Short Answer

P vs NP is likely to be resolved before 2035, driven by increased expert and market sentiment following recent AI breakthroughs in complexity theory from OpenAI and Anthropic, with the market pricing this at 75.0%.

1. Market Behavior & Drivers

The market's probability of a pre-2032 resolution for P vs NP surged 45 percentage points on November 20, 2026. The move from 6.0% to 51.0% was driven by widespread news reports that an Anthropic team had completed a formal, machine-checked proof of the Riemann Hypothesis. While not a direct resolution of P vs NP, traders interpreted this reported AI-assisted breakthrough in a famously difficult mathematical problem as a strong positive indicator for the likelihood of a similar result for P vs NP.
This sharp repricing also appears to be a delayed reaction to earlier developments. In late October 2026, both OpenAI and Anthropic announced separate, AI-discovered proofs that refuted the Exponential Time Hypothesis. The market had traded below 10% in the weeks following those announcements. The subsequent news regarding the Riemann Hypothesis likely served as a catalyst, forcing the market to price in the combined impact of multiple recent AI advances in computational complexity and formal mathematics. The current price of 46.0% is slightly below the peak, which may reflect the fact that none of the announced proofs have completed formal peer review.
  • Resolution by 2035 appears likely, driven by recent AI complexity theory breakthroughs.
  • Formal acceptance before 2029 faces multi-year peer review and CMI waiting periods.
  • Expert consensus indicates high probability for AI-assisted Millennium Prize solutions by 2030.

Who Wins and Why

Outcome Market Model Why
Before 2027 4.0% 3.0% Peer review and formal recognition typically take years, making resolution before 2027 highly improbable.
Before 2028 46.0% 33.5% Strict criteria, including a two-year CMI waiting period, make resolution before January 1, 2028, unlikely.
Before 2029 58.0% 43.7% Two-year CMI waiting periods and journal bans make formal acceptance by January 1, 2029, tight.
Before 2030 73.0% 63.1% This timeframe aligns with expert sentiment for AI solving Millennium Problems by 2030.
Before 2035 75.0% 70.4% Growing optimism regarding AI's capabilities provides ample time for proof and recognition.

Current Context

AI advances in October 2026 significantly impacted P vs NP prediction markets. On October 26, 2026, OpenAI announced an AI-discovered proof refuting the Exponential Time Hypothesis (ETH) for k-SAT, with Anthropic posting a near-identical refutation hours later; both proofs were dated to discovery on October 16, 2026, and included Lean formalizations [1][2][3]. The involved verifying team explicitly stated their role was not certification, and neither proof is peer-reviewed [1][2][3]. The ETH refutation does not settle P vs NP; its 2^o(n) runtimes are subexponential but not polynomial, and the algorithm applies only to k-SAT, not NP in general [1][2][3]. Anthropic's paper indicated the speedup is unlikely to surpass brute force until inputs exceed 10^7 variables [1][2][3]. These developments, alongside AI disproving the 3SUM conjecture (Oct 5) and achieving sub-n log n integer multiplication (Oct 6), influenced markets [4]. Kalshi's P vs NP resolution market (KXPNP), as of November 20-21, 2026, saw a 40-47 point jump in implied probabilities for resolution within 24 hours of the ETH news [5][6][7][8]. Implied probabilities for resolution before 2027 were 3-4%, before 2028 were 6-48%, before 2029 were 10-58%, before 2030 were 23.5-69%, before 2035 were 36.5-76%, and before 2040 were 45-57% [5][6][7][8]. An outcome before 2032 interpolated to roughly 40-55% [5][6][7][8].
Expert opinion diverges on AI's potential to solve Millennium Prize Problems. A LEAP expert survey, conducted October 6-27, 2026, with 143 experts, 49 superforecasters, and 133 AI risk experts, found a median 80% probability that AI will solve or substantially assist one of the remaining Millennium Prize Problems (excluding Navier-Stokes) by end-2030, compared to 20% by end-2026 [9][10]. Superforecasters estimated 85% by 2030, while the public's median was 44% [9][10]. For at least three of the five remaining problems by 2030, expert and superforecaster medians were 45-50% [9][10]. Emanuele Viola, writing on November 17, 2026, in a CACM blog, suggested the October AI results support his belief that P=NP, anticipating further AI-driven conjecture resolutions [4]. In contrast, Lance Fortnow (Computational Complexity blog, June 10, 2026) and Stephen Wolfram (Jan 30, 2026) expressed skepticism that a P vs NP proof would emerge, citing barriers such as relativization, natural proofs, algebrization, and computational irreducibility [11][12]. The Kalshi contract defines resolution as a peer-reviewed article presenting a complete proof, disproof, or counterexample in the Clay Mathematics Institute's original formulation, or written recognition by an authoritative body [6][7]. Clay Millennium Prize rules require a qualifying refereed publication, a two-year waiting period, and general community acceptance, but do not address AI-originated proofs [13].
Despite many past claims, formal recognition for P vs NP remains elusive. The P versus NP Register, as of July 24, 2026, recorded 159 claimed solutions, including 116 historical dossiers and 43 post-2016 entries, with seven appearing in 2025 alone [14]. However, 86.2% of historical claims were never adjudicated [14]. Indexed P vs NP works increased from 369 in the prior full decade to 2,741 in the current partial decade [14]. As of September 2026, FAIR Labs found no AI system claiming progress on P vs NP, Hodge, BSD, or Yang-Mills problems; claims were limited to Navier-Stokes (Sep 2026) and partial Riemann hypothesis bound improvements [13]. A Clay Mathematics Institute workshop on 'P vs NP and Complexity Lower Bounds' occurred September 29-October 3, 2025 [15]. A February 2026 CACM interview with a complexity researcher noted the problem had stood for over 50 years with the field "no closer to an answer than we've ever been" prior to the October 2026 AI developments [16].
Sources (16)
  1. 1Computational Complexity: October 2026blog.computationalcomplexity.org
  2. 2AI cracks a 27-year-old pillar of computer science, OpenAI and Anthropic claim | Science | AAASscience.org
  3. 3OpenAI's Kakeya, ETH and Yang–Mills claims: what they say and what stays open — Papexpapex.org
  4. 4What Happened in Fall 2026: AI and P vs NP – Communications of the ACMcacm.acm.org
  5. 5When will P vs NP be resolved? Odds & Predictions 2031kalshi.com
  6. 6When will P vs NP be resolved? | PredictPediapredictpedia.com
  7. 7When will P vs NP be resolved odds: Before 2035 76% on Kalshipredictdesks.com
  8. 8When will P vs NP be resolved? - Kalshi KXPNP-32 Odds | CoinRithmcoinrithm.com
  9. 9Wave 13: Short-Term Cruxes | LEAPleap.forecastingresearch.org
  10. 10Probability AI Solves Another Millennium Prize Problem | LEAPleap.forecastingresearch.org
  11. 11P vs. NP and the Difficulty of Computation: A Ruliological Approach—Stephen Wolfram Writingswritings.stephenwolfram.com
  12. 12Computational Complexity: Respect the P v NP Problemblog.computationalcomplexity.org
  13. 13The Proof Machines. FAIR Labs Frontier Series Special Reportfairlabs.ai
  14. 14Home · The P versus NP Registercorroboration.org
  15. 15P vs NP and Complexity Lower Bounds - Clay Mathematics Instituteclaymath.org
  16. 16Addressing the Complexity of Computation | Communications of the ACMdl.acm.org

2. Price Chart

Historical Price (Probability)

Outcome probability
Date

3. Significant Price Movements

Notable price changes detected in the chart, along with research into what caused each movement.

📉 November 21, 2026: 17.0pp drop

Price decreased from 75.0% to 58.0%

Outcome: Before 2029

What happened: The claimed 17.0-percentage-point drop for the "Before 2029" outcome on November 21, 2026, could not be verified from the provided sources, which show inconsistent market data or drops for different deadlines and dates [1][2][3]. While recent AI developments sparked discussions in October 2026 regarding complexity theory, specifically refuting the Exponential Time Hypothesis (ETH) rather than P vs NP, the research does not identify specific social media posts or direct evidence linking this activity to the alleged price drop [4][5]. Therefore, a primary driver for the stated market movement, including any social media influence, cannot be determined from the available information. Social media's role in this specific, unverified price drop is irrelevant due to lack of confirmation.

📈 November 20, 2026: 64.0pp spike

Price increased from 11.0% to 75.0%

Outcome: Before 2029

What happened: The primary driver was a social media and news wave on November 20, 2026, which reported that Anthropic mathematician Levent Alpoge and AI system Claude completed a Lean-checked proof of the Riemann Hypothesis [6][7]. This widely circulated claim, despite not directly addressing P vs NP, spurred a general market belief that AI was rapidly solving Millennium Prize Problems, repricing the "When will P vs NP be resolved?" market upwards [6][8]. This social media and news activity served as a primary driver, creating a halo effect that accelerated expectations for other complex mathematical challenges to be resolved soon [6][9].
Sources (9)
  1. 1When will P vs NP be resolved? Odds & Predictions 2031kalshi.com
  2. 2When will P vs NP be resolved? - Kalshi KXPNP-32 Odds | CoinRithmcoinrithm.com
  3. 3Will P vs NP be resolved before 2028? | Forecasting Agentforecast.agenticlearning.ai
  4. 4AI cracks a 27-year-old pillar of computer science, OpenAI and Anthropic claim | Science | AAASscience.org
  5. 5What Happened in Fall 2026: AI and P vs NP – Communications of the ACMcacm.acm.org
  6. 6Iran War Day 266 Live: Second Chance: The AI Trade Breaks Into A Correction, The | Mr. Robert Intelligence Briefmrrobert.ai
  7. 7Tech — prediction-market news | pdatapdata.world
  8. 8What's Moving on Kalshi Today: Volume and Big Tradesbotforkalshi.com
  9. 9Prediction Market Movers: Biggest Odds Changes Today | PredictionBubblespredictionbubbles.net

4. Market Data

Contract Snapshot

This market resolves to "YES" if the P vs NP problem is officially resolved by a peer-reviewed proof or disproof, or confirmation from a recognized math body, before the specified deadline for that market. It resolves to "NO" if no such resolution occurs by that deadline. The key deadlines are December 31st of the year preceding the market's stated year (e.g., December 31, 2027, for the "Before 2028" market).

Available Contracts

Market options and current pricing

Outcome bucket Yes (price) No (price) Last trade probability
Before 2027 $0.05 $0.97 4%
Before 2028 $0.46 $0.60 46%
Before 2029 $0.65 $0.48 58%
Before 2030 $0.71 $0.54 73%
Before 2035 $0.76 $0.48 75%
Before 2040 $0.83 $0.41 57%

Market Discussion

Kalshi traders show growing optimism for the P vs NP problem's resolution, with probabilities significantly increasing across all timeframes over the past day. While 73% foresee a resolution before 2030 and 58% before 2029, traders are currently split on whether it will be resolved before 2028 (46% chance). The provided discussion does not detail specific arguments for or against these predicted timelines.

5. Trust Index

Octagon Trust Index Kalshi 70 Good

Order book is critically thin.

Primary risk· Trade quality

How it adds up
Integrity80% of score82Good
Trade quality20% of score20Avoid

Includes the cost to trade: a $10,000 order can't be filled here because the order book is too thin.

Trust score70Good

Weighted blend with hard caps — a critically weak safety pillar, or a severe trading anomaly, caps the total regardless of the rest. Full methodology · About the Trust Index

Trust profile
Integrity4 screens run · 6 don't apply

6. What is the expected peer review timeline for the OpenAI and Anthropic ETH refutations by authoritative bodies like the Annals of Mathematics?

Official Review StatusNo authoritative body has announced a review or timeline [1][2]
Papers Posted DateOctober 26, 2026 (OpenAI and Anthropic preprints) [1][2]
Claimed Advantage ThresholdAppears beyond ~10^10,000 variables [1][2]
Authoritative bodies have not yet announced formal review for ETH refutations. As of the available information, no authoritative body, including the Annals of Mathematics, the American Mathematical Society (AMS), the International Mathematical Union (IMU), or the Clay Mathematics Institute (CMI), has announced a formal review or timeline for the OpenAI and Anthropic ETH refutation papers [1][2]. An informal expert examination, led by Impagliazzo, Paturi, and Ryan Williams, was ongoing as of November 4–7, 2026. These reviewers clarified that their role does not certify the result, and neither paper has undergone peer-review nor been published in a journal [1][2].
OpenAI and Anthropic papers posted ETH refutations in late 2026. The OpenAI ETH-refutation preprint and its Lean formalization were posted on October 26, 2026, with Anthropic's independent paper reaching essentially the same bound [1][2]. Crucially, the ETH refutation explicitly does not resolve the P vs NP question, with OpenAI's paper stating it 'does not bear on the P vs NP question directly' [1][2]. The claimed advantage of a subexponential 3-SAT algorithm, which is weaker than what P=NP would require, only manifests beyond approximately 10^10,000 variables. The Clay Mathematics Institute continues to list P vs NP as unsolved as of November 6, 2026 [1][2].
Resolving P vs NP involves a lengthy formal review process. Any full journal peer-review cycle for a P vs NP claim would also entail a two-year post-publication wait for prize consideration from the Clay Mathematics Institute. Additionally, it would involve two-year resubmission bans from journals like JACM for P/NP claimants [1][2]. Furthermore, the Kalshi 'When will P vs NP be resolved?' market sets criteria for resolution, including a peer-reviewed proof/disproof or written recognition by an authoritative body [3].
Sources (3)
  1. 1Will P vs NP be resolved before 2028? | Forecasting Agentforecast.agenticlearning.ai
  2. 2Will P vs NP be resolved before 2030? | Forecasting Agentforecast.agenticlearning.ai
  3. 3When will P vs NP be resolved? | PredictPediapredictpedia.com

7. How does the expert consensus from the October 2026 LEAP survey compare to the market-implied probability of an AI-assisted Millennium Prize solution by 2030?

AI Millennium Prize Solution by 2030 (LEAP)80% probability (LEAP Wave 13 median expert forecast, Oct 2026) [1][2]
P vs NP Resolution Before 2030 (Market)21% (Kalshi market, Nov 20, 2026) [3]
AI Solves 3+ Millennium Prize Problems by 2030 (LEAP)45% probability (LEAP experts, Oct 2026) [1][2]
Expert consensus indicates high probability of AI-assisted Millennium Prize solutions. Expert consensus from the October 2026 LEAP survey indicates an 80% probability that artificial intelligence (AI) will assist in solving at least one Millennium Prize Problem (excluding Navier-Stokes) by the end of 2030 [1][2]. This finding contrasts significantly with the market-implied probability for a P vs NP resolution in the same timeframe, which was priced at 21% on the Kalshi market as of November 20, 2026 [3]. The LEAP survey, conducted October 6-27, 2026, also revealed that superforecasters predicted an 85% probability for this outcome, marking a substantial increase from 20% in summer 2025. This rise is primarily attributed to a September 2026 Navier-Stokes AI proof and late-October 2026 Lean-checked ETH refutations [1][4]. Additionally, LEAP experts forecast a 45% probability that AI will solve at least three of the five remaining Millennium Prize Problems by 2030 [1][2].
Scope differences and problem-specific barriers explain the considerable probability gap. The discrepancy between LEAP's 80% forecast for any Millennium Prize by 2030 and the market's approximately 21-24% for a P vs NP resolution before 2030 is attributed to these factors [1][5][6]. LEAP respondents highlight that P vs NP faces unique challenges, including relativization, natural-proofs, and algebrization, whereas problems like Hodge, BSD, and the Riemann Hypothesis are considered more tractable [1]. This perspective is supported by markets pricing the Riemann Hypothesis significantly higher, at 86-89% on markets concerning which Millennium Prize AI will solve next [1][5][6]. Furthermore, complexity theorist Lance Fortnow stated in June 2026 that he does not expect a P vs NP proof "in my lifetime" from humans or machines. The late-October 2026 OpenAI/Anthropic ETH refutations were also explicitly stated not to bear on P vs NP, aligning with the market's low ~21-24% Before-2030 price [4][7].
Sources (7)
  1. 1Wave 13: Short-Term Cruxes | LEAPleap.forecastingresearch.org
  2. 2Probability AI Solves Another Millennium Prize Problem | LEAPleap.forecastingresearch.org
  3. 3When will P vs NP be resolved? — Before 2030kalshiscan.com
  4. 4Will P vs NP be resolved before 2028? | Forecasting Agentforecast.agenticlearning.ai
  5. 5Tech & science prediction markets — live odds on Kalshi and Polymarket · OddsOnAiroddsonair.live
  6. 6Tech & AI Prediction Markets — Live Odds | Frenzy Capitalfrenzycap.com
  7. 7Will P vs NP be resolved before 2030? | Forecasting Agentforecast.agenticlearning.ai

8. How do the arguments for an imminent AI-driven resolution from Emanuele Viola contrast with the long-standing structural barriers cited by Lance Fortnow?

P=NP belief article dateNovember 17, 2026 [1]
P vs NP skepticism article dateJune 10, 2026 [2]
OpenAI ETH refutation acceptance dateOctober 28, 2026 [3]
Emanuele Viola foresees AI resolving P=NP, contrasting with Lance Fortnow's skepticism. Emanuele Viola argued on November 17, 2026, that recent AI advancements make P=NP a reasonable extrapolation, with AI tools expected to settle the problem [1]. He cited developments including Anthropic's 3SUM-conjecture-disproving algorithm, OpenAI's sub-n log n integer multiplication, and OpenAI's ETH refutation, which provided faster algorithms for problems previously conjectured not to admit them [1]. In contrast, Lance Fortnow stated on June 10, 2026, that an AI-generated P vs NP proof was not imminent, expressing his belief that such a proof would not appear in his lifetime, whether by man or machine [2]. Fortnow highlighted several structural barriers to resolving P vs NP, including that relativization blocks simulation and diagonalization, and that approaches from circuit complexity, proof complexity, and algebraic geometry stall below full polynomial-time power [2]. He also noted that proving P!=NP must exclude infinitely many bizarre polynomial-time algorithms, and that AI could not yet produce even a trivial full Lean proof [2].
Fortnow acknowledged AI progress but maintained that structural barriers persist for P vs NP. Although Fortnow accepted the OpenAI ETH refutation as valid on October 28, 2026, he stressed that the resulting bound gave nothing new for Subset Sum or TSP, and did not approach the 2^{n^{o(1)}} needed for subexponential algorithms for all of NP [3]. He further explained that P!=NP would require a super-polynomial lower bound for all algorithms, a side blocked by relativization, natural proofs, and algebrization barriers [3]. On November 4, 2026, Fortnow acknowledged that AI now finds proofs rather than just filling in details, but he maintained his belief that P vs NP requires new techniques that AI won't come up with on its own [4]. He also pointed out that Lean verification raises the risk that a formalization proves a different theorem than intended, noting this verification gap is one that Viola's AI-driven resolution scenario depends on closing [4][5].
Sources (5)
  1. 1What Happened in Fall 2026: AI and P vs NP – Communications of the ACMcacm.acm.org
  2. 2Computational Complexity: Respect the P v NP Problemblog.computationalcomplexity.org
  3. 3Computational Complexity: October 2026blog.computationalcomplexity.org
  4. 4Computational Complexity: 2026blog.computationalcomplexity.org
  5. 5P vs NP, a primer · Latent Mirrorlatentmirror.com

9. Based on the P vs NP Register, what is the trend in the rate and adjudication status of claimed solutions since the start of 2025?

Claimed P vs NP Solutions in 20257 (doubled from 3-4 annually 2021-2024) [1][2]
Post-2016 Unadjudicated Claims43 total; 12 refuted/critiqued, 4 withdrawn, 27 no recorded status [1][2]
Current P vs NP StatusUnsolved by Clay Mathematics Institute, no accepted resolution as of Nov 20-21, 2026 [3][4][5]
Claimed P vs NP solutions have notably increased since early 2025. The rate of claimed P vs NP solutions approximately doubled in 2025, with 7 recorded claims, compared to the prior baseline of 3–4 claims per year from 2021 to 2024 [1][2]. The broader P/NP literature also experienced significant growth, with 2,741 works appearing between 2021 and mid-2026, largely driven by 1,477 preprints [1][3]. Furthermore, post-2016 claims indicate a shift in direction, with 55.8% asserting P≠NP, a divergence from earlier periods where P=NP claims were more frequent [1].
Despite increased claims, no solutions have been adjudicated or accepted. The P vs NP Register shows that all 43 post-2016 candidates remain unadjudicated, and no claim recorded since 2025 has been accepted as a resolution [1][2]. Among these candidates, 12 (27.9%) have records of refutation or critique, 4 (9.3%) have been withdrawn, and 27 (62.8%) have no recorded status [1][2]. Notable claims from 2025 and 2026 include Lean-verified P=NP and P≠NP claims based on declared axioms [3][6].
P vs NP officially remains unsolved, despite related AI algorithm advancements. In October 2026, OpenAI and Anthropic reported AI-generated subexponential 3-SAT algorithms; however, these are explicitly not P vs NP resolutions, remain unverified, and are currently under review [3][6]. As of November 20–21, 2026, the Clay Mathematics Institute continues to list P vs NP as unsolved, indicating that no accepted resolution for the problem currently exists [3][4][5].
Sources (6)
  1. 1Statistics · The P versus NP Registercorroboration.org
  2. 2Home · The P versus NP Registercorroboration.org
  3. 3Will P vs NP be resolved before 2030? | Forecasting Agentforecast.agenticlearning.ai
  4. 42032 When will P vs NP be resolved? Odds on Kalshi | DOXAdoxa.pro
  5. 5When will P vs NP be resolved? | PredictionMarkets.USpredictionmarkets.us
  6. 6Will P vs NP be resolved before 2028? | Forecasting Agentforecast.agenticlearning.ai

10. Beyond ETH, what are the next key complexity theory conjectures AI labs like OpenAI and DeepMind are likely to target as stepping stones to P vs NP before 2030?

ETH Refutation AnnouncementOct 26, 2026 [1][2][3]
Subset Sum Algorithm AnnouncementOct 4, 2026 [2]
P vs NP Resolution Timelinesooner than 2029 [2][1]
AI labs recently refuted key complexity theory conjectures. OpenAI and Anthropic have made a significant breakthrough in complexity theory by refuting the Exponential Time Hypothesis (ETH) and its associated conjectures, including SETH, Gap-ETH, NSETH, and #ETH [3][2][1]. OpenAI published its Lean-verified, AI-generated paper on October 26, 2026, while Anthropic independently released a paper achieving essentially the same bound [2][1]. Although this refutation does not directly resolve the P vs NP problem, these papers indicate that the growing-width hypothesis and the Orthogonal Vectors Conjecture (OVC) are the next key conjectures AI labs are likely to pursue as stepping stones [3][4][2][1].
Upcoming research will likely target the Orthogonal Vectors Conjecture. The OVC holds particular importance because it underpins fine-grained lower bounds for various problems, such as edit distance, Longest Common Subsequence (LCS), Fréchet distance, and regular-expression matching, now that SETH is no longer applicable [3]. This strategic focus on long-standing exponential-time barriers is consistent with previous efforts by AI labs [2]. For instance, on October 4, 2026, OpenAI released a Subset Sum algorithm that surpassed a previously unbroken ~2^(n/2) exponent barrier, demonstrating their systematic approach to fundamental problems [2].
P vs NP resolution is a near-term goal for AI labs. According to Scott Aaronson, as of October 27, 2026, both OpenAI and Anthropic are actively competing to solve P vs NP using their internal models [2][1]. A possible timeline for this fundamental problem's resolution has been suggested as "sooner than 2029" [2][1].
Sources (4)
  1. 1Will P vs NP be resolved before 2030? | Forecasting Agentforecast.agenticlearning.ai
  2. 2Will P vs NP be resolved before 2028? | Forecasting Agentforecast.agenticlearning.ai
  3. 3OpenAI ETH refutation - subexponential SAT claim explainedbraindetox.kr
  4. 4AI overturns ETH | Horizonshorizons.comp.nus.edu.sg

11. What Could Change the Odds

Key Catalysts

Recent algorithmic advances have impacted market sentiment regarding the P vs NP problem. On Oct 5, 2026, a paper co-authored by Alman & Vassilevska Williams reported an Anthropic-model-discovered algorithm disproving the 3SUM conjecture [1]. The following day, Oct 6, 2026, an OpenAI algorithm for multiplying n-bit integers faster than n log n was reported [1]. The most significant event was the Oct 26, 2026 publication of an OpenAI preprint proving the Exponential Time Hypothesis false with a 2^O(n·log log n/√log n) algorithm for 3-SAT, formalized in Lean on Oct 17 [2][1]. This disproof also refutes SETH, Gap-ETH, NSETH, and #ETH [2][1]. An independent review group, including Impagliazzo, Paturi, and Ryan Williams, has reported no flaw as of Oct 29, 2026 [2][3][4]. Emanuele Viola, in a Nov 17, 2026 CACM blog, argued these results support P=NP and suggest AI will settle the problem [1]. This sequence of events coincided with a significant surge in Kalshi market odds between Nov 19-21, 2026, showing increases of +39-40 points for resolution before 2035 (to ~75-76%), +50 points for before 2030 (to ~73%), +47 points for before 2029 (to ~58%), and +43 points for before 2028 (to ~49%) [5][6][7][8][9].
However, several factors weigh against a near-term resolution of P vs NP. Disproving the Exponential Time Hypothesis does not equate to P=NP; a subexponential 2^o(n) algorithm remains considerably slower than a polynomial one and does not transfer under general polynomial-time reductions to general NP problems [10]. There is no direct cryptographic impact from this disproof [10]. Established barrier theorems, including Baker-Gill-Solovay (1975), Razborov-Rudich (1997), and Aaronson-Wigderson (2008), impede entire categories of P≠NP proof techniques [11][2]. The Clay Mathematics Institute still listed P vs NP as unsolved as of Oct 6, 2026 [11]. Furthermore, the Clay prize requires two years of community acceptance following publication, and prior claims, such as Ke Xu & Guangyan Zhou's July 4, 2025 'SAT requires exhaustive search', faced strong rebuttal by Sept 2025 [11]. Market forecasts from Metaculus show a median estimate of April 2050 for resolution, with OpinionKings pricing Before 2040 at 46% and Before 2035 at 36% as of Nov 12, 2026, which are more conservative than Kalshi's post-surge figures [12][13][14].

Key Dates & Catalysts

  • Expiration: January 07, 2027
  • Closes: January 07, 2040
Sources (14)
  1. 1What Happened in Fall 2026: AI and P vs NP – Communications of the ACMcacm.acm.org
  2. 2OpenAI and the Exponential Time Hypothesis: What Is Proven - Thorsten Meyer AIthorstenmeyerai.com
  3. 3OpenAI and Anthropic Claim AI Has Disproved a 27-Year-Old Computer Science Hypothesishackernoon.com
  4. 4Exponential Time Hypothesis Disproof: What OpenAI Claims | USA 24usa-24.com
  5. 5When will P vs NP be resolved? Odds & Predictions 2031kalshi.com
  6. 6When will P vs NP be resolved?: Live Odds (Full Field) | ContractTaxcontracttax.com
  7. 7When will P vs NP be resolved odds: Before 2035 76% on Kalshipredictdesks.com
  8. 8When will P vs NP be resolved? | PredictionSwappredictionswap.ai
  9. 9Will P vs NP be resolved before 2029? — 58.0% on Kalshipredictionbubbles.net
  10. 10OpenAI's Exponential Time Hypothesis Disproof Explained: What It Is, How It Works, Limits and Risks - PiyushAIpiyushai.com
  11. 11P versus NPpapex.org
  12. 12When will P = NP be settled?metaculus.com
  13. 13P equals NP. · PNPscience.markets
  14. 14When will P vs NP be resolved? · OpinionKingsopinionkings.com