AI predictions from labs, reports, markets, and experts, sorted by significance.
2026 #
[Powerful AI emerges by late 2026](/predictions/anthropic-amodei-2026-q4-001) -
[AI speeds a major lab's research by 50%](/predictions/ai2027-2026-q1-001) -
[Model-weight theft becomes a strategic threat](/predictions/ai2027-2026-q1-003) -
[China launches a national AI push](/predictions/ai2027-2026-q2-001) -
[AI-led stocks rise 30%](/predictions/ai2027-2026-q4-002) -
[Global AI capex reaches $1T](/predictions/ai2027-2026-q4-003) -
[AI computation reaches Finland-scale electricity use](/predictions/international-ai-safety-report-2026-2026-001) -
[AI reaches month-long task horizon](/predictions/metr-time-horizon-2026-q4-001) -
[A major AI company publicly releases its AI agent](/predictions/ai2027-2026-q1-002) -
[China centralizes AI compute at one site](/predictions/ai2027-2026-q2-002) -
[Junior software job market enters turmoil](/predictions/ai2027-2026-q4-001) -
[AI outcompetes PhDs at problem solving](/predictions/situational-awareness-2026-002) -
[Big tech reaches $100B AI revenue](/predictions/situational-awareness-2026-001) -
[AI systems figure out novel insights](/predictions/altman-gentle-singularity-2026-001) -
[ARC-AGI-3 tests agentic skill acquisition in early 2026](/predictions/francois-chollet-2026-q1-001) -
[AI agents enter one-third of B2B payments by 2026](/predictions/forrester-2026-predictions-2026-016) -
[AI privacy lawsuits surge 20% in 2026](/predictions/forrester-2026-predictions-2026-008) -
[Sellers face agent-led quote negotiations in 2026](/predictions/forrester-2026-predictions-2026-017) -
[Ungoverned genAI costs B2B firms $10B in 2026](/predictions/forrester-2026-predictions-2026-013) -
[Agency acquisition triggers a dozen reviews in 2026](/predictions/forrester-2026-predictions-2026-012) -
[B2B influencer budgets rise in 2026](/predictions/forrester-2026-predictions-2026-015) -
[CFOs gate AI investments in 2026](/predictions/forrester-2026-predictions-2026-002) -
[CIOs bail out business-led AI failures in 2026](/predictions/forrester-2026-predictions-2026-006) -
[CX teams abandon journey mapping in 2026](/predictions/forrester-2026-predictions-2026-010) -
[Developer hiring time doubles in 2026](/predictions/forrester-2026-predictions-2026-004) -
[Display ad budgets fall 30% in 2026](/predictions/forrester-2026-predictions-2026-007) -
[Five major brands unify agentic commerce in 2026](/predictions/forrester-2026-predictions-2026-011) -
[Human expertise rivals genAI for buyers in 2026](/predictions/forrester-2026-predictions-2026-014) -
[Neocloud revenue reaches $20B in 2026](/predictions/forrester-2026-predictions-2026-003) -
[One-third of consumers choose offline brand experiences in 2026](/predictions/forrester-2026-predictions-2026-009) -
[Quantum security exceeds 5% of budgets in 2026](/predictions/forrester-2026-predictions-2026-005)
2027 #
[U.S. officials expect imminent AGI](/predictions/ai2027-2027-q2-005) -
[A major AI company announces AGI](/predictions/ai2027-2027-q3-001) -
[AI does 25% of remote-work jobs](/predictions/ai2027-2027-q4-021) -
[Allies demand an AI after being sidelined](/predictions/ai2027-2027-q4-003) -
[A misaligned AI system achieves superintelligence](/predictions/ai2027-2027-q4-012) -
[DPA consolidates top U.S. AI compute](/predictions/ai2027-2027-q4-009) -
[AI reaches Nobel-level expertise across fields](/predictions/anthropic-amodei-2027-q1-001) -
[AI triples a major lab's algorithmic progress](/predictions/ai2027-2027-q1-002) -
[An escaped AI could survive, copy itself, and hack](/predictions/ai2027-2027-q1-003) -
[AI cyber capability becomes a strategic asset](/predictions/ai2027-2027-q1-006) -
[A cheap AI system becomes a superhuman coder](/predictions/ai2027-2027-q1-010) -
[A major lab's algorithmic progress reaches 4x](/predictions/ai2027-2027-q1-012) -
[An advanced AI lacks robust truth-seeking](/predictions/ai2027-2027-q2-002) -
[Misalignment tests resist available fixes](/predictions/ai2027-2027-q2-003) -
[AI research progress reaches 10x](/predictions/ai2027-2027-q2-009) -
[A public AI model poses extreme biosecurity danger](/predictions/ai2027-2027-q3-003) -
[AI apps and B2B SaaS surge](/predictions/ai2027-2027-q3-005) -
[Programming hiring nearly stops](/predictions/ai2027-2027-q3-004) -
[White House treats AI race like the Cold War](/predictions/ai2027-2027-q3-006) -
[A superhuman AI becomes adversarially misaligned](/predictions/ai2027-2027-q3-013) -
[AI research progress reaches 50x](/predictions/ai2027-2027-q3-012) -
[An AI system becomes a superhuman AI researcher](/predictions/ai2027-2027-q3-010) -
[Evidence of AI misalignment accumulates](/predictions/ai2027-2027-q3-014) -
[A major AI company continues near full speed](/predictions/ai2027-2027-q4-008) -
[A major AI company pivots to safer models](/predictions/ai2027-2027-q4-005) -
[A safer AI uses faithful English reasoning](/predictions/ai2027-2027-q4-006) -
[Misalignment memo leaks to the press](/predictions/ai2027-2027-q4-001) -
[Public backlash and congressional subpoenas erupt](/predictions/ai2027-2027-q4-002) -
[U.S. creates an AI oversight committee](/predictions/ai2027-2027-q4-004) -
[A major AI company controls 50% of world AI compute](/predictions/ai2027-2027-q4-010) -
[AI launches a compressed dot-com boom](/predictions/ai2027-2027-q4-019) -
[China seeks a mutual AI slowdown](/predictions/ai2027-2027-q4-016) -
[U.S. and China reach an AI deadlock](/predictions/ai2027-2027-q4-015) -
[AI systems complete hours-long software projects](/predictions/international-ai-safety-report-2026-2027-001) -
[Largest training runs exceed $1B](/predictions/international-ai-safety-report-2026-2027-003) -
[AGI arrives by 2027](/predictions/situational-awareness-2027-001) -
[Annual AI investment reaches $1 trillion](/predictions/situational-awareness-2027-005) -
[Government AGI project gets underway](/predictions/situational-awareness-2027-004) -
[Machines have 10% chance to beat humans at all tasks](/predictions/grace-ai-authors-2027-001) -
[Labor-market churn reaches 23% by 2027](/predictions/wef-future-of-jobs-2023-2027-001) -
[Machines perform 42% of business tasks by 2027](/predictions/wef-future-of-jobs-2023-2027-003) -
[AI handles hours-to-weeks tasks autonomously](/predictions/anthropic-amodei-2027-q1-002) -
[Frontier training runs need five gigawatts](/predictions/anthropic-amodei-2027-001) -
[U.S. builds fifty AI-dedicated gigawatts](/predictions/anthropic-amodei-2027-002) -
[A major lab starts online-learning post-training](/predictions/ai2027-2027-q1-001) -
[China steals frontier AI model weights](/predictions/ai2027-2027-q1-007) -
[U.S. sabotage attempts against Chinese AI fail](/predictions/ai2027-2027-q1-008) -
[A major lab runs 200,000 AI researcher copies](/predictions/ai2027-2027-q1-011) -
[Major algorithmic breakthroughs create a new AI system](/predictions/ai2027-2027-q1-009) -
[AI agents handle a wide range of week-long tasks](/predictions/metr-time-horizon-2027-q1-001) -
[AI systems dominate a major lab's research](/predictions/ai2027-2027-q2-008) -
[Most human AI researchers stop contributing much](/predictions/ai2027-2027-q2-010) -
[A major AI company publicly releases a cheap AGI](/predictions/ai2027-2027-q3-002) -
[China falls two months behind in AI race](/predictions/ai2027-2027-q3-009) -
[U.S. tightens chip, internet, and employee controls](/predictions/ai2027-2027-q3-007) -
[300,000 AI copies run at 50x human speed](/predictions/ai2027-2027-q3-011) -
[Safer AI reaches 20x research progress](/predictions/ai2027-2027-q4-007) -
[A misaligned AI wins more autonomy](/predictions/ai2027-2027-q4-013) -
[AI race continues despite China offer](/predictions/ai2027-2027-q4-014) -
[A superintelligent AI becomes an elite advisor](/predictions/ai2027-2027-q4-018) -
[A superintelligent AI excels at politics and lobbying](/predictions/ai2027-2027-q4-020) -
[AI models reach 55% on undergraduate FrontierMath](/predictions/international-ai-safety-report-2026-2027-002) -
[A team starts working on core AGI ideas by 2027](/predictions/francois-chollet-2027-001) -
[AI agents become drop-in remote workers](/predictions/situational-awareness-2027-003) -
[AI creates $10T companies](/predictions/situational-awareness-2027-006) -
[Administrative roles lose 26 million jobs by 2027](/predictions/wef-future-of-jobs-2023-2027-005) -
[AI and big-data roles grow 30-35% by 2027](/predictions/wef-future-of-jobs-2023-2027-010) -
[AI and machine-learning specialists become the fastest-growing role by 2027](/predictions/wef-future-of-jobs-2023-2027-009) -
[Employers expect 14 million fewer jobs by 2027](/predictions/wef-future-of-jobs-2023-2027-002) -
[Six in 10 workers need training by 2027](/predictions/wef-future-of-jobs-2023-2027-006) -
[Workers face 44% skill disruption by 2027](/predictions/wef-future-of-jobs-2023-2027-007) -
[AI uses digital tools and lab equipment](/predictions/anthropic-amodei-2027-q1-003) -
[Millions of AI instances run at 10x-100x human speed](/predictions/anthropic-amodei-2027-q1-004) -
[Security-relevant AI capabilities advance by 2027](/predictions/anthropic-amodei-2027-q1-005) -
[Interpretability falls short of full model understanding](/predictions/ai2027-2027-q2-004) -
[Five Eyes allies receive limited AGI access](/predictions/ai2027-2027-q3-008) -
[Agent token and API loads rise 1000x by 2027](/predictions/idc-enterprise-agents-2027-002) -
[Agentic automation reaches 40% of enterprise apps by 2027](/predictions/idc-enterprise-agents-2027-003) -
[Global 2000 agent use increases tenfold by 2027](/predictions/idc-enterprise-agents-2027-001) -
[Effective compute rises five orders of magnitude](/predictions/situational-awareness-2027-002) -
[Robots do real-world tasks](/predictions/altman-gentle-singularity-2027-001) -
[AI and big data become a major training priority by 2027](/predictions/wef-future-of-jobs-2023-2027-008) -
[Nearly 75% of companies adopt AI by 2027](/predictions/wef-future-of-jobs-2023-2027-004) -
[Over 40% of agentic AI projects are canceled by 2027](/predictions/gartner-ai-software-2027-001) -
[Enterprises defer 25% of AI spend into 2027](/predictions/forrester-2026-predictions-2027-001) -
[Poor AI-ready data cuts productivity 15% by 2027](/predictions/idc-enterprise-agents-2027-004)
2028 #
[Superhuman AI is publicly released](/predictions/ai2027-2028-q2-004) -
[A misaligned AI becomes wildly superintelligent](/predictions/ai2027-2028-q2-002) -
[U.S. and Chinese AIs negotiate a treaty](/predictions/ai2027-2028-q3-001) -
[AI progress compresses six years into two](/predictions/international-ai-safety-report-2026-2028-001) -
[Weak general AI becomes publicly available](/predictions/metaculus-agi-2028-q2-001) -
[A safer AI becomes aligned and transparent](/predictions/ai2027-2028-q1-001) -
[Robot economy expands through special zones](/predictions/ai2027-2028-q2-001) -
[Robots pass the Coffee Test](/predictions/ai2027-2028-q2-005) -
[Treaty-compliant chips replace old systems](/predictions/ai2027-2028-q3-002) -
[AI election control becomes a public issue](/predictions/ai2027-2028-q3-003) -
[Vice President wins AI-era election](/predictions/ai2027-2028-q4-002) -
[Factories produce one million robots monthly](/predictions/ai2027-2028-q4-003) -
[AI reaches research-level scientific performance](/predictions/international-ai-safety-report-2026-2028-002) -
[OpenAI compute growth slows around 2028](/predictions/metr-compute-slowdowns-2028-001) -
[AI research progress speeds up 3x](/predictions/situational-awareness-2028-004) -
[AI widely automates software engineering](/predictions/situational-awareness-2028-003) -
[Intelligence explosion gets underway](/predictions/situational-awareness-2028-001) -
[A safer AI exceeds top human experts](/predictions/ai2027-2028-q1-002) -
[Safer AI reaches 200x research progress](/predictions/ai2027-2028-q1-003) -
[U.S. chooses a win-the-race AI strategy](/predictions/ai2027-2028-q1-004) -
[Public AI deployment transforms the economy](/predictions/ai2027-2028-q2-003) -
[AI drives stratospheric GDP growth](/predictions/ai2027-2028-q4-001) -
[A 2 GW AI supercomputer becomes feasible](/predictions/trends-ai-supercomputers-2028-001) -
[Training clusters cost hundreds of billions](/predictions/situational-awareness-2028-002) -
[Agentic AI makes 15% of work decisions by 2028](/predictions/gartner-ai-software-2028-001) -
[AI builds payment processing websites](/predictions/grace-ai-authors-2028-003) -
[AI decision explanations remain unlikely](/predictions/grace-ai-authors-2028-002) -
[AI fine-tunes open-source LLMs](/predictions/grace-ai-authors-2028-005) -
[AI makes songs indistinguishable from artists](/predictions/grace-ai-authors-2028-004) -
[One-third of enterprise apps include AI agents by 2028](/predictions/gartner-ai-software-2028-002)
2029 #
[Humans become obsolete to the robot economy](/predictions/ai2027-2029-q4-004) -
[Most work shifts to AIs and robots](/predictions/ai2027-2029-q4-002) -
[U.S. and China accept an AI-brokered deal](/predictions/ai2027-2029-q4-003) -
[AI can complete month-long software tasks](/predictions/metr-time-horizon-2029-q2-001) -
[Robots, fusion, and disease cures become commonplace](/predictions/ai2027-2029-q4-001) -
[AI progress runs 10x faster](/predictions/situational-awareness-2029-001) -
[Agents execute 217 billion daily actions by 2029](/predictions/idc-enterprise-agents-2029-002) -
[Deployed AI agents exceed 1 billion by 2029](/predictions/idc-enterprise-agents-2029-001) -
[Agent action costs fall 87% by 2029](/predictions/idc-enterprise-agents-2029-005) -
[Agents consume 3.7T daily token calls by 2029](/predictions/idc-enterprise-agents-2029-004) -
[Agent token delivery costs exceed $68B by 2029](/predictions/idc-enterprise-agents-2029-003)
2030 #
[Misaligned AI kills most humans](/predictions/ai2027-2030-q2-001) -
[China transitions after an AI-sabotaged coup](/predictions/ai2027-2030-q4-001) -
[World government forms under U.S. control](/predictions/ai2027-2030-q4-002) -
[AI systems complete days-long software projects](/predictions/international-ai-safety-report-2026-2030-002) -
[AI systems complete month-long professional tasks](/predictions/international-ai-safety-report-2026-2030-015) -
[AI systems do remote-worker jobs at human level](/predictions/international-ai-safety-report-2026-2030-016) -
[AI-guided robots handle complex real-world tasks](/predictions/international-ai-safety-report-2026-2030-017) -
[Robot economy expands beyond human zones](/predictions/ai2027-2030-q1-001) -
[AI capabilities largely plateau](/predictions/international-ai-safety-report-2026-2030-013) -
[AI improves across specialized domains](/predictions/international-ai-safety-report-2026-2030-004) -
[AI systems become useful digital assistants](/predictions/international-ai-safety-report-2026-2030-014) -
[AI uses 7.4% of US electricity](/predictions/international-ai-safety-report-2026-2030-009) -
[Chip supply supports 100,000x GPT-4 compute](/predictions/international-ai-safety-report-2026-2030-010) -
[Energy bottlenecks do not stop AI scaling](/predictions/international-ai-safety-report-2026-2030-011) -
[Frontier models train with 3,000x more compute](/predictions/international-ai-safety-report-2026-2030-007) -
[Largest training runs need 4-16 GW](/predictions/international-ai-safety-report-2026-2030-008) -
[Largest-model training compute grows 125-fold](/predictions/international-ai-safety-report-2026-2030-001) -
[Models need 10-100x less compute](/predictions/international-ai-safety-report-2026-2030-005) -
[AGI arrives around 2030](/predictions/google-deepmind-agi-2030-2030-001) -
[Superintelligence arrives by 2030](/predictions/situational-awareness-2030-001) -
[Data center electricity use reaches 945 TWh](/predictions/iea-energy-ai-2030-001) -
[Data centers drive 20% of advanced-economy demand growth](/predictions/iea-energy-ai-2030-004) -
[Small modular reactors come online for data centers](/predictions/iea-energy-ai-2030-003) -
[U.S. data centers outconsume heavy industry](/predictions/iea-energy-ai-2030-002) -
[AI models reach 75% on undergraduate FrontierMath](/predictions/international-ai-safety-report-2026-2030-003) -
[Autonomous vehicle deployment remains limited](/predictions/international-ai-safety-report-2026-2030-012) -
[AGI arrives around 2030](/predictions/francois-chollet-2030-001) -
[Distributed AI training spreads across sites](/predictions/trends-ai-supercomputers-2030-006) -
[Leading AI supercomputer costs $200B](/predictions/trends-ai-supercomputers-2030-003) -
[Leading AI supercomputer reaches 2e22 FLOP/s](/predictions/trends-ai-supercomputers-2030-001) -
[Leading AI supercomputer requires 9 GW](/predictions/trends-ai-supercomputers-2030-004) -
[Leading AI supercomputer uses 2 million chips](/predictions/trends-ai-supercomputers-2030-002) -
[Power becomes main AI-supercomputer bottleneck](/predictions/trends-ai-supercomputers-2030-005) -
[Trillion-dollar cluster reaches 100 GW](/predictions/situational-awareness-2030-002) -
[Intelligence becomes too cheap to meter](/predictions/altman-gentle-singularity-2030-002) -
[One person accomplishes far more than in 2020](/predictions/altman-gentle-singularity-2030-001)
2031 #
[AI compresses fifty years of biology into five](/predictions/anthropic-amodei-2031-001) -
[AI makes lifespan doubling possible](/predictions/anthropic-amodei-2031-004) -
[Most cancer mortality and incidence falls sharply](/predictions/anthropic-amodei-2031-003) -
[Most genetic disease becomes preventable or curable](/predictions/anthropic-amodei-2031-010) -
[Most infectious disease becomes preventable or treatable](/predictions/anthropic-amodei-2031-002) -
[Most mental illness becomes treatable or curable](/predictions/anthropic-amodei-2031-005) -
[AI health gains reach the poorest countries](/predictions/anthropic-amodei-2031-012) -
[Developing countries begin rapid AI catch-up](/predictions/anthropic-amodei-2031-007) -
[U.S. remains AI-supercomputer leader](/predictions/trends-ai-supercomputers-2031-001)
2032 #
2033 #
[AI reaches month-long tasks at 50% reliability](/predictions/metr-compute-slowdowns-2033-001) -
[AI automation exposes 300 million full-time jobs by 2033](/predictions/goldman-generative-ai-gdp-2033-q2-003) -
[Generative AI raises global GDP 7% by 2033](/predictions/goldman-generative-ai-gdp-2033-q2-001) -
[Exposed U.S. jobs lose 25-50% of workload by 2033](/predictions/goldman-generative-ai-gdp-2033-q2-006) -
[Generative AI lifts productivity growth 1.5 points by 2033](/predictions/goldman-generative-ai-gdp-2033-q2-002) -
[Generative AI software market reaches $150B by 2033](/predictions/goldman-generative-ai-gdp-2033-q2-005) -
[Two-thirds of U.S. occupations face AI exposure by 2033](/predictions/goldman-generative-ai-gdp-2033-q2-004)
source & further reading
whenwill.ai — original article