Meta Platforms is not in a correction. It is being reclassified — and the market hasn’t caught up.
The stock closed at $572.1 on April 1, 2026, down roughly 12% in a single month while the Nasdaq Composite sits 11% off its recent peak. The reflexive read is that META is oversold, a quality name dragged down by macro noise, crude oil near $110, and general risk-off sentiment. That read is comfortable. It is also wrong in a specific and important way.
The price chart tells you about the Nasdaq. It tells you nothing about what just happened in a U.S. courtroom.
The Section 230 Workaround No One Wants to Say Out Loud
A U.S. social media addiction trial has returned a verdict treating Meta’s recommendation algorithm as a defective product — not as editorial content, not as a neutral platform, but as a manufactured good with identifiable design flaws. The legal architecture matters enormously. Section 230 has historically functioned as a near-total liability shield by framing platforms as passive conduits for third-party speech. The “defective product” theory doesn’t argue with that framing. It walks around it entirely. If the harm comes from the algorithm’s design — the weight given to engagement, the feedback loops, the targeting mechanics — then Section 230’s content protections are simply irrelevant. You’re not suing over what someone posted. You’re suing over the machine that decided to show it to a 14-year-old at 11pm, repeatedly, until something broke.
Meta has spent two decades building its legal moat on Section 230 ground. That ground just shifted. The liability that comes with a “defective product” theory is not a fine or a consent decree. It is a per-plaintiff, scalable, potentially class-action-structured exposure that no one — not Meta’s lawyers, not its CFO, not the analysts covering it — can put a number on right now. Unquantified liability in a capital-intensive cycle is not a footnote risk. It is a ceiling on the multiple.
Australia and Indonesia have launched separate investigations into whether Meta violated protections for users under 16. The “defective product” logic, once established in one jurisdiction, travels. Plaintiff’s attorneys in other countries read American verdicts.
Burning the Furniture
While the legal architecture crumbles on one side, the financial architecture strains on the other. Per SEC filings, Meta’s capital expenditures hit $69.7 billion in 2025 — an 87.1% increase year-over-year from $37.3 billion in 2024. Against 2025 revenue of $201.0 billion, that puts the capex-to-revenue ratio at 34.7%, up from 22.6% the prior year. R&D spending reached $57.4 billion according to the same filings, a 30.8% increase, pushing that ratio to 28.6% of revenue.
Add those two lines together and Meta spent roughly 63 cents of every revenue dollar on building and researching in 2025. Revenue grew 22.2%. The cost of generating that revenue grew at nearly four times that rate on the infrastructure side alone.
The macro is a meat grinder. The power infrastructure Meta is vertically integrating in Louisiana — designed to bypass grid constraints — is a genuine strategic move, but it is also a confession that the marginal cost of AI compute is now so structurally high, and so exposed to energy prices, that the company has decided to become its own utility. At $110 crude, that bet carries its own set of risks that S&P Global has flagged directly. A high-margin software business does not build power plants. A capital-intensive infrastructure company does. The multiple compression that follows that transition is arithmetic. Is this a visionary pivot, or a legacy giant throwing billions at a ghost?
The analyst community has noticed the capex surge. Most price targets sit somewhere around twice the current trading price. The market is laughing in their face. That gap — the distance between where the stock trades and where the consensus says it should trade — is not insight. It is a hallucination maintained by people who are structurally incentivized to be wrong. Buy-side pressure, deal flow relationships, the simple career reality that a bullish call on a beloved mega-cap is a safer error than a bearish one. The 100% price-target gap on META is not a contrarian opportunity. It is a reason to read the footnotes more carefully than the headline figure.
Meta has positioned its “Responsible AI” frameworks and content moderation infrastructure as a selling point to enterprise customers — a signal to nervous CIOs that the platform is safe to build on. That positioning is a marketing placebo. It does nothing to address the “defective product” theory, which is not about whether Meta has a content policy. It is about whether the algorithm’s core design — the engagement-maximizing architecture that is also Meta’s primary revenue mechanism — constitutes a foreseeable harm. You cannot safety-label your way out of a product liability case.
The silent variable underneath all of this is what insurance markets are beginning to call “algorithmic duty of care” — a standard being developed not by regulators first, but by underwriters. If insurers start pricing AI-driven recommendation engines the way they price chemical plants or pharmaceutical compounds, the compliance cost structure for every social platform changes permanently. This is not in any current estimate I can find. It is not in Meta’s filings. It is a direction of travel rather than a line item — but the direction is not ambiguous.
The weakest assumption in the bear case is that the “defective product” ruling survives appeal and becomes a template rather than an outlier — that hasn’t been established yet. The advertising business remains durable, the user numbers are extraordinary, and the AI infrastructure buildout could generate returns that justify the spend if the legal environment stabilizes and energy costs don’t compound.
The capex risk is priced in; the consensus trade is to buy the dip and wait for the multiple compression to reverse. The legal risk is not. That trade rests on the assumption that the “defective product” ruling is an isolated event rather than an opening brief for an entirely new liability category. The next verdict will test that assumption. The one after that will price it.
$572.1 is not obviously cheap when the product itself might be what’s on trial.
Meta’s entire legal defense for two decades was essentially “we’re just the pipe, don’t blame the pipe.” Now the pipe has an algorithm, the algorithm has a design, the design has victims, and the victims have lawyers. Turns out the pipe was a product the whole time — they just forgot to read their own terms of service.