Ahead of Consensus.
Intelligence across tech and capital markets, for investors, policymakers, and business leaders.
9 minute read
Eleven percent is not a figure markets hand out casually to a company already worth the better part of two trillion dollars. Yet that is what happened to Meta Platforms on September 21, when its shares closed at $741.25, adding close to $190 billion in market value in a single session and delivering the stock’s steepest one-day advance in more than a year. Volume ran to nearly 48.7 million shares, roughly two and a half times the three-month average, and the close left Meta within about six percent of its 52-week high. The gain also carried the stock roughly 21 percent higher since September 8, the day Meta introduced Muse, and extended its rebound from mid-August lows past 35 percent.
The timing deserves attention, because it complicates the tidy version of this story. The surge landed two days before Meta Connect 2026, the company’s annual showcase opening Wednesday in Menlo Park, which means Monday’s buyers were not applauding anything Mark Zuckerberg had yet said. They were pricing in an outcome. What actually moved the stock were two developments that had already occurred: a fresh upgrade from Wall Street, and ten days of hard evidence that a new Meta product was finding an audience on its own merits. Connect itself remains, for the moment, a promise rather than a catalyst already banked.
It is a striking reversal for a company that spent much of the past three years being pressed by the same analysts to justify its AI spending rather than accelerate it. Wells Fargo’s Ken Gawrelski supplied Monday’s spark. On September 21 he lifted his price target on Meta to $796 from $640 while holding an Overweight rating, rebuilding the valuation on 25 times projected 2027 earnings, up from the 20 times multiple he had used before. That richer multiple is not extravagant by the standards of this AI cycle, though it sits well above where Meta itself traded for most of the past two years. His note credited early signs of a new product cycle, anchored in the performance of Meta’s Muse Spark model and the assistant built on top of it, and it simultaneously trimmed his 2026 operating income estimate by roughly 12 percent, a sign the higher target was a wager on where earnings power is heading rather than a verdict on the next two quarters.
The firm did not act alone. JPMorgan upgraded Meta to Overweight from Neutral the same week and set its own target at $820, while Rosenblatt and Citigroup had already moved to $886 and $800 the week prior. Wall Street’s average target now sits near $780, weighted heavily toward Buy and Strong Buy calls. Options traders confirmed the conviction: Monday’s volume ran more than four times its 30-day average, with roughly $3.9 billion in premium changing hands and calls outpacing puts by better than two to one. Even so, at least one large position wagered against the rally, a useful reminder that not every desk on Wall Street was buying the story at face value.
What shifted sentiment was not a keynote slide but ten days of downloads. Meta launched Muse on September 8 as a standalone personal AI agent built to execute tasks rather than merely answer questions: booking travel, managing email, filling out forms, making purchases once linked to a user’s accounts. The app runs on the Muse Spark 1.3 model, processes each session on a dedicated virtual machine inside Meta’s own cloud, and remains limited to American users eighteen and older across iOS, Android, WhatsApp, and the web. Meta offers it free at the base tier, with heavier users paying $20 or $100 a month.
By September 18, ten days in, Muse had reached the top of Apple’s US App Store free chart, ahead of ChatGPT, Gemini, Claude, and even Meta’s own Instagram, with Sensor Tower estimating well over 700,000 American downloads across that stretch, a pace that rivals ChatGPT’s own breakout in 2023. For a company that has poured extraordinary sums into AI infrastructure without a defining consumer hit to justify the spending, an assistant people are choosing to pay for is the first real proof the investment might yield something beyond marginally better ad targeting. It also supplies a rare answer to a question that has trailed the entire industry: whether consumers will actually pay for artificial intelligence, rather than simply sample it for free.
Underneath the enthusiasm sits a more complicated set of numbers. Meta’s second-quarter results, filed with the Securities and Exchange Commission on July 29, showed revenue of $60.8 billion, up 28 percent year over year and modestly ahead of consensus, while diluted earnings of $6.18 a share missed estimates by more than 14 percent. Operating margin fell to 31 percent from 43 percent a year earlier as total costs climbed 55 percent to $42 billion, weighed down by $2.4 billion in legal charges and $1.2 billion in severance from a May headcount reduction. Capital expenditures reached $31.1 billion for the quarter alone, and free cash flow collapsed 91 percent to $784 million. Zuckerberg told investors on the call that artificial intelligence was already accelerating the core business, even as that same investment squeezed margins in real time. Management guided third-quarter revenue to $61 billion to $64 billion and raised its full-year expense outlook to $165 billion through $169 billion.
A separate, one-time obligation compounds the picture. In late August, Meta agreed to pay up to $18 billion to settle a multistate lawsuit over harm to young users, with $12.7 billion guaranteed across ten years and the balance conditional on YouTube and TikTok adopting comparable safeguards. Meta disclosed it would book roughly $10 billion of that as a third-quarter legal charge, the same figure Wells Fargo cited when trimming its earnings forecast even as it raised its price target. The settlement also erased an overhang that court filings had once priced as high as $1.4 trillion, which goes some distance toward explaining why investors were willing to absorb the near-term charge without flinching.
Connect 2026 opens Wednesday with an evening keynote and a developer session the following morning, and early coverage points toward next-generation smart glasses and a lightweight mixed-reality headset, reportedly code-named Phoenix, as likely centerpieces, alongside deeper ties between Muse and Meta’s hardware lineup. None of that is confirmed. Meta has promised only the latest in virtual reality, wearables, and artificial intelligence, language vague enough to leave genuine room for disappointment against a market that has already priced in success.
That gap is the risk sitting beneath Monday’s exuberance. Meta now trades at a multiple that assumes Muse’s early numbers become durable subscriptions rather than a passing curiosity, and that Connect supplies hardware distribution rather than simply restating what investors already believe. Competition from OpenAI and Google has not paused, and converting free users into twenty or hundred-dollar subscribers remains, as it always has, harder in practice than in a research note. What can be said with confidence today is narrower, and more durable: Meta’s stock recorded its largest one-day move in over a year on a rating change and genuine early demand for a new product, arriving ahead of, not because of, the event Wall Street is now waiting to see.