Shares rose about 6% on the launch of Meta's personal AI agent. The same stock rose 6.06% five months ago on the model behind it, and neither move arrived with a usage or revenue figure attached.
FINANCIALMARKETS.COM | AFTERNOON EDITION
Meta Platforms launched Muse on September 8, describing it in its own release as a personal AI agent available on iOS, Android and the web, with a free tier and paid plans at $20 and $100 a month, built on a model the company calls Muse Spark.
On Wednesday the stock traded as high as $657.86 and sat near $651.50 in the afternoon, up 6.20% from a prior close of $613.48, on volume of roughly 23.2 million shares. It did that against a falling market, with nearly 70% of US issues lower on oil-driven risk aversion, and while Apple traded down during its own product event. Those two facts together make a company-specific driver more plausible than a sector one.
The part missing from most of Wednesday's coverage is that this already happened. On April 8, 2026, Meta shares rose 6.06%, to $650.65, on the release of the Muse Spark model that underlies Wednesday's product. Same ticker, same product family, nearly the same percentage move, and nearly the same absolute share price. Meta had originally planned to ship the full consumer agent in April and delayed it to address security concerns before releasing it in September.
Neither move came with a number. Meta has not disclosed a Muse download figure, an active-user figure, a subscription count or a revenue contribution, and no such figure exists in any public source. The bull case is therefore resting on analyst conviction rather than company data. KeyBanc reiterated an Overweight rating and a $780 target. Morgan Stanley put a roughly $30 trillion addressable market on consumer agents while explicitly noting that adoption evidence is still required. Mizuho commented the same day. None of those were new rating or price-target actions.
The most useful bear framing is not a bear at all. Bank of America, in a note dated September 7, one day before launch, carried a Buy rating and an $810 target while valuing the shares at 18 times 2027 GAAP earnings against a 21 times historical average, a discount it tied to advertising cyclicality, AI capital-expenditure pressure on margins, infrastructure inflexibility and regulatory overhang. That is a bullish analyst pricing in exactly the unresolved-monetization skepticism the April-to-September repeat raises.
Separately, security and privacy issues were flagged during Meta's internal testing before launch, including a test scenario in which private photos were exposed. Meta has not addressed those findings publicly.
Meta Connect runs September 23 and 24. It is the company's own developer conference and the most plausible venue for the first Muse adoption figure. Until one exists, the April comparison is the only quantitative test available for whether this is a product cycle or a pattern.