Things started getting interesting for MetaMETA-- two weeks ago. The company printed Q2 revenue up 28% year-over-year to $60.80 billion — the ad machine still humming, with ad impressions up 14%, average price per ad up 12%, and 3.60 billion people opening at least one of Instagram, WhatsApp, Facebook or Messenger daily. And yet operating income fell 8% to $18.78 billion, GAAP diluted EPS fell 13% to $6.18, and the operating margin compressed from 43% to 31%. Revenue beat. Earnings missed. Free cash flow collapsed to $784 million from $8.5 billion a year ago because capital expenditure alone ate $31.1 billion in the quarter.
The official explanation was clean enough. Two "non-recurring" items did the damage: a $2.40 billion charge related to legal proceedings, and $1.18 billion of severance from the May 2026 headcount reduction of roughly 8,000 employees. Strip both, CFO Susan Li said, and operating income would have risen 9% year-over-year. A simple pre-tax EPS add-back of those disclosed charges implies roughly $7.58 per share; Meta did not publish a separate non-GAAP EPS figure in the release or call, so I treat that number as an illustrative bridge rather than a reported result.
Q2 2026 GAAP operating income of $18.8B becomes $22.4B on the disclosed-charge add-back; a simple pre-tax EPS bridge is about $7.58, while reported diluted EPS was $6.18.
Sources: Meta Q2 2026 results, earnings call transcript, and Meta Q2 2025 results.
| period | GAAP operating income (B) | Comparable add-back operating income (B) | GAAP diluted EPS | Illustrative pre-tax EPS add-back |
|---|---|---|---|---|
| Q2 2025 | 20.441 | 20.441 (same under narrow add-back) | 7.14 | 7.14 (same under narrow add-back) |
| Q2 2026 | 18.775 | 22.355 | 6.18 | 7.58* |
* Calculated as $6.18 + ($2.40B + $1.18B) / 2.566B diluted shares; tax effects are not modeled.
Walk that bridge and you arrive at consensus 2026 EPS of $32.81, raised from $29.65 pre-print — even as the consensus price target was cut from $855 to $827, with the cut attributed to higher legal expenses. Read those two revisions together: analysts raised the EPS they are modeling but cut the multiple they will pay for it. That is the contradiction the Oakland trial is about to resolve. Consensus 2027 EPS sits at roughly $33, still embedding a legal run-rate well below the Q2 print. Both numbers assume the legal line normalizes toward zero. That assumption is the trade.
Yes, Meta could keep rallying from here. The ad business is genuinely accelerating, AI-driven recommendation improvements are showing up in the price-per-ad print, and the bull case says Reality Labs is one Connect event away from a re-rating. I am not interested in disputing the bull case on its own terms. I am interested in the plumbing underneath it — and the plumbing says the legal line is not a one-off, it is a serial cost generator, and the moment sell-side models have to mark it as recurring the multiple compresses.
Last Thursday — August 6 — New Mexico First Judicial District Judge Bryan Biedscheid ordered Meta to create a $567 million abatement fund to address youth mental health harms. The bulk, $420 million, funds treatment services for young people; the remainder goes to awareness, prevention, and screening over the next five years. That ruling sits on top of the $375 million in civil penalties a Santa Fe jury imposed in March after finding Meta liable for 75,000 willful violations of New Mexico's Unfair Practices Act. New Mexico alone is now $942 million.
New Mexico's $942M combined rulings and Q2's $2.4B legal charge are the known dollar figures. The Oakland case carries a reported theoretical claim ceiling of about $1.4T for four state-law plaintiffs; Tennessee remained unquantified, and the coordinated actions covered 41 states plus Washington, D.C.
Sources: New Mexico DOJ, Associated Press, Meta Q2 2026 results, Oakland court filing, Reuters, and New York AG.
| case / scope | amount or status | data note |
|---|---|---|
| New Mexico civil penalties (March 2026 jury) | $375M | Jury award |
| New Mexico abatement fund (Aug 6, 2026 ruling) | $567M | Court-ordered fund |
| New Mexico total (civil + abatement) | $942M | Calculated: $375M + $567M |
| Q2 2026 legal proceedings charge | $2.4B | Meta disclosed charge |
| Oakland federal MDL (29 states) | Not separately quantified | No standalone public dollar estimate for the 29-state COPPA claim |
| Oakland four-state state-law claims | ~$1.4T theoretical ceiling | Plaintiff calculation reported in court filings/media; not an expected loss |
| Tennessee state trial | Not publicly quantified | Trial ongoing as of August 7, 2026 |
| Coordinated child-safety actions | 41 states + Washington, D.C. | 42 AGs / jurisdictions in the 2023 actions; non-dollar count |
The dollar figure understates the structural piece. Judge Biedscheid ordered Meta to build banner and informational screens explaining its protection features, to continue improving AI-based age assurance, to attempt developing a dedicated under-13 prediction model within two years, to partner with schools on a reporting portal, and to file semiannual compliance reports. Those are engineering and compliance costs that sit in operating expense, not in a one-time legal line. Meta said it will appeal. No stay pending appeal has been reported, which means the compliance clock is already running.
New Mexico is one case. The pipeline behind it is what makes the legal line recurring rather than transitory. At the coordinated October 2023 filing, 41 states plus Washington, D.C. — 42 attorneys general or jurisdictions — took action against Meta. The current Oakland proceeding includes 29 states' federal COPPA claims alongside four states' state-law claims; Tennessee is a separate state-court trial. Meta itself warned in the Q2 release: "we have a number of youth-related trials scheduled for this year in the U.S., which may ultimately result in a material loss." That is not my characterization of the risk. That is management's.
Understanding what I understand about how analyst models actually work tells me the next leg of this trade is not the verdicts themselves — it is the basis shift. Consensus 2026 EPS of $32.81 is built on the assumption that the Q2 legal charge does not repeat. The $2.40 billion legal charge is, in effect, added back to forward estimates. Once verdicts cluster — New Mexico $942 million, Oakland MDL pending, Tennessee pending — the legal charge can no longer be flagged as non-recurring. At that point sell-side models have to shift their valuation basis from adjusted EPS to GAAP EPS.
The arithmetic is unforgiving. GAAP diluted EPS for Q2 was $6.18. A simple pre-tax add-back of the two disclosed charges is about $7.58, but that is an illustrative calculation rather than a reported adjusted EPS. The roughly $1.40-per-share bridge is the dollar size of the "one-time" assumption on that narrow basis. Same stock price. Lower earnings basis. Higher multiple. That is what multiple compression looks like when the basis shifts, not when the price moves.
The positioning tells you the market has not priced this. META closed August 6 at $589.90 — down 10.63% year-to-date, down 23.4% on a rolling annual basis, trading below both its 50-day simple moving average ($600.62) and its 200-day ($631.99). RSI 14 sits at 47.0. MACD DIFF is deeply negative at -8.16. The options market is the more telling tell: per Ainvest data, average implied volatility of 37.4%, put/call volume ratio of 0.32, put/call open interest ratio of 0.44. That is a complacency structure, not a fear structure. There is no put wall being defended.
The capital-flow data is the part that should make you nervous if you are long. Per Ainvest data, large-order outflow was $146.8 million against $117.8 million inflow — institutional distribution, not accumulation. Block flow was a wash. Retail inflow was $542.7 million against $528.7 million outflow — net positive $14 million. Same stock. Same verdicts. Different impact — because the smart money is quietly handing shares to the retail money while the legal pipeline is still being framed as a one-off. That is the kind of divergence I check before I check anything else.
Conditional chain. If the Oakland MDL — 29 states — produces a verdict in the same neighborhood as New Mexico, the legal line stops being add-back-able and consensus 2026 EPS revises toward the GAAP print. If Tennessee lands first, same mechanism, same re-rating, just earlier. If both land before year-end, the basis shift happens inside the Q3 or Q4 print and the stock has to reprice against a higher multiple on lower earnings. If, on the other hand, Meta wins Oakland or settles cheaply, the adjusted-EPS basis holds and the bull case survives the plumbing check. That is the break condition. I am not making a price call. I am telling you which mechanism is now in motion and what would invalidate it.
What to watch, in order. Oakland MDL jury selection the week of August 10 — the speed of jury selection tells you whether the trial starts on schedule. Tennessee verdict timing. Any analyst note that moves the valuation basis from adjusted EPS to GAAP — that is the tell that the plumbing has shifted. The put/call ratio climbing off 0.32 — first sign the option market is starting to price legal-event risk. And whether META dealer gamma flips negative, because once it does, the mean-reversion regime that has been suppressing vol breaks and bad news starts to amplify instead of fade.
Same plumbing. Same verdicts. Different basis. That is the trade.
Comments
No comments yet