Meta’s Muse AI Shows First Cracks in Growth Story
TREE NEWS reports: Meta’s AI agent Muse, launched on September 8, sparked a furious rally that lifted the stock 27% in 12 trading sessions, peaking at $777.59 and expanding its forward P/E from 14x to 21x. One month later, the growth narrative is showing its first real fractures. Alternative data from TickerTrends, analyzed by Hunterbrook Media, shows daily net new users collapsing from roughly 198,000 at the September peak to about 74,000 — a 62.4% decline. Downloads have fallen more than 21% from their peak, and app reviews are down 11.6% week-over-week.
The Cost Problem Morgan Stanley Just Quantified
Morgan Stanley published its most detailed Muse cost model to date, and the numbers are sobering. Serving a single Muse user costs between $3 and $130 per month, with a base case of $37 — roughly $445 annually. That exceeds Meta’s global average revenue per user of about $70 per year. GPU inference scales linearly with token consumption, while CPU sandbox costs scale with activation frequency and virtual machine runtime. At full utilization, the $20 Power tier would cost about $488 per month in inference alone; the $100 Maximum tier, roughly $2,925.
Monetization Is Years Away
Morgan Stanley’s Brian Nowak is explicit: subscription revenue will be “very limited for quite some time.” Meta is focused on user growth and engagement, not monetization. The real prize — a real-time bidding auction market where enterprises pay for agent-completed transactions — targets a $30 trillion consumer spending pool, but won’t arrive until late 2027 to 2028 at the earliest. Bank of America sees no material revenue before 2028; Goldman Sachs calls monetization timing the key question and privacy and security risks the single largest threat.
Capacity vs. Reality: A 366x Gap
Morgan Stanley’s capacity analysis is built on 1 billion daily active users. Reality: TickerTrends shows 2.73 million DAU, Similarweb 2.18 million, Sensor Tower about 1.8 million — none above 3 million. Based on TickerTrends data, Meta has already purchased roughly 366x the CPU capacity needed for its current user base. At the current pace of 74,000 net new users per day, reaching 1 billion DAU would take approximately 37 years.
Ecosystem Blockades and the Amazon Standoff
Amazon blocked Muse from its platform on September 20, citing failure to identify itself as an AI agent and undisclosed storage of user credentials. Costco and Netflix have reportedly imposed similar restrictions, and telecom carriers are routing Muse-initiated AI customer service calls away from human agents. Morgan Stanley reads the Amazon conflict as a negotiation over merchant data rights, shopping experience control, and data flows back to Muse — likely the first agent cooperation agreement and a precedent for future deals.
Key Takeaways for Investors
- Valuation re-rating is story-dependent: The move from 14x to 21x forward earnings was bought with a “1 billion user” narrative now contradicted by hard data.
- Cost structure is a double-edged sword: High costs deter smaller competitors, but they also mean Muse is deeply value-destructive at scale until monetization arrives.
- Watch November 4: Meta’s Q3 earnings will be the real test, when the market will demand Meta’s own Muse user data.
- China angle: Tencent’s closed ecosystem, mini-program infrastructure, and WeChat Pay rails may offer a stronger agent moat than Meta’s open-web approach.
