations to enhance compute capacity while reducing energy consumption. Perhaps he’ll take a look.

Ultimately, Muse’s profitability hinges on Meta’s ability to lower its compute costs. Zuckerberg himself pointed out that the compute-intensive nature of the service has limited its global rollout.

The 10 million downloads from just the US and Canada are impressive, but Zuckerberg believes that broader availability would significantly increase that number. He acknowledges that some users—and crucially, businesses—may not immediately grasp the value of Muse. “We’ll work with folks on the terms that make sense for them,” he stated, expressing confidence that Muse will prove valuable over time.

Yet, this is the same individual who introduced the Metaverse. Only time will tell how this unfolds.