A new American Civil War is coming?

Analyst Targets

• Goldman Sachs: Buy, $725 price target (reiterated earlier this week; cautious Friday note on hyperscaler return-on-investment)
• Oppenheimer: Perform (no target), citing monetization uncertainty and compute-capacity strain
• KeyBanc (Justin Patterson): Overweight, $900 target (raised Tuesday from $780)
• Tigress Financial: Strong Buy, $995 target (raised Thursday from $945)
• JPMorgan: Overweight, $920 target
• Wedbush / BMO Capital: Hold, reaffirmed Friday

What Happened Friday

Meta Platforms shares fell 3.7% in mid-day trading after Goldman Sachs struck a less optimistic tone, prompting investors to lock in profits following the viral success of the Muse AI personal agent. The stock settled at about $751, leaving it roughly flat on the week after touching $780 intraday on Thursday.

Muse had reached 2.8 million downloads within two weeks of its September 9 launch, topping Apple’s App Store and Google Play charts in the United States and Canada, and had catapulted the stock roughly 16% over the course of the week before Friday’s reversal. That kind of move in five sessions demands a clean catalyst to sustain it. Friday offered three reasons to pause instead.

The Three Headwinds

1. Goldman’s Return-on-Investment Warning

Goldman warned that AI hyperscalers need to generate roughly $300 billion a year in AI services revenue just to break even on their capital spending, and that AI users may need to spend roughly $1 trillion annually on AI applications for hyperscalers to generate solid returns on investment. That framing reoriented the conversation from Muse’s download chart to the denominator: major Big Tech AI spending is rising sharply year over year.

Goldman still carries a Buy rating on the stock. But the note landed differently on a Friday after a 16% week, when investors were already looking for permission to trim.

2. The Muse Security Flaw

An ethical hacker discovered a flaw in Meta’s Muse AI agent that could potentially grant unauthorized access to a user’s cloud-based private information, according to The Information. The vulnerability was reported through Meta’s bug bounty program and was initially classified as a SEV-2, Meta’s third-highest severity level, typically used for incidents with significant impact.

To exploit the vulnerability, an attacker would need to trick a user into asking Muse to summarize a link to a compromised webpage, then have the user manually select “allow” on a system prompt containing a security notice. Meta’s response was procedural rather than architectural: the company is making safety warnings clearer and more prominent when it detects malicious sites. The Muse safety warning changes represent Meta’s response to another security issue surfacing in its personal AI agent this month. For a product whose commercial premise rests on users granting it access to email, calendars, and financial accounts, the accumulation of incidents matters more than any single flaw.

3. Oppenheimer’s Monetization Math

Oppenheimer reiterated a Perform rating, citing uncertainty around Muse’s monetization and concerns about fiscal 2027 operating expense estimates. Analyst Jason Helfstein noted that despite approximately 340,000 U.S. daily downloads, monetization remains uncertain. The firm’s arithmetic was sobering: an acceleration to 500,000 downloads per day for more than a year would increase 2027 estimated revenue by only 4%, assuming 100% download-to-DAU conversion, zero churn, and 28% of users paying.

Muse has reached approximately 700,000 daily active users, but rapid adoption is exposing potential limitations in computing capacity, with resource overconsumption, feature downgrades, and task-execution failures suggesting infrastructure requirements could increase rapidly. The cost side of that equation is steep: Meta’s 2026 capital expenditure outlook is $130 billion to $145 billion.

Forward Scenarios

Bull: Muse daily active users sustain above 1 million, Meta scales its agentic commerce partnerships with Shopify and PayPal into a fee-based revenue stream that proves Oppenheimer’s subscription math wrong. Compute costs fall as Meta’s in-house MTIA 450 chip reaches data centers in H1 2027. The stock reclaims $780 and pushes toward the $900 targets from KeyBanc and JPMorgan.

Base: Muse download momentum holds but meaningful monetization arrives no earlier than 2028, in line with Bank of America’s estimate. Security incidents stay contained and do not affect user growth. Goldman’s break-even framework weighs on sentiment intermittently. META trades in a $720-$800 range through year-end as the market waits for Q3 earnings to clarify compute cost trajectory.

Bear: A fourth or fifth Muse security incident triggers a measurable user backlash, slowing daily active user growth. Capex revisions push 2027 estimates above $200 billion, compressing free cash flow. Goldman’s $725 target becomes the floor to test rather than the discount it looks today. The stock revisits $680.

Technical Overlay

Meta traded at approximately $751, still well above the $520.26 52-week low set earlier in the year, reflecting how dramatically sentiment has shifted since the Muse launch. The Friday reversal came precisely at the $780 area that capped Thursday’s intraday high, making that level the near-term resistance to watch. The 20-day moving average sits closer to $680, meaning Friday’s pullback barely registered against the trend. A close below $720 would be the first technical signal that the Muse rally is unwinding rather than simply consolidating.

Bottom Line

Friday’s 3.4% decline is not the Muse story reversing. It is the market doing basic arithmetic it skipped during a euphoric week. The product is real, the downloads are real, and the competitive advantage of running an agent inside a dedicated cloud virtual machine with access to email and transactions is genuinely differentiated. What is also real: Meta projects capital expenditures of $130 to $145 billion for 2026, and while the advertising business continues to show robust growth, the sheer dimension of infrastructure costs is forcing investors into a new sobriety. The security incidents do not break the thesis, but they introduce a trust variable into a product whose entire value depends on users handing over the keys. Watch whether the SEV-2 flaw and its predecessors affect September retention data when Q3 earnings arrive.