Traders just piled into a whimsical Manifold market asking whether “an AI will do Magic before EOY 2027.” The contract sits at 19% implied probability of YES as of September 12, 2026, and roughly two-thirds of its entire lifetime volume changed hands in the past 24 hours. The trigger was not a stage magician. It was a week in which OpenAI claimed a Millennium Prize math result, Sam Altman publicly agreed to “try” for a room-temperature superconductor, and four frontier models shipped inside days of each other.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What the market actually asks
The contract, created by a user named calour on Manifold Markets, is titled “Will an AI do Magic before EOY 2027? [See description].” Despite the name, it has nothing to do with card tricks or illusions. The “[See description]” tag is doing the heavy lifting. The resolution text lists a menu of science-fiction-grade feats that would count as “Magic,” including room-temperature superconductors, factoring RSA-2048, mind uploading, raising human intelligence by 15 IQ points, reversing aging, transmuting lead to gold, curing depression in 95% or more of cases, building a portable fusion reactor, or “solving alignment.” The creator caps the list with a catch-all: anything that “clearly exceeds the Magicalness bar,” resolved on an “I’ll know it when I see it” basis.
In other words, this is a subjective, tongue-in-cheek market about whether artificial intelligence produces a genuinely world-altering breakthrough before the end of 2027. It is closer to a cultural barometer than a precise forecasting instrument, which is exactly why it is worth reading when it moves.
The odds, and what 19% implies
As of September 12, 2026, the market prices YES at 19% on Manifold. On a prediction market, the trading price is read as a rough implied probability: a 19% line means the pool of traders collectively thinks there is a little less than a one-in-five chance that an AI clears the “Magic” bar before EOY 2027. That signal carries more weight than it once did: industry-wide prediction-market volume jumped from $15.8 billion in 2024 to $63.5 billion in 2025, according to a July 2026 report from Skynet, so a 19% line now reflects a far deeper and more liquid pool of traders than a comparable contract would have a year earlier. Manifold runs on play money called mana rather than cash, so the dollars are not real, but the incentive structure that pushes prices toward the crowd’s best guess still applies. For readers new to that mechanic, our primer on how model comparisons and benchmarks get framed in Llama 4 vs DeepSeek V4 vs GPT-5.6 covers how capability claims tend to get stress-tested.
The more striking number is the velocity. The market reports that about 68% of its all-time volume traded in a single 24-hour window, placing it among the most active contracts on the platform that day. On a thinly traded novelty market, a burst like that usually signals that a news event, not a slow drift in sentiment, pulled money in.
What set off the sudden bet
The cause is not a single announcement but a cluster of them in early September 2026, each of which maps onto an item in the market’s “Magic” list. No verified event has resolved the market YES, and none is close to doing so. What changed is the perceived trajectory. Below is the week that moved the line.
| Date (2026) | Event | Maps to market criterion |
|---|---|---|
| Sep 1-3 | Four frontier models ship within days: Claude Fable 5.1, Gemini 3.8 Flash, Meta’s Muse Spark 1.3, and OpenAI’s GPT-6 Astra | General capability jump |
| Sep 8 | OpenAI says an internal model helped produce a proof related to the Navier-Stokes Millennium Prize Problem | “Solving” an open problem |
| Early Sep | Sam Altman replies “you know what, let’s try” to a dare about chasing a room-temperature superconductor | Room-temperature superconductors |
| 2026 (ongoing) | AI-assisted screening surfaces two new superconducting materials, YRu3B2 and LuRu3B2 | Room-temperature superconductors |
| 2026 | RSA-260 (an 862-bit modulus) is factored; RSA-2048 remains far out of reach | Factoring RSA-2048 |
The superconductor banter
The most direct spark came from OpenAI chief executive Sam Altman. After a satirical post floated a rumor that a rival lab had quietly found a room-temperature superconductor, Altman replied “you know what, let’s try,” as reported by BeInCrypto. The remark was banter, not a roadmap, and no lab has announced any such material. But a room-temperature superconductor is an explicit item on the market’s “Magic” menu, and a sitting frontier-lab CEO treating it as a near-term target is the kind of signal that nudges a speculative crowd.
The underlying science is moving, if slowly. An international consortium led by Aalto University reported using machine learning combined with quantum-geometry methods to screen candidate materials faster, work that has already surfaced two new superconductors, per Aalto University. Those materials are not room-temperature superconductors, and the consortium’s stated goal runs to 2033, well past this market’s EOY 2027 horizon. That gap between the headline and the fine print is the recurring theme of this whole episode.
The Navier-Stokes claim and the credit fight
On September 8, OpenAI said a set of agents running on a next-generation model helped produce a proof related to the Navier-Stokes equations, one of the seven Clay Mathematics Institute Millennium Prize Problems, each carrying a $1 million reward, as documented on OpenAI’s own write-up. The company said it was not claiming the prize and that the wider mathematical community would need to verify the work.
The announcement was quickly overshadowed by a credit dispute. As Axios reported, New York University mathematician Tristan Buckmaster, who had been working toward a related result with Anthropic-affiliated mathematician Levent Alpoge, pushed back on how the contribution was framed, noting his team had solved a forced Euler variant that is a step toward, not a full resolution of, Navier-Stokes. For the Manifold crowd, the nuance mattered less than the optics: an AI system credibly linked to progress on a century-defining math problem reads like the “Magic” list coming to life, even if the proof is still under review.
Who and what is involved
The players behind the move are the major AI labs and the researchers checking their claims. The model launch burst, tracked by Artificial Analysis, put new flagships from Anthropic, Google DeepMind, Meta, and OpenAI into the market within a single week, with six labs now fielding a model above 50 on the Artificial Analysis Intelligence Index. Several shipped twice under split access policies, a generally available version and a gated variant for vetted security or research organizations. Anthropic’s new flagship is covered in our walkthrough of the Claude Fable 5.1 API, and the competitive backdrop across assistants shows up in our look at the Gemini Windows integration gap.
On the other side are the mathematicians and materials scientists whose verification work will determine whether any of these claims actually clear a bar. That tension, between lab marketing and independent confirmation, is the same dynamic that shaped coverage of OpenAI-linked security incidents earlier this cycle.
Why a joke market matters to a tech and finance reader
Novelty markets like this one are not investment signals, and the mana at stake is not money. Their value is as a sentiment gauge, and that gauge is reading a much bigger crowd than it used to: real-money rival Polymarket logged 678,342 unique users in April 2026 alone, according to data Luvkaizen reported in May 2026, underscoring how mainstream prediction markets have become even as Manifold’s version stays play-money only. When a subjective “will AI do something miraculous” contract sees two-thirds of its volume arrive in a day, it is a compact readout of how a technically literate crowd is reacting to a news cycle. The read here is ambivalence: even after a Millennium Prize claim, a CEO’s superconductor dare, and a four-model launch week, the crowd still prices the “Magic” threshold at just 19% by EOY 2027. Enthusiasm about capability is real, but skepticism about verified, world-changing results remains the dominant position.
That split matters for the broader AI trade. The same companies generating these headlines are the ones whose valuations and funding rounds, such as the raise detailed in our Mistral AI valuation coverage, depend on the gap between demonstrated progress and promised breakthroughs. A market that refuses to price “Magic” above one in five is, in its own informal way, repricing that gap in real time.
What to watch next
Three things could move this line again before 2027. First, independent verification of the Navier-Stokes proof: if the Clay Mathematics Institute or the broader math community confirms a full solution, expect a sharp repricing. Second, any credible materials result that edges toward a room-temperature superconductor rather than a faster screening method. Third, the next frontier-model launch cycle, since each capability jump feeds the narrative even when it clears no specific item on the list. Absent a concrete, verified breakthrough, the most likely path is continued drift driven by headlines rather than resolutions.
Frequently asked questions
Is this market about AI performing magic tricks? No. Despite the title, the resolution criteria define “Magic” as a list of major scientific and technological feats, such as room-temperature superconductors or factoring RSA-2048, plus a subjective catch-all judged by the market creator.
What are the current odds? The market prices YES at 19% implied probability as of September 12, 2026, on Manifold Markets. Prediction-market prices are read as rough probabilities, not guarantees.
Has any AI actually met the bar? No. None of the early-September 2026 events, including OpenAI’s Navier-Stokes claim, has resolved the market YES, and the superconductor and RSA-2048 items remain unmet. The claims are also still under independent review.
Is real money involved? No. Manifold uses play-money mana rather than cash, a detail Tech Insider confirmed in its September 2026 reporting on this market, so it functions as a sentiment gauge rather than a financial instrument. Real-money rivals like Polymarket are a different story: they run on wallet infrastructure from providers such as Magic Labs, which had provisioned more than 60 million wallets by July 2026, according to Architect Partners.
The Bottom Line
Sources
- Manifold Markets — Will an AI do Magic before EOY 2027? [See description]
- Axios — OpenAI’s historic math solution overshadowed by credit controversy
- OpenAI — On the Navier-Stokes Millennium Prize Problem
- BeInCrypto — Is Sam Altman Serious About a Room-Temperature Superconductor? He Said ‘Let’s Try’
- Aalto University — AI accelerates the search for a room-temperature superconductor
- Artificial Analysis — Four frontier launches in eight days
Prediction markets carry risk and are not investment or betting advice. Market availability is restricted by jurisdiction: Polymarket is not available to US persons, while Kalshi is a CFTC-regulated US exchange. Manifold Markets operates with play-money mana. The wallet infrastructure underpinning real-money venues has scaled fast: Magic Labs said it had served more than 200,000 developers across over 18,000 applications by July 2026 and had supported $8.9 billion in cumulative Polymarket transaction volume through November 2025, per a September 2026 Architect Partners report. Participation is limited to those 18+ or 21+ as applicable in your jurisdiction. If gambling is a problem for you or someone you know, call 1-800-GAMBLER for confidential help.
Related Coverage
- Millennium Prize Problems Bet: Manifold Odds Jump to 24% on OpenAI Claim
- Millennium Problems Bet: Manifold Prices 99% Odds Two Fall by 2040 After OpenAI's Navier-Stokes Claim
- Superconductor Bet Heats Up: Manifold Traders Give a Room-Temperature Breakthrough 15% Odds Before 2031
- Manifold Traders Push AI Masters Minecraft by 2035 Market to 95%
- Polymarket's LAPTOP $1B FDV Market Swings +24 Points After Hunter Biden Token's Wild Debut
