"Please Don't Stop The Museic": Everything You Need To Know Ahead Of Meta's Earnings
"Please Don't Stop The Museic": Everything You Need To Know Ahead Of Meta's Earnings
Two months ago, Facebook Meta was the Magnificent 7's designated AI loser: lots of capex, nothing to show for it, and a stock that bottomed at $543.67 on August 18. Then on September 8 Zuckerberg launched Muse, followed in October by the Muse Wonder frontier model, and the stock ripped 43% in five weeks to a $777.59 high on September 24 (despite exploding churn that would make Cablevision blush).
Tonight, a week earlier than usual, Meta reports Q3 earnings, and the bar is about as high as it gets this earnings season. As JPMorgan's TMT trader Mark Schilsky puts it in his latest Tech Sketch earnings preview, " Please Don't Stop the Museic " ( available to pro subs ):
"A lot can change in two months. Two months ago, META was trading ~$550 because investors viewed it as a relative AI loser (with a pretty meaningful legal overhang). They thought META had nothing to show for all of Zuckerberg's largesse. Well, he delivered: Muse and Wonder. As such, the stock is up nearly +40% off the lows and investor sentiment is white hot . Can management keep the party going? All eyes are on 2027 – especially the Opex guide – if we even get it this print... But we've a LOT of goodness recently. We're going to need some more surprises to break out to $800+ …"
A small correction: the stock is up closer to 30% from the lows now (and falling). Meta just snapped a five-session losing streak, its longest since July, and has closed in the red in nine of the last 11 trading days . At around $708 it sits 9% below the September peak. Which raises a question that a lot of people on the buyside are asking quietly, and UBS asked out loud:
"Meta reports a week earlier than normal – trying to get ahead of the (cap raise) crowd?" - UBS
— zerohedge (@zerohedge) October 18, 2026
We'll get to the cap raise. First, the bar.
Expectations: 9 Out Of 10
Schilsky grades investor expectations for every big Internet name reporting over the next two weeks on a 1-to-10 scale. Netflix, which reported last night, got a 3. Alphabet gets a 5. Amazon, Spotify and Etsy each get a 7. Meta gets a 9 , the only red box on the sheet ( note the asterisk on the 2027 guides: "If we get it" ):
Goldman's desk sees the same thing from the positioning side. The bank's cash desk has Meta positioning at 8/10 , and as the Goldman US equities desk put it in its Mid-Day Color note today, the " stock remains an overweight for most investors following an explosive move higher post-Muse release ." Goldman's AIS team adds that the buyside is "largely braced for a top-line beat, with focus on FY26 capex guide and whether Muse plus ads monetization lets management keep raising OPEX into 2027 without the stock punishing them for it ."
Meta has plenty of company. Even as hedge funds dumped tech at the fastest pace in a decade (Info Tech saw " the largest 2-week % net selling on our record ," -3.3 standard deviations, per Goldman Prime), the one corner nobody wants to sell is the megacaps: Mag 7 net exposure on the Goldman Prime book just hit a new record high , now around 22% of total US net exposure:
Put differently, everyone is in the room. Goldman's sales desk framed tonight as " the first live test of whether a pure play hyperscaler holds the line on 2026/27 capex while FCF is under pressure."
How big a move is priced in? According to Goldman's derivatives desk (GS Vol Views, available to pro subs ), the implied move of 6.57% "looks cheap vs avg realized move 9.9% past 5 prints" , while "flat near-term put call skew (20th %-ile vs 10y) shows constructive positioning." We checked the math, and Goldman is right: Meta has moved at least 8% the day after each of its last five reports, in both directions. Notably, the last two went the wrong way, and so did the last two Q3 prints:
Meta's earnings are big enough to move the index too: Goldman's TMT specialists flagged that Friday's QQQ straddle is pricing a 2% move around the print, and see "attractive risk-reward in owning QQQ or META gamma." (The last time anyone tried that, on October 13, META won the gamma squeeze lottery .)
The Quarter: "Just" The High End, Please
On the actual Q3 numbers, the bar is simple. Meta guided to $61-64 billion in revenue, and according to Schilsky, "while channel checks have reportedly been solid, investors are expecting 'just' the high end of the $61B to $64B guide this quarter. As such, the bogey is ~$64B (+26% Y/Y FXN) – with a slight upward bias. The Street is at ~$63.25B."
Here is the full Bloomberg consensus. Note that the median revenue estimate is $63.33 billion, while the single highest sellside estimate, $64.07 billion, is basically where the buyside bogey starts. The EPS range, $2.61 to $4.43, is also unusually wide for a company with this much coverage, probably because not everyone has penciled in the same legal charges (more on that below). And Q3 capex is expected at $41 billion, vs. less than $19 billion a year ago:
The good news for bulls: the ad market is cooperating. Goldman's internet analyst Eric Sheridan, in his " US Digital Ad Q3'26 Preview " ( also available to pro subs ), sees "a resilient digital advertising environment into Q3, with solid momentum exiting the quarter," with AI "increasingly becoming a structural driver of ad share gains." Sheridan reiterated his Buy and raised his price target to $860 from $725 . Deutsche Bank's Benjamin Black (Buy, $820 target, now at 27x 2027E earnings "to reflect the strong adoption of the Muse agent") is at $63.7 billion and cites agency checks showing CPM growth "accelerated ~350bps sequentially in 3Q after accelerating roughly 650bps in 2Q," which "compare favorably to Street expectations for approximately 150bps of FX-neutral deceleration."
For Q4, the Street is at ~$73.5 billion (+24% FXN). Schilsky says investors "will tolerate no more than a 1-point FXN decel on the top end of the guidance range," so the expected guide is $72-75 billion . Deutsche hears a slightly lower top end, "roughly $74-74.5bn."
In other words, a beat on the quarter is the price of admission. In July, it was the revenue guide and another capex hike that sent the stock down 8%. As we tweeted at the time:
*META SEES FY CAPEX $130B TO $145B, SAW $125B TO $145B CDS limit up you say?
— zerohedge (@zerohedge) July 29, 2026
Muse: Show Us The Numbers
The reason the stock is where it is, of course, is Muse. And since the agent launched with only three weeks left in the quarter, all the data so far comes from third parties. Schilsky's ask is blunt: "Will we get any hard and fast Muse-related KPIs from Zuckerberg? DAUs? Downloads? Anything? We all know it's been a big success, but with that success comes a LOT more Capex and Opex . Investors need to have confidence in forward-looking user metrics in order to credibly underwrite Zuckerberg's investments."
The bull data: per Sensor Tower, Muse hit 10.9 million downloads since launch, 10.2 million of them in the US, making it one of only a dozen apps ever to reach 10 million US downloads in 40 days (the last one was Meta's own Threads, which, well, we all know how that ended). Deutsche Bank's Figure 1 shows Muse running far ahead of ChatGPT, Sora and Gemini on post-launch downloads, users and time spent, though still a rounding error vs. ChatGPT's 50 million US DAUs:
The bear data: as we detailed in "Peak Muse?" earlier this month, daily net user additions have slowed more than 60% from the September surge , Amazon blocked Meta's agents from its store within two weeks of launch, and banks are now following. Bloomberg's Dave Lee adds another : much of Muse's success "has come from sheer brute force," with almost 39% of Meta's entire marketing spend this month going to pushing Muse , per Sensor Tower. Nothing says organic growth like buying it.
And then there is monetization, which is still mostly a promise. Muse can make money "through subscriptions, a commerce take-rate model, interchange spread, and advertising on Muse's Feed tab," per Deutsche, and the $100-a-month "Maximum" tier we broke down in September is a start. But Schilsky warns that " Muse monetization may not be terribly meaningful in 2027 ; it's set to hopefully ramp in 2028." Which brings us to the number that actually matters.
The $455 Billion Question
Forget Q3. The debate on Meta is 2027 spending, and on that front the sellside is still playing catch-up with the buyside. Per Schilsky:
2027 opex: "While nearly everyone agrees that Street estimates of ~$203B are too low, I have heard a wide range of possibilities from $210B to $230B... expect a guide that looks like $210/$215B to $220/$225B – if we even get it this print."
2027 capex: "The Street's ~$200B remains too low. Most investors I spoke with are in the range of $225B to $250B... As for 2028 investor expectations…'higher'."
2026: the opex guide "has to be upped by +$10B on both the top and bottom ends due to META's recent AG settlement. But no one really cares." Capex should end near the high end of $130-145 billion, with the bottom of the range possibly lifted to $135 billion.
Deutsche's checks are in the same place: investors expect "$210-225bn in Opex, and $230-$240bn in CapEx next year," and the bank just raised its own 2027 capex forecast to $232 billion from $214 billion. Bloomberg's consensus has capex at about $140 billion this year, $202 billion in 2027 and $218 billion in 2028, with Meta projected to burn free cash flow until at least 2028 .
Since Meta can book data center spending either as capex (building it) or opex (leasing it from the likes of Crusoe), Schilsky says to watch the sum: " Buyside Consensus for 2027 Opex + Capex is in the range of $435B to $475B – that's a midpoint of $455B ." The Street's sum is $403 billion, a gap of more than $50 billion:
To put that in context, Goldman's internet team estimates Meta's AI-related capex across 2026-27 alone at $308.7 billion, part of a ~$1.73 trillion "Phase 2" buildout across the hyperscalers (and SpaceX):
As a reminder, Goldman calculated that this group needs ~$1.42 trillion of cumulative 2028-30 revenue to earn a 15% ROIC on 2026-27 AI capex, roughly $11.6 billion of annual revenue per gigawatt (a math problem we dissected in September ).
Here is the catch: Meta said last year it would stop giving next-year opex and capex guides on the Q3 call, so the most important number of the night may not show up at all. Schilsky thinks CFO Susan Li will at least repeat last year's language that capex dollar growth will be "notably larger" next year, while opex verbiage might flip to "grow at a slower rate in 2027 than 2026." And he explains why the opex number is the one that moves the stock: " Opex growth matters WAY MORE than Capex growth because investors care so MUCH MORE about EPS than FCF ." There are "real concerns" that the first 2027 opex guide, whenever it comes, knocks Street 2027 EPS of $34 down to as low as ~$32. The real bulls are already looking past it to 2028, with EPS estimates "in the low $40's vs. the Street's ~$39."
Translation: Meta is being priced on 2028 earnings from a product that won't make meaningful money until 2028, funded by spending that will hit 2027 earnings. What could go wrong.
Debt Or Equity? Pick Your Poison
Which brings us back to UBS's question. According to Schilsky, the cap raise is no longer a fringe idea on the buyside:
"Yes, a majority of investors still believe that an Equity Raise is likely post-Q3 , assuming the stock is high enough. (Certainly >$800. ~$850 would be 'better'.) We all know that META is going to burn FCF next year; the only question is whether the cost of debt or the cost of equity will be lower when they need to raise capital."
For context, $800 is about 13% above where the stock trades now and $850 is 20%. Which may explain the timing: a strong print tonight gets Meta into raise territory before Microsoft, Amazon and Alphabet report next week and potentially muddy the waters.
As for the cost of debt, it has been going up all year. The $27 billion bond funding Meta's Hyperion data center (the RPLDCI 6.581% of 2049, via the Project Beignet SPV), the largest corporate bond ever sold, hit a new lifetime low on Monday, at 91.5, down from 110 a year ago:
The rest of Meta's paper isn't faring much better. Earlier this month we noted that Goldman's prime brokerage data showed long-dated Meta paper among the most-shorted bonds in IG credit:
GS Prime Brokerage shows 9 of the 10 most-shorted US IG bonds are long-dated META, NVDA, ORCL, GOOGL and "SPCX" paper. CoreWeave's 9¼% 2030s are the #1 most-shorted HY bond. Oracle has the largest IG market value on loan.
— zerohedge (@zerohedge) October 1, 2026
So it's probably no surprise that, per Goldman's FICC desk, " entirety of credit market has their focus on META earnings ." And that's before counting what doesn't show up on the balance sheet at all: as Needham's Laura Martin calculated in early October, Meta has $628 billion in off-balance sheet commitments vs. $112 billion of on-balance sheet debt and leases. Add them to enterprise value, and the stock gets 35% more expensive. Regular readers know we have been warning since last October that the AI buildout runs on debt; Meta is now the test of how much more of it the market will take before it asks for equity instead.
And speaking of bills, there's a fresh one. Just hours before the print, a New Mexico judge ordered Meta to pay $25 billion after a jury found the company misled consumers about data privacy in the Cambridge Analytica case. The state had asked for $35-40 billion; Meta argued for a cap of $3.45 billion. We flagged it as soon as it hit:
*META ORDERED TO PAY $25B ON CAMBRIDGE ANALYTICA DATA BREACH: AP
— zerohedge (@zerohedge) October 21, 2026
That's on top of the ~$10 billion attorneys general settlement that Schilsky says is already headed for the 2026 opex guide (and which "no one really cares" about). We'll see if the market cares about two and a half times that. Meta will probably appeal, but for a company about to burn free cash flow, it's one more line in the funding plan.
Everyone Wants In, Nobody Wants The Bill
So here is the setup into tonight:
expectations at 9/10,
positioning at 8/10,
Mag 7 exposure at a record,
an implied move that Goldman calls cheap,
a buyside bar for 2027 spending that's $50 billion above the Street,
a product whose growth may already have peaked, and...
an equity raise that the majority of the buyside expects as soon as the stock lets it.
Bloomberg points out that at 19x forward earnings, Meta is the second-cheapest Mag 7 stock after Nvidia. Well, it probably should be.
The bull case is that Zuckerberg beats, guides Q4 to $75 billion, drops a Muse DAU number and keeps 2027 vague, and the stock grinds to $800 and the raise. The bear case is that he does all that and then says the words "notably larger" about 2027 capex. Judging by Meta's last two Q3 prints (-4% and -11% last October ), and the last two prints overall (-9% and -8%), we know which one the tape has favored lately: the market keeps paying Meta for products and charging it for the bill.
Schilsky's own closing question sums it up: "Can management keep the party going?" Maybe. But with the stock already 9% off the highs before the print, and Meta reporting a week early in what UBS suspects is a race to the equity window, we suspect the party is closer to last call. And the moment the cap raise lands, the bill arrives for everyone in the room.
Much more in the full JPM Tech Sketch " Earnings Preview Bonanza ", Goldman's " US Digital Ad Q3'26 Preview " and Deutsche Bank's " Meta – Buying Into The Muse-Hype ", all available to pro subs .
Tyler Durden
Wed, 10/21/2026 - 15:09