There's a number that should sit at the top of any memo about a new Meta platform, and it isn't a user count. It's the cumulative operating loss Meta's Reality Labs division has racked up since the company began breaking out the segment in 2020 — north of $46 billion as of the company's own filings through 2023, and still climbing by roughly the cost of a mid-sized acquisition every quarter. That figure is the proper context for the latest piece of Meta platform development to surface in the press: a reported effort to build a prediction-market product, where users wager on the outcome of future events.
Hold the number in your head, because the temptation with every new Meta experiment is to evaluate it on its own terms. Don't. Evaluate it on the company's terms — which are, demonstrably, a willingness to spend at a scale that would bankrupt anyone else, for years, on bets that may never clear.
What the $46 billion actually measured
Reality Labs is the augmented- and virtual-reality segment: Quest headsets, the Ray-Ban smart glasses, the broader "metaverse" infrastructure. The reported losses are operating losses for that segment — revenue minus the cost of building and selling those products, including a very large R&D line. It is a clean, auditable number because Meta is required to disclose it, and it has become the single most quoted piece of evidence that the company tolerates enormous, sustained losses in service of a platform it hasn't proven anyone wants yet.
What that figure captures well: the company's appetite for long-horizon, capital-intensive platform construction. What it captures precisely is one division's bet. And that precision is exactly the problem when you try to use it to reason about a prediction-market app.
What the number doesn't measure
The $46 billion is Reality Labs. It does not include the graveyard of experimental platforms that lived and died outside that segment — and that graveyard is the more relevant data set for a prediction-market launch.
Consider the pattern. Libra, later Diem, the global stablecoin project that regulators strangled before it shipped a single transaction. Bulletin, the Substack-style newsletter platform, shut down after about a year. Facebook Sports Stadium, a second-screen sports product, quietly retired. The NFT and digital-collectibles integration across Instagram and Facebook, announced with some fanfare and wound down inside roughly a year. Each of these was an attempt to extend Meta into a new transactional or content vertical. Each was cheaper than Reality Labs and arguably more disposable.
So the honest read is that the $46 billion measures Meta's tolerance for a specific kind of loss — the slow, board-approved, multi-year kind. It tells you almost nothing about how long the company will support a smaller experimental platform that touches money, prediction, and the attention of regulators. Those have a much shorter half-life.
Why prediction markets sit in a contested zone
A prediction market lets people buy and sell contracts tied to whether something happens — an election outcome, a Fed rate decision, whether a film clears a box-office threshold. The price of a contract reads as an implied probability. Proponents argue these markets aggregate information better than polls. Skeptics, with some justification, note that a contract paying out on a real-world event is difficult to distinguish from a bet.
That ambiguity is the regulatory story. In the United States, event contracts have generally fallen under the Commodity Futures Trading Commission, while anything that reads as wagering risks colliding with state-level gambling law, which varies enormously jurisdiction to jurisdiction. The recent history is instructive without being settled: platforms in this space have spent years in litigation and negotiation with the CFTC over which event contracts are permissible, and the lines have moved more than once. Treat any specific ruling as a snapshot, not a rule — the regime is actively contested as of writing.
For Meta, that contested zone is the entire risk. A company carrying the regulatory scar tissue of Libra knows what it looks like when financial regulators decide a tech platform is moving into their domain uninvited. The prediction-market category invites three sets of regulators at once: commodities, gambling, and the consumer-protection apparatus that watches how platforms market anything wager-shaped to a billion-plus users.
The case for doing it anyway
The cynical reading writes itself: another Meta platform, another vertical chased after a competitor proved the engagement, destined for the shutdown blog post in eighteen months. That reading is probably correct more often than not, and the failure record supports it.
But there's a less cynical version worth holding alongside it. Prediction markets generate two things Meta values structurally. The first is high-intensity, recurring engagement — people check positions the way they check scores, repeatedly, with emotional stakes. The second is intent data of unusual quality: a user putting money on an outcome reveals far more than a user clicking a poll. For an advertising business, knowing what people will actually pay to be right about is a genuinely valuable signal, separate from whether the market itself ever turns a profit.
That reframes the regulatory exposure. Meta may not need the prediction market to be a clean, profitable business. It may only need it to run long enough, in permissible enough jurisdictions, to learn something the rest of the platform can use. Which is a very different — and more durable — reason to spend than the one the failure record predicts.
What to do with this before your next update
Here is something concrete to try this week, modest enough to be honest. Pull Meta's segment reporting and find the line where this experiment would actually land. A prediction-market product is almost certainly not Reality Labs; it would sit inside Family of Apps or be folded into "other" R&D, where it will be far harder to track and impossible to isolate. So set a single tripwire: note today's Family of Apps operating margin, and flag in your model the quarter Meta first acknowledges a wagering or event-contract product in an earnings call or a CFTC filing. The gap between announcement and that first regulatory acknowledgment is your real signal — not the press leak, and not the $46 billion that taught everyone to look at the wrong number.
Not sure which tool to use?
Compare the top AI music and sound tools side by side — honest reviews, real pricing, no sponsorships.
