The Prediction Ledger
What the industry said would happen, and by when — with the source for every line. When a date arrives the line is marked happened, did not happen, walked back or disputed, with evidence. The pattern is the point, not the person. Data as of 2026-08-31. Calendar of the due dates (ICS) · the bubble, month by month.
50 dated claims · 7 happened 13 did not 3 walked back 0 due, awaiting review 22 pending
| Who | What they said | When · source | Due | Status |
|---|---|---|---|---|
| Sergey Brin, co-founder | “You can count on one hand the number of years until ordinary people can experience this.” Self-driving cars are available to ordinary people within five years, i.e. by 2017. | 2012-09-25 Computerworld (California SB 1298 signing at Google HQ) | 2017-09-25 | did not happen by the date Waymo did not open fully driverless rides to the general public in Phoenix until October 2020, and only within a limited service area. evidence |
| Elon Musk, CEO Tesla | (paraphrase) Musk said Tesla would demonstrate a fully autonomous drive from a home in Los Angeles to Times Square, New York, without a single touch, by the end of 2017. A Tesla drives itself coast to coast with no human input by the end of 2017. Fuller wording in reports: 'a demonstration drive of full autonomy all the way from LA to New York... by the end of next year. Without the need for a single touch, including the charger.' | 2016-10-19 NBC News (Tesla press call, Oct 19, 2016) | 2017-12-31 | did not happen by the date Tesla never staged the demonstration; the cross-country drive is still cited in 2026 as the first of a decade of missed autonomy deadlines. evidence |
| Elon Musk, CEO Tesla | “I feel very confident predicting autonomous robotaxis for Tesla next year” Tesla operates autonomous robotaxis, over a million of them, on public roads by the end of 2020. He also said 'over 1 million robotaxis on the road next year' and, of past deadlines, 'Sometimes I am not on time, but I get it done.' | 2019-04-22 CNBC (Tesla Autonomy Investor Day) | 2020-12-31 | did not happen by the date Tesla's first paid robotaxi rides did not begin until June 2025, in a small geofenced Austin pilot with safety monitors aboard. evidence |
| John Zimmer, co-founder and president Lyft | “within five years a fully autonomous fleet of cars will provide the majority of Lyft rides across the country.” Most Lyft rides nationwide are provided by fully autonomous vehicles by September 2021. Same essay: 'By 2025, private car ownership will all-but end in major U.S. cities.' | 2016-09-18 John Zimmer, Medium essay 'The Third Transportation Revolution' | 2021-09-18 | did not happen by the date In October 2022 Lyft still offered only a handful of partner-run driverless pilots, all with safety drivers, and Zimmer acknowledged the timeline was wrong. evidence |
| Ford Motor Company (CEO Mark Fields) Ford | (paraphrase) Ford announced it would have a high-volume, fully autonomous SAE Level 4 vehicle, with no steering wheel or pedals, in commercial ride-hailing or ride-sharing operation in 2021. A driverless Ford ride-hailing vehicle is in commercial operation in 2021. Fields called it 'a transformational moment in our industry'. Ford's own release: media.ford.com, Aug 16, 2016. | 2016-08-16 Associated Press via Phys.org (Ford Palo Alto event) | 2021-12-31 | did not happen by the date Ford's autonomy partner Argo AI was shut down in October 2022 and Ford said profitable fully autonomous vehicles at scale were 'a long way off'. evidence |
| Elon Musk, CEO Tesla | “I am highly confident the car will be able to drive itself with reliability in excess of a human this year.” Tesla vehicles achieve full (Level 5) self-driving more reliable than a human during 2021. | 2021-01-27 TechCrunch (Q4 2020 earnings call; DMV memo) | 2021-12-31 | did not happen by the date Weeks later Tesla's own Autopilot director told California regulators the tweeted timeline 'does not match engineering reality' and described the system as Level 2; FSD remains supervised in 2026. evidence |
| Klarna (official press release; CEO Sebastian Siemiatkowski) Klarna | “It is doing the equivalent work of 700 full-time agents” Klarna's AI assistant replaces the work of 700 agents and drives a $40 million profit improvement in 2024. Also reported by Bloomberg, May 8, 2025: 'Klarna Turns From AI to Real Person Customer Service'. | 2024-02-27 Klarna press release | 2024-12-31 | walked back In May 2025 Klarna said it was recruiting human customer-service staff again; Siemiatkowski said cost had been 'a too predominant evaluation factor' and the result was 'lower quality'. evidence |
| 16 signatory companies (Amazon, Anthropic, Google, Meta, Microsoft, OpenAI, xAI and others) Frontier AI Safety Commitments, AI Seoul Summit | “publish a safety framework focused on severe risks by the upcoming AI Summit in France” Every signatory publishes a frontier safety framework before the Paris AI Action Summit of February 10-11, 2025. Signatory list later grew to 20 companies. | 2024-05-21 UK Government, Frontier AI Safety Commitments | 2025-02-10 | disputed Most major signatories published frameworks by February 2025 (several only in the final days), but some were late or minimal; xAI, for example, had only a draft at the deadline per METR's tracker. evidence |
| Apple (official newsroom announcement, WWDC) Apple | “some features, software platforms, and additional languages will come over the course of the next year” The more personal Siri (personal context, onscreen awareness, in-app actions) ships by mid-2025. Delay statement given to Daring Fireball, Mar 7, 2025: 'It's going to take us longer than we thought to deliver on these features.' | 2024-06-10 Apple Newsroom, 'Introducing Apple Intelligence' | 2025-06-10 | did not happen by the date Apple said in March 2025 the features would take longer and arrive 'in the coming year'; as of April 2026 the Gemini-powered personal Siri was still promised only for 'later in 2026'. evidence |
| Brad Smith, vice chair and president Microsoft | “In FY 2025, Microsoft is on track to invest approximately $80 billion to build out AI-enabled datacenters to train AI models and deploy AI and cloud-based applications around the world.” Microsoft spends about $80 billion on AI-enabled data centers in fiscal 2025 (July 2024-June 2025). | 2025-01-03 Microsoft On the Issues blog | 2025-06-30 | happened Microsoft's fiscal-Q4 2025 capex including finance leases was $24.2 billion, bringing the four quarters of FY2025 to roughly $88 billion, above the $80 billion figure; FY2026 guidance rose further. evidence |
| Elon Musk, CEO Tesla | “We're going to be launching unsupervised full self-driving as a paid service in Austin in June.” A paid, unsupervised robotaxi service launches in Austin in June 2025. Same call: 'I'm confident that we'll release unsupervised FSD California this year as well' (did not happen in 2025). In May 2026 Musk again promised unsupervised FSD 'widespread in the US by the end of this year.' | 2025-01-29 Tesla Q4 2024 earnings call transcript (Motley Fool) | 2025-06-30 | disputed Paid rides began June 22, 2025 with about ten Model Ys in a South Austin geofence, invite-only, and a Tesla employee in the front passenger seat as a safety monitor, so 'unsupervised' is contested. evidence |
| Mustafa Suleyman, co-founder and CEO of Inflection AI (now CEO, Microsoft AI) Inflection AI | “Go make $1 million on a retail web platform in a few months with just a $100,000 investment... could be as little as two years away.” An AI can autonomously turn $100,000 into $1 million on a retail platform within about two years, i.e. by mid-2025. Suleyman framed it as 'something like this could be as little as two years away', a soft deadline. | 2023-07-14 MIT Technology Review (interview) | 2025-07-14 | did not happen by the date No AI system has been shown to pass the test; in Anthropic's Project Vend (June 2025) a Claude agent running a small shop for a month lost money. evidence |
| Elon Musk, CEO of Tesla and SpaceX Tesla | “we're headed toward a situation where AI is vastly smarter than humans” AI becomes vastly smarter than humans in less than five years, i.e. by July 2025. Musk added the five-year frame did not mean 'everything goes to hell', only that 'things get unstable or weird.' In April 2024 he re-dated the claim to 'around the end of next year' (2025) or 2026. | 2020-07-25 CNBC (reporting Musk's July 25, 2020 New York Times interview) | 2025-07-25 | did not happen by the date No lab claimed human-surpassing general intelligence by mid-2025; at Davos in January 2026 Google DeepMind's Demis Hassabis said current systems were 'nowhere near' human-level AGI. evidence |
| Dario Amodei, CEO Anthropic | “I think we'll be there in three to six months—where AI is writing 90 percent of the code” AI writes 90% of code within three to six months (by September 2025) and essentially all code within twelve months (by March 2026). Same answer: 'in twelve months, we may be in a world where AI is writing essentially all of the code.' | 2025-03-10 Council on Foreign Relations event transcript | 2025-09-10 | did not happen by the date By the deadline no industry-wide measure showed AI writing 90% of code; in January 2026 Amodei re-dated the claim, saying models would replace software developers 'within a year'. Anthropic does say most of its own code is now AI-written. evidence |
| Kyle Vogt, CEO Cruise (General Motors) | (paraphrase) Vogt said Cruise aimed to reach $1 billion in annual revenue by 2025. Cruise generates $1 billion of revenue in 2025. Vogt repeated in April 2023 that Cruise was 'on track' for the target. | 2022-09-12 CNBC (Goldman Sachs Communacopia conference) | 2025-12-31 | did not happen by the date GM stopped funding Cruise's robotaxi business in December 2024 after more than $10 billion of investment, folding the unit into its driver-assistance work. evidence |
| Gary Marcus, cognitive scientist, NYU professor emeritus, author Independent (Marcus on AI) | “There could continue to be no 'GPT-5 level' model (meaning a huge, across the board quantum leap forward as judged by community consensus) throughout 2025.” No model delivering a consensus quantum leap over GPT-4 appears in 2025. Post published in early January 2025; day not shown, so set to 01. Same list predicted AI-company valuations 'could well' start to fall in 2025, which did not happen. | 2025-01-01 Gary Marcus, Substack '25 AI Predictions for 2025' | 2025-12-31 | disputed OpenAI shipped GPT-5 on Aug 7, 2025 and called it 'a significant leap in intelligence', but much of the community judged the gain incremental, so whether a 'GPT-5 level' leap occurred is contested. evidence |
| Sam Altman, CEO OpenAI | “We believe that, in 2025, we may see the first AI agents 'join the workforce' and materially change the output of companies.” AI agents will join the workforce and materially change company output during 2025. Same post: 'We are now confident we know how to build AGI as we have traditionally understood it.' Posted Jan 5-6, 2025 (US time). | 2025-01-06 Sam Altman blog, 'Reflections' | 2025-12-31 | disputed Agent products shipped in 2025 (Operator, ChatGPT agent, Codex), but an MIT NANDA study in Aug 2025 found roughly 95% of enterprise generative-AI pilots produced no measurable P&L impact. evidence |
| Mark Zuckerberg, CEO Meta | (paraphrase) Zuckerberg said Meta would invest $60-65 billion in capital expenditures in 2025, bring about 1 GW of compute online and end the year with more than 1.3 million GPUs. Meta spends $60-65 billion on capex in 2025. Pattern: capex guidance was exceeded, not missed. Post called 2025 'a defining year for AI'. | 2025-01-24 CNBC (reporting Zuckerberg's Facebook post) | 2025-12-31 | happened Meta's actual 2025 capital expenditures including finance leases were $72.22 billion, above the original range after two upward revisions during the year. evidence |
| Elon Musk, CEO Tesla | “Will we succeed in building 10,000 exactly by the end of December this year? Probably not, but will we succeed in making several thousand? Yes.” Tesla builds several thousand Optimus humanoid robots during 2025. | 2025-01-29 Tesla Q4 2024 earnings call transcript (Motley Fool) | 2025-12-31 | did not happen by the date By late July 2025 Tesla had built only 'in the hundreds' of Optimus units against a 5,000-unit plan, and the production line was paused for a redesign; 2026 trackers put the 2025 total in the hundreds. evidence |
| Mark Zuckerberg, CEO Meta | “2025 will be the year when it becomes possible to build an AI engineering agent that has coding and problem-solving abilities of around a good mid-level engineer.” An AI agent with the abilities of a good mid-level engineer becomes buildable in 2025. Said more bluntly on the Joe Rogan Experience (Jan 10, 2025). On the same call he hedged: 'I don't think you're going to see this year like an AI engineer that is extremely widely deployed.' | 2025-01-29 Meta Q4 2024 earnings call transcript (Meta IR) | 2025-12-31 | did not happen by the date No mid-level-engineer agent was announced in 2025; in July 2026 Zuckerberg was reported telling staff that AI-agent development had gone slower than Meta expected (reported). evidence |
| Mark Zuckerberg, CEO Meta | “the year when a highly intelligent and personalized AI assistant reaches more than 1 billion people, and I expect Meta AI to be that leading AI assistant” Meta AI reaches more than 1 billion people during 2025. Measured as monthly actives across Facebook, Instagram, WhatsApp and Messenger, where the assistant is built in. | 2025-01-29 Meta Q4 2024 earnings call transcript (Meta IR) | 2025-12-31 | happened At Meta's annual meeting on May 28, 2025 Zuckerberg said Meta AI had one billion monthly active users across the company's apps. evidence |
| Sam Altman, CEO OpenAI | (paraphrase) Altman said OpenAI would ship GPT-4.5 within weeks as its last non-chain-of-thought model, then release GPT-5 within months as a system unifying its o-series and GPT-series models. GPT-5, unifying OpenAI's model lines, ships within months of February 2025. Paraphrase of an X post; exact wording not independently fetched. | 2025-02-12 TechCrunch (reporting Altman's X post roadmap) | 2025-12-31 | happened GPT-4.5 shipped Feb 27, 2025 and GPT-5 launched Aug 7, 2025 as a unified system with built-in reasoning. evidence |
| CoreWeave (company guidance; CEO Michael Intrator) CoreWeave | (paraphrase) CoreWeave raised its full-year 2025 revenue guidance to $5.15-5.35 billion. CoreWeave's 2025 revenue lands between $5.15 and $5.35 billion. A supply-side miss (data-center delivery), not a demand miss; backlog was said to be unaffected. | 2025-08-12 CNBC (Q2 2025 results) | 2025-12-31 | walked back On Nov 10, 2025 CoreWeave cut the range to $5.05-5.15 billion, saying a third-party data-center developer was behind schedule and pushing revenue into 2026. evidence |
| Sam Altman, CEO OpenAI | (paraphrase) Altman said OpenAI expected to end 2025 above a $20 billion annualized revenue run rate and to grow to hundreds of billions of dollars in revenue by 2030. OpenAI exceeds a $20 billion annualized revenue run rate by end of 2025 (and reaches hundreds of billions by 2030). The 2030 'hundreds of billions' leg remains pending (due 2030-12-31). Same post cited more than $1.4 trillion of infrastructure commitments. | 2025-11-06 CNBC (reporting Altman's X post) | 2025-12-31 | happened OpenAI CFO Sarah Friar said in January 2026 that annualized revenue surpassed $20 billion in 2025, up from $6 billion in 2024. evidence |
| Elon Musk, CEO Tesla | “we'll be starting production in April next year” Cybercab robotaxi production begins at Giga Texas in April 2026. Included as a counter-example: a Tesla date that held, albeit with a slow-ramp caveat. | 2025-11-06 TechCrunch (Tesla annual shareholder meeting) | 2026-04-30 | happened On the Q1 2026 call (Apr 22, 2026) Musk said 'We have just started production of Cybercab', while warning initial output would be 'very slow' before ramping later in the year. evidence |
| Sam Altman, CEO, and Jakub Pachocki, chief scientist OpenAI | (paraphrase) OpenAI said it was tracking toward an intern-level automated AI research assistant by September 2026 and a fully automated 'legitimate AI researcher' by 2028. An intern-level automated AI researcher exists by September 2026; a fully autonomous AI researcher by 2028. Pachocki also said deep-learning systems may be 'less than a decade away from superintelligence.' | 2025-10-28 TechCrunch (OpenAI livestream, Oct 28, 2025) | 2026-09-30 | pending In March 2026 Pachocki told MIT Technology Review the September 2026 intern milestone was still the plan; no delivery announcement found as of Aug 31, 2026. evidence |
| Dario Amodei, CEO Anthropic | “I think it could come as early as 2026, though there are also ways it could take much longer.” 'Powerful AI' (smarter than a Nobel laureate across fields, able to act autonomously for days) could arrive as early as 2026. Essay page shows 'October 2024'; published Oct 11, 2024. Explicitly hedged ('as early as'). | 2024-10-11 Dario Amodei essay, 'Machines of Loving Grace' | 2026-12-31 | pending |
| Sam Altman, CEO OpenAI | “2026 will likely see the arrival of systems that can figure out novel insights. 2027 may see the arrival of robots that can do tasks in the real world.” AI systems that can produce novel insights arrive in 2026; robots doing real-world tasks arrive in 2027. Same post asserts '2025 has seen the arrival of agents that can do real cognitive work.' | 2025-06-10 Sam Altman blog, 'The Gentle Singularity' | 2026-12-31 | pending |
| Colette Kress, CFO (restating Jensen Huang's Oct 28, 2025 GTC figure) Nvidia | “we currently have visibility to a half a trillion dollars in Blackwell and Rubin revenue from the start of this year through the end of calendar year 2026” Nvidia books about $500 billion of Blackwell and Rubin revenue between January 2025 and December 2026. Huang first gave the $500 billion figure at GTC Washington on Oct 28, 2025; Kress said about $50 billion had shipped by the October quarter. | 2025-11-19 Nvidia Q3 FY2026 earnings call transcript (Motley Fool) | 2026-12-31 | pending |
| OpenAI (official blog; team led by Ilya Sutskever and Jan Leike) OpenAI | (paraphrase) OpenAI set itself the goal of solving the core technical challenges of superintelligence alignment within four years, dedicating 20% of its secured compute to a new Superalignment team. OpenAI will solve the core technical challenges of superintelligence alignment by mid-2027, dedicating 20% of secured compute to the effort. Paraphrase: the original post could not be re-fetched for exact wording. The same post said superintelligence 'could arrive this decade'. | 2023-07-05 OpenAI blog, 'Introducing Superalignment' | 2027-07-05 | walked back The Superalignment team was dissolved in May 2024, less than a year in, after both leads left; Jan Leike said safety culture had 'taken a backseat to shiny products.' evidence |
| David Solomon, CEO Goldman Sachs | “I wouldn't be surprised if in the next 12 to 24 months, we see a drawdown with respect to equity markets.” An equity-market drawdown occurs by October 2027, with much AI capital failing to earn returns. He added that 'a lot of capital that's deployed... will turn out to not deliver returns'. In November 2025 he put the likely drawdown at 10-20%. | 2025-10-03 CNBC (Italian Tech Week, Turin) | 2027-10-03 | pending |
| Dario Amodei, CEO Anthropic | “we could have AI systems equivalent to a 'country of geniuses in a datacenter' as soon as 2026 or 2027” AI systems collectively equivalent to a country of geniuses exist by 2027. Page shows only 'April 2025', so the day is set to 01. Amodei made the same 2026-or-2027 claim at Davos in January 2025. | 2025-04-01 Dario Amodei essay, 'The Urgency of Interpretability' | 2027-12-31 | pending |
| Andy Jassy, CEO Amazon | “In the next few years, we expect that this will reduce our total corporate workforce as we get efficiency gains from using AI extensively across the company.” Amazon's total corporate headcount falls within a few years because of AI-driven efficiency. 'Next few years' has no fixed date; due set to three years out. Amazon framed the cuts as culture and layers as well as AI. | 2025-06-17 Amazon (Jassy memo to employees) | 2028-06-17 | happened Amazon cut about 14,000 corporate roles in October 2025, in a memo calling this generation of AI 'the most transformative technology we've seen since the Internet', and announced further corporate cuts in January 2026. evidence |
| Shane Legg, co-founder and chief AGI scientist Google DeepMind | “I think there's a 50% chance that we have AGI by 2028.” Even odds of AGI by the end of 2028. Legg has held the 2028 estimate since about 2009-2011 and says he would not be surprised if it slips. | 2023-10-26 Dwarkesh Podcast transcript | 2028-12-31 | pending |
| Michael Burry, founder Scion Asset Management | (paraphrase) Burry estimated that by extending the useful lives of their servers, the hyperscalers would understate depreciation by about $176 billion between 2026 and 2028, overstating earnings. Big-tech AI earnings are overstated by roughly $176 billion of under-recognized depreciation over 2026-2028. His filing showed put options against Nvidia and Palantir as of Sept 30, 2025. Accounting-estimate claim; framed by Burry as an industry-wide practice. | 2025-11-10 CNBC (reporting Burry's X posts and 13F filing) | 2028-12-31 | pending |
| OpenAI, SoftBank, Oracle and MGX (joint announcement; Sam Altman, Masayoshi Son, Larry Ellison) Stargate Project | “The Stargate Project is a new company which intends to invest $500 billion over the next four years... We will begin deploying $100 billion immediately.” $100 billion of AI data-center investment deploys immediately and $500 billion within four years (by January 2029). Announced at the White House Jan 21, 2025 (SoftBank release dated Jan 22 JST). The four-year $500B leg is pending; the 'immediately' leg is the disputed part. | 2025-01-21 SoftBank Group press release, 'Announcing The Stargate Project' | 2029-01-21 | disputed By late 2025 OpenAI said Stargate had nearly 7 GW of planned capacity and 'over $400 billion in investment over the next three years' across sites, but most of that was commitments; only the first Abilene buildings were live, so whether $100 billion deployed 'immediately' is contested. evidence |
| Dario Amodei, CEO Anthropic | (paraphrase) Amodei warned AI could wipe out half of all entry-level white-collar jobs and push US unemployment to 10-20% within one to five years. Half of entry-level white-collar jobs disappear and unemployment reaches 10-20% by mid-2030. Direct quote from the interview: 'We, as the producers of this technology, have a duty and an obligation to be honest about what is coming.' Second report: https://fortune.com/2025/05/28/anthropic-ceo-warning-ai-job-loss | 2025-05-28 Axios interview (confirmed via Fortune) | 2030-05-28 | pending |
| Oracle (financial analyst meeting; co-CEO Clay Magouyrk) Oracle | (paraphrase) Oracle projected Oracle Cloud Infrastructure revenue of $166 billion in fiscal 2030, raising a $144 billion target given a month earlier. OCI revenue reaches $166 billion in the fiscal year ending May 2030, from roughly $10 billion in fiscal 2025. Primary transcript not fetched; figure corroborated by multiple outlets. The prior $144 billion target was given with Oracle's Sept 9, 2025 results. | 2025-10-16 Reuters (Oracle AI World analyst meeting) | 2030-05-31 | pending |
| Daron Acemoglu, Institute Professor of economics MIT | “no more than a 0.66% increase in total factor productivity (TFP) over 10 years” AI adds less than 1% to US total factor productivity over the decade to 2034 (a skeptic's counter-forecast to transformative-growth claims). NBER working paper dated May 2024; day set to 01. Included as a public economist's dated skeptical forecast. | 2024-05-01 NBER Working Paper 32487, 'The Simple Macroeconomics of AI' | 2034-05-31 | pending |
| Andrej Karpathy, AI researcher, OpenAI co-founder and former Tesla AI director Independent (Eureka Labs) | “It's the decade of agents” Reliable autonomous AI agents (continual learning, computer use, multimodality) take roughly a decade, i.e. until about 2035, not one year. A deliberate counter to the industry's 'year of agents' framing in 2025. | 2025-10-17 Dwarkesh Podcast transcript | 2035-12-31 | pending |
| Demis Hassabis, CEO Google DeepMind | (paraphrase) Hassabis said today's systems are nowhere near human-level AGI and that genuine AGI is probably five to ten years away, with roughly even odds within the decade. AGI arrives between 2031 and 2036, not in the next two years. Davos week, Jan 19-23, 2026; exact panel day not confirmed. Hassabis has given the same 5-10 year range since 2025. | 2026-01-22 Fortune (World Economic Forum, Davos 2026) | 2036-01-31 | pending |
| Kai-Fu Lee, CEO of Sinovation Ventures, former president of Google China Sinovation Ventures | (paraphrase) Lee said that within 15 to 25 years about 40 percent of the world's jobs would become displaceable by AI. Roughly 40% of the world's jobs are displaceable by AI within 15 to 25 years, i.e. by 2034-2044. Lee stressed 'displaceable' rather than certainly displaced; judged at the outer bound of his range. The same forecast appears in his 2018 book 'AI Superpowers'. | 2019-01-13 CBS News, 60 Minutes | 2044-01-13 | pending |
| Ilya Sutskever, co-founder and CEO of Safe Superintelligence; OpenAI co-founder Safe Superintelligence | “I think like 5 to 20.” A system that learns as well as a human and becomes superhuman arrives in 5 to 20 years, i.e. between 2030 and 2045. Answer to a direct question about his timeline; lower bound is 2030. | 2025-11-25 Dwarkesh Podcast transcript | 2045-12-31 | pending |
| David Cahn, partner Sequoia Capital | “AI's $200B question is now AI's $600B question.” AI revenue must eventually fill a roughly $600 billion gap to justify GPU capex; otherwise investors, not builders, bear the losses. Cahn framed speculative frenzies as normal for technology; no date attached to the reckoning. | 2024-06-20 Sequoia Capital essay | — | pending |
| Sam Altman, CEO OpenAI | “It is possible that we will have superintelligence in a few thousand days (!)” Superintelligence may arrive within a few thousand days (roughly the early-to-mid 2030s). No explicit deadline; 'a few thousand days' implies roughly 2030-2035. | 2024-09-23 Sam Altman essay, 'The Intelligence Age' | — | pending |
| Sam Altman, CEO OpenAI | “Are we in a phase where investors as a whole are overexcited about AI? My opinion is yes.” Investors are collectively overexcited about AI, i.e. the market is in a bubble phase, even though the technology is real. Remarks at a reporters' dinner the week of Aug 11, 2025; day approximate. He added: 'When bubbles happen, smart people get overexcited about a kernel of truth.' | 2025-08-14 CNBC (remarks to reporters, first reported by The Verge) | — | pending |
| Jeff Bezos, founder and executive chair Amazon | “This is a kind of industrial bubble... the benefits to society from AI are going to be gigantic.” AI is in an industrial-style bubble in which many investments fail but society keeps the infrastructure and the benefits. Distinguished 'industrial' from purely financial bubbles: 'AI is real, and it is going to change every industry.' | 2025-10-03 CNBC (Italian Tech Week, Turin) | — | pending |
| Ray Dalio, founder Bridgewater Associates | “There's a lot of bubble stuff going on... bubbles don't pop, really, until they are popped by tightness of monetary policy.” Markets are in a late-1990s-style bubble that will not burst until monetary policy tightens. He compared the setup to 1998-1999 and 1927-1928 while declining to predict timing. | 2025-10-28 CNBC (Future Investment Initiative, Riyadh) | — | pending |
| Jerome Powell, Chair US Federal Reserve | “these companies—the companies that are so highly valued—actually have earnings and stuff like that” The AI investment boom differs from the dot-com bubble because today's highly valued companies have earnings and business models. He continued: in the 1990s 'these were ideas rather than companies... So it's really a different thing.' A central banker's bubble denial. | 2025-10-29 Federal Reserve, FOMC press conference transcript | — | pending |
| Jensen Huang, CEO Nvidia | “There's been a lot of talk about an AI bubble. From our vantage point, we see something very different.” AI demand is not a bubble; it reflects durable, multi-phase demand for accelerated computing. A bubble denial, logged alongside the bubble calls. | 2025-11-19 Nvidia Q3 FY2026 earnings call transcript (Motley Fool) | — | pending |
Rules: verbatim quotes only from primary sources (transcripts, company posts, filings) or top-tier reporting; paraphrases are labeled; statements are judged against their own stated dates. Nothing here is about a person's character. JSON.