Beyond the Price Tag: What Meta’s Muse Is Really For

Reading Time: 3 minutes

Muse isn’t Meta’s way out of advertising; it’s the newest way Meta is doubling down on it.

Coverage of Meta’s push to bring Muse to small businesses tends to read it as a search for revenue beyond advertising, or to judge it by what users visibly pay for AI. Both readings share a blind spot: they look for income with a visible price tag, and the arithmetic shows why that search comes up short, since subscriptions bought by individuals or small businesses are far too small to matter at Meta’s scale. The more plausible story is that Muse is a new instrument for the advertising business that already generates nearly all of Meta’s revenue, with the payoff arriving indirectly through ad spend rather than subscription fees.

What the coverage says

When Meta took Muse, its personal AI agent, to small businesses on 29 September, Reuters (via The Star) presented it as part of Meta’s effort to diversify its revenue beyond advertising. The next day, TechCrunch’s analysis of consumer AI economics leaned on PNC data, relayed by Andreessen Horowitz, showing that only about 2.2% of consumers pay for AI, and described Meta as already exploring the enterprise angle. To its credit, the piece also notes that Meta’s personalized ad targeting gives Muse more options for monetization and more time. My quarrel is with the weighting: the story is told in subscriptions and enterprise contracts, while the money may be flowing somewhere else.

The arithmetic of subscriptions

Start with the denominator. In the second quarter of 2026, Meta earned $59.4 billion from advertising out of $60.8 billion in total revenue, about 98%. Muse’s paid plans cost $20 a month for the entry tier and $100 for the top one, and most uses are free. At $20, a subscriber is worth $60 a quarter, so matching just 5% of Meta’s ad revenue would take roughly 49.5 million subscribers, and even 1% would take about 10 million. If every subscriber paid $100, 5% would still need around 10 million. For scale, in February OpenAI announced 50 million paying ChatGPT subscribers, a company-reported figure: Muse would need a paid base roughly that large just to reach 5%. Subscriptions, whether bought by individuals or small businesses, cannot be the point.

The blind spot

Both readings assume that revenue without a price tag isn’t revenue. Meta, however, earns its money from advertising, not from price lists, and that is where Muse plugs in. Reuters reports that it connects to Instagram analytics, Facebook Pages and Meta ad accounts; TechRadar lists drafting new ad campaigns among its uses; and Meta says it won’t publish, send or buy anything without approval. Think of it as a helper standing at the on-ramp to Meta’s ad products. In the same Q2 results, Meta’s CEO says AI is accelerating the core business today while opening the door to enterprise opportunities. I read Muse as the first half of that sentence: another brick in a very old wall.

Limits

This is an argument from structure. As far as I can tell, Meta hasn’t disclosed Muse-specific revenue or costs, and my claim covers only Muse for individuals and small businesses: Meta’s wider enterprise ambitions could still diversify its income. I’d change my mind if Meta showed Muse earning its keep mainly through something other than ad spend, such as subscription or enterprise revenue becoming a material line in its results. Muse launched in September, so Meta’s next report is the first to include it; I’ll be watching “other revenue” (about $1.0 billion in Q2, against $59.4 billion from ads) and any new Muse line.

The takeaway

The lesson reaches beyond Meta. When a company that already earns its money elsewhere launches a new product, the price list is rarely the whole story; ask which existing business it feeds. Muse’s price list is small, and that is the point: it is there to make a very large business bigger, not to become a new one. That is what Muse is really for.

References

  1. Anzar Mehraj, Meta expands Muse AI agent for small businesses, Reuters via The Star, 29 September 2026.
  2. Russell Brandom, The ugly economics of consumer AI, TechCrunch, 30 September 2026.
  3. Meta Platforms, Meta Reports Second Quarter 2026 Results, press release, 29 July 2026.
  4. Craig Hale, Meta targets SMBs with Muse AI that can run your business for you – but will anyone actually need it?, TechRadar, 30 September 2026.
  5. Aisha Malik, ChatGPT reaches 900M weekly active users, TechCrunch, 27 February 2026.

AI engine used: Claude Sonnet 5.5 (Anthropic), at max reasoning effort and with web search. Gemini (Google, set to “Automatic Gemini 3 models” on high accuracy) helped me optimise the master prompt I gave Claude. The full Gemini chat is not linked because it also contains material from an unrelated project.

How Claude was used: I set Claude up as a Socratic partner, so the thesis, arguments and decisions are mine. It asked probing questions, challenged my reasoning, suggested analytical lenses (user versus buyer, principal-agent), checked sources and figures with web search, caught a units error in my subscription arithmetic, polished the wording of my thesis, proposed the title, subtitle, outline and source shortlist, and drafted this text from the ideas I stated, adding a few supporting details that I reviewed and kept.

Prompt log: the full conversation with Claude, including every prompt, is linked here:
https://claude.ai/share/ee24a292-f035-4011-a977-d54c7337f1c1

One thought on “Beyond the Price Tag: What Meta’s Muse Is Really For”

  1. Maciej Jaczewski says:

    Really sharp argument – the subscriber math is what makes it land. One thing I’d add which is the existence of paid tiers doesn’t need to contradict your point. Even with $20 or $100 plans on offer, meta doesn’t need many people to actually pay since the subscription revenue was never going to move the needle. The paid tiers can just be there to cover costs or test willingness to pay, while the real value is still the free usage feeding ad targeting and ad spend.

Leave a Reply