Meta Platforms has a rare hit on its hands with its Muse AI agent. The question now is if it will make a big enough splash to justify hundreds of billions of dollars in capital spending. The answer is almost certainly no.

More than 10 million people in the U.S. and Canada have downloaded the Muse app, Meta Chief Executive Mark Zuckerberg said on the company’s earnings call Wednesday.

That is impressive. ChatGPT’s app took about three times as long to get to 10 million downloads, according to data from Sensor Tower, a market-intelligence firm. Elon Musk’s Grok and Anthropic’s Claude took even longer.

Meta is already monetizing Muse’s growth, with subscription tiers costing $20 or $100 a month for heavier users. Zuckerberg also raised the possibility last month that Muse could charge a small fee on transactions, and he suggested Wednesday that small businesses could adopt it to bring in customers and operate more efficiently.

And Meta is a world-leader in online advertising, which could bring in further revenue through in-app ads or sponsored recommendations.

That sounds respectable. But for Muse to pay a meaningful chunk of Meta’s AI bills, its growth and monetization would need to kick into a different gear.

Even if Muse skyrockets to 100 million users next year paying $20 a month, it would only add $24 billion to Meta’s annual revenue—about 12% of next year’s anticipated capital spending, per FactSet.

Associated advertising and fees could add more to that pot, but probably not a ton: Meta made about $67 of revenue per daily active user, almost all of it from ads, in the year to the end of the third quarter. For 100 million users, that would equate to another $6.7 billion—hardly enough to pay the bills.

The risk for Meta is that those bills pile up well before AI ventures are profitable enough to pay them.

Meta can already barely afford its current capital spending, which analysts expect to reach nearly $200 billion next year.

The company’s AI-infused ad business drove 26% revenue growth in the third quarter and is expected to keep humming. But spending is so large that more money is leaving the business than is entering it: Free cash flow was negative $2.1 billion in the third quarter, a deficit expected to balloon to negative $14.1 billion in the fourth quarter, according to FactSet.

Meta’s long-term debt has surged from $28.8 billion at the end of 2024 to $80.9 billion in the third quarter. And the company on Wednesday set the stage to raise equity later by issuing a new class of nonvoting shares to existing shareholders—a move that spooked investors. Meta’s stock fell almost 7% on Thursday.

The company needed to move quickly because competitors in the AI race were also raising money, which “resulted in an increasingly challenging environment for raising capital,” it said in a regulatory filing.

The obvious solution for Meta is for Muse to grow faster, reaching several hundred million users. Zuckerberg on Wednesday said the company was preparing for a billion of them.

That is extraordinarily optimistic, given that Muse, for all its fanfare, only has around 2.5 million daily active users now, per Sensor Tower. And analyst estimates of user growth and revenue potential are all over the map, underlining how uncertain its future remains.

Citizens analyst Andrew Boone estimated that Muse with 100 million users could generate $10 billion of revenue. Raymond James analyst Josh Beck suggested a middle-of-the-road case for Muse was 175 million users paying $20 a month, bringing in around $50 billion of revenue when adding up subscription fees, advertising and transaction fees.

It is also unclear how much it will cost Meta to serve its Muse users. Those people require more computing power to support than a typical Facebook or Instagram user. That power costs money. Analysts’ estimates of how much computing power Muse might require in the next couple of years range from a fraction of a gigawatt to several.

Constraints on computing power, independent of its cost, could be another headwind for Muse if user growth continues at a torrid pace. Hundreds of millions of users could eat up all of Meta’s planned additional computing capacity, which analysts at Morgan Stanley said illustrated one reason why the adoption path for agentic AI was “likely to be long.”Importantly, there is another limit to Muse’s growth that a few of its competitors don’t face: Users just don’t trust Meta much.

Privacy violations and persistent allegations that its platforms harmed users have made many people wary of the company and Zuckerberg. Meta in August reached a settlement with 48 U.S. states over harms to teenagers that led to a $10 billion charge in its third-quarter results. And on Wednesday, a New Mexico court ordered Meta to pay $25 billion to the state in a data-privacy case.

Only 8% of people in a recent survey said they would trust Meta with a password, compared with 30% for Google and 23% for Apple. But agents’ usefulness is limited if users won’t let them to email accounts and e-commerce sites so they can manage their lives and buy stuff.

Right now, Muse is the leading agent because it is capable, has been well marketed and is a first mover among the big tech companies. Google offers an agent, too, but it doesn’t have a dedicated agent app like Muse, while OpenAI’s Dots only launched after Muse took the spotlight. Anthropic on Wednesday unveiled its entry in the race, called Claude Duet.

When that dynamic changes and more trusted companies make agents the center of their AI strategies, Meta’s time in the agentic AI spotlight might turn out to be short.