Meta Platforms' Muse personal agent isn't even three weeks old, but the stock is up 27% in that period. Muse may turn out to be a game-changer for both Meta and the entire consumer tech landscape, but there is a long road ahead, and it doesn't change any of the near term trends in Meta's financials, where growing investments and commitments are weighing on profitability.
During the run, the stock went from trading at a small discount to the S&P 500 index's forward price/earnings ratio to getting a 21% premium, which is higher than 99% of trading days in the past five years.
Muse and similar artificial-intelligence-based personal agents aren't just apps. They represent an entirely new human-computer interface. Instead of going to Expedia to book a vacation and operating the website directly with a mouse or a finger on a screen, the user has a conversation with Muse, which then books the vacation. The graphic interface of the website fades into the background. If personal agents become widespread, websites and apps will likely have separate ways for humans and machines to interact with them. (My colleague Josh Scafer has a very useful perspective on what it is like to let Muse into your personal life)
But new user interfaces can be slow to take hold. There was a decade between the Apple Macintosh introducing the graphical user interface (GUI) in 1984 and the mass adoption of it driven by Windows 95. People will first have to break the deeply ingrained habits of using graphical interfaces on their PCs and phones and start feeling comfortable talking to an agent instead. This week's hype aside, it won't happen overnight.
At Wednesday's Connect conference keynote address, CEO Mark Zuckerberg said that "millions of people" were already using Muse. That is progress but it is a long way from the 3.6 billion daily users of Meta's social-media and messaging products. A year ago, OpenAI's Sora video app was a hit at first, racing to the top of the U.S. iOS App Store rankings in three days and staying there for more than three weeks before falling off a cliff. Sora was shut down in March after only six months. While Meta is far better than OpenAI at building apps, there is a lot of distance between "shiny new AI toy" and "people's new interface with the digital world."
The more Muse succeeds, the more it will face competition, including from Apple, which is in the process of building the very same thing in Siri AI. Meta will also have to get over the hurdle of being one of the least-trusted corporations.
Putting those serious caveats aside, if we look ahead to a future where billions of people use Muse, there is still the unanswered question of how it can be monetized.
"Muse also has a novel business model," Zuckerberg told the developers in the audience during his Connect keynote. "We believe that Muse will make you money. And we're standing behind this by making Muse free for a huge number of tokens with the expectation that over time we will profit by taking a small fee from transactions."
Call it the "Muse Tax."
There will have to be a lot of transactions for those small fees to add up to profits. After spending $72 billion last year on data centers to support Muse and other AI services like the Meta AI chatbot, the company may double those expenditures this year. Debt has risen, free cash flow has evaporated, and share repurchases have been halted. Without a cloud service for direct revenue from all that infrastructure like some of its competitors, the company's operating margin is declining, even while internal use of AI boosts its sales growth. Next year, Wall Street analysts expect revenue to grow by 20%, but earnings per share are seen rising just 9%.
On top of the capital expenditures, Meta has nearly $700 billion in future supply and lease commitments in addition to those already on its balance sheet. A month from now, when the company reports its third-quarter earnings, these trends are unlikely to have changed, regardless of what happens with Muse.
With the stock trading at a premium to the S&P 500 that hasn't been seen for over a year, even bulls may want to think about trimming their positions. The last time Meta stock was this pricey relative to the index, it declined 5.8% over the next month.
E-commerce may not be enough to offset all the costs. Another avenue could be premium subscriptions for people who quickly adopt Muse as a new user interface and use it heavily. Also, Meta hopes to transform its Reality Labs hardware division from its 2025 $19 billion loss into a profit center with a range of conversational AI-powered wearables (primarily smartglasses). But margins on hardware are famously slim.
Advertising didn't get a single mention in the 55-minute keynote even though Meta still generates 98% of its revenue that way. No one is better at collecting user data from free digital services and turning it into ad sales than Meta. That is thanks to great app design and a previous indifference toward user privacy.
That is a tough setup for a company now pushing a personal agent and promising that user interactions with the agent will soon be hidden even from its own view. Muse will effectively be a black box, and it will be tough to advertise through the service, despite Meta's ad-first DNA.
As difficult as monetizing Muse proves to be, the bigger challenge will be getting people to fundamentally rethink how they interact with computers-and Meta itself. For Zuckerberg & Co that hard work is still to come.
Comments