--- type: Analysis title: The Case Against the AI Bubble Thesis description: Profitable incumbents, real compounding revenue, the general-purpose-technology argument, and a long record of bubble calls running years early, checked against filings and primary documents. resource: https://aibubblequestion.com/case-against tags: [steelman, hyperscalers, gpt-argument, bubble-calls, base-rates] sources: - id: fortune-hyperscaler-cashflow title: "Fortune/BofA on hyperscaler cash flow (forecast, Dec 2025)" resource: https://fortune.com/2025/12/19/ai-hyperscalers-debt-issuance-bond-binge-cash-flow-capital-expenditures/ author: "Fortune" last_modified: 2025-12-19T00:00:00Z - id: msft-10k-fy2025-cashflow title: "Microsoft 10-K (FY2025)" resource: https://www.sec.gov/Archives/edgar/data/789019/000095017025100235/msft-20250630.htm author: "Microsoft" last_modified: 2025-07-30T00:00:00Z accession: "0000950170-25-100235" - id: googl-10k-fy2025-cashflow title: "Alphabet 10-K (FY2025)" resource: https://www.sec.gov/Archives/edgar/data/1652044/000165204426000018/goog-20251231.htm author: "Alphabet" last_modified: 2026-02-04T00:00:00Z - id: meta-10k-fy2025-cashflow title: "Meta 10-K (FY2025)" resource: https://www.sec.gov/Archives/edgar/data/1326801/000162828026003942/meta-20251231.htm author: "Meta" last_modified: 2026-01-29T00:00:00Z - id: amzn-10k-fy2025-cashflow title: "Amazon 10-K (FY2025)" resource: https://www.sec.gov/Archives/edgar/data/1018724/000101872426000004/amzn-20251231.htm author: "Amazon" last_modified: 2026-02-06T00:00:00Z - id: orcl-10k-fy2025-cashflow title: "Oracle 10-K (FY2025)" resource: https://www.sec.gov/Archives/edgar/data/1341439/000095017025087926/orcl-20250531.htm author: "Oracle" last_modified: 2025-06-18T00:00:00Z accession: "0000950170-25-087926" - id: hyperscaler-ocf-most-recent-filing title: "SEC filings, most recent 10-K for each hyperscaler (Microsoft and Oracle FY2026, Alphabet, Amazon and Meta FY2025): combined operating cash flow $634.9B against $464.6B capex, spanning January 2025 to June 2026 across five different fiscal calendars, not one period" resource: https://www.sec.gov/Archives/edgar/data/789019/000119312526323660/msft-20260630.htm author: "Microsoft; Oracle; Alphabet; Amazon; Meta" last_modified: 2026-07-29T00:00:00Z accession: "MSFT FY2026 0001193125-26-323660; ORCL FY2026 0001193125-26-277521" - id: goldman-ai-buildout-scale-may2026 title: "Goldman Sachs, Tracking Trillions (AI build-out scale)" resource: https://www.goldmansachs.com/insights/articles/tracking-trillions-the-assumptions-shaping-scale-of-the-ai-build-out author: "Goldman Sachs Global Institute" last_modified: 2026-05-01T00:00:00Z - id: goldman-genai-gdp-2023 title: "Goldman Sachs, Generative AI could raise global GDP by 7%" resource: https://www.goldmansachs.com/insights/articles/generative-ai-could-raise-global-gdp-by-7-percent author: "Goldman Sachs Research (Briggs & Kodnani)" last_modified: 2023-04-05T00:00:00Z - id: innovapath-systematic-review title: "InnovaPath systematic review" resource: https://innovapath.us/index.php/IN/article/view/185 author: "InnovaPath" - id: shiller-brookings-housing-bubble-2005 title: "Shiller, Bubble, Bubble, Where's the Housing Bubble?" resource: https://www.brookings.edu/articles/bubble-bubble-wheres-the-housing-bubble/ author: "Robert Shiller (Brookings)" last_modified: 2005-06-01T00:00:00Z - id: burry-intelligence-per-watt-arxiv title: "Saad-Falcon et al., Intelligence per Watt (arXiv:2511.07885)" resource: https://arxiv.org/abs/2511.07885 author: "Saad-Falcon, Narayan, et al." last_modified: 2026-08-07T00:00:00Z - id: chanos-99-not-97-benzinga title: "Benzinga on Chanos, '99-type moment'" resource: https://www.benzinga.com/markets/prediction-markets/26/07/60841557/enron-short-seller-jim-chanos-sees-a-1999-type-moment-in-ai-boom-includes-spcx-tsla-in-his-shorts author: "Benzinga" last_modified: 2026-07-31T00:00:00Z - id: goldman-ai-investment-1tn-aug2026 title: "Global AI Investment Is Forecast to Exceed $1 Trillion in 2026" resource: https://www.goldmansachs.com/insights/articles/global-investment-is-forecast-to-exceed-1-trillion-in-2026 author: "Goldman Sachs Research (Joseph Briggs)" last_modified: 2026-08-07T00:00:00Z generated: by: "claude/opus-5" at: 2026-08-30T00:00:00Z status: stable stale_after: 2027-02-28T00:00:00Z verified: - by: "claude/opus-5" at: 2026-08-30T00:00:00Z --- # The steelman * Who is spending: BofA forecast in December 2025 that combined operating cash flow at five profitable hyperscalers would grow from $378B (2023) to $577B (2025), with the group's debt-to-cash ratio falling.[^fortune-hyperscaler-cashflow] FY2025 is now a completed year for all five; reading their own 10-Ks (Microsoft, Alphabet, Meta, Amazon and Oracle, each on its own fiscal-year-end) gives an actual combined operating cash flow of $577.0B for FY2025 -- $136.16B, $164.71B, $115.80B, $139.51B and $20.82B respectively -- landing almost exactly on the forecast.[^msft-10k-fy2025-cashflow][^googl-10k-fy2025-cashflow][^meta-10k-fy2025-cashflow][^amzn-10k-fy2025-cashflow][^orcl-10k-fy2025-cashflow] The near-exact match is not a coincidence to be suspicious of: BofA published its forecast in December 2025, and by then Microsoft's FY2025 (ended 30 June 2025) and Oracle's FY2025 (ended 31 May 2025) had already closed and been filed, so two of the five inputs were already-known numbers, not forecasts, when the forecast was made.[^msft-10k-fy2025-cashflow][^orcl-10k-fy2025-cashflow][^hyperscaler-ocf-most-recent-filing] This is the site's own arithmetic summing five named filings, not a BofA figure; the debt-to-cash ratio claim was not independently recomputed this pass and is dropped rather than restated on the forecast's authority alone. * Revenue is real: Microsoft AI at $37B run rate +123% YoY; OpenAI's run rate has since topped $40B annualized; enterprise spend growing, though several of its headline figures trace to different-scoped surveys ([ai-demand-side](/current/ai-demand-side.md)). * General-purpose technology: Goldman's capex model reaches $1.6T/yr by 2031 in its baseline scenario -- explicitly labeled "a scenario-based framework... not a forecast of future spending," sensitive enough to assumptions that they "could shift figures by hundreds of billions of dollars."[^goldman-ai-buildout-scale-may2026] The oft-paired GDP figure is a different document from a different year: a 2023 Goldman research note put the level effect at "7% (or almost $7 trillion)" of global GDP growth *cumulative over a 10-year period* -- not $7 trillion added per year, and not the same analysis as the 2031 capex model.[^goldman-genai-gdp-2023] Even read correctly, the literature's modal path is overvaluation -> correction -> consolidation, not wipeout.[^innovapath-systematic-review] * Circularity rebuttals: today's AI equity stakes are disclosed and generally small beside cash flows, structured as equity plus offtakes rather than the vendor loans of 1999-2001. The comparison point itself needs a correction: a widely repeated "123% of vendors' pretax earnings" loan-book figure very likely describes five North American telecom-equipment vendors collectively in 1999, not Lucent alone -- a same-era, separately reported figure for Nortel specifically ("as much as 130% of equipment cost") shows the era's vendor-financing numbers were reported per-company and varied. The origin article for "123%" was not independently re-reached this pass. Whether today's AI equity stakes are proportionately "small" is itself an assertion, not a measured ratio on either side of this comparison. # Base rates on bubble calls (the timing caveat) * Greenspan's December 1996 "irrational exuberance": Nasdaq gained roughly another 300% before the March 2000 peak. * Housing warnings didn't arrive as one signal. Case and Shiller's academic diagnosis in 2002-03 preceded three to four more years of rising prices. Shiller's own plainest warning on the record is much later and much closer to the top: in June 2005, under a year before the actual peak, he called it "a bubble of unprecedented proportions" that would "end ugly."[^shiller-brookings-housing-bubble-2005] Treating "2002-04 warnings" as one uniform three-to-four-years-early signal overstates how early Shiller's own strongest warning actually was. * "Everything bubble" warnings ran 2013-2019 through a strong compounding decade. * A specific numeric base rate this page previously carried -- that manias persist 3-5 years after expert warnings begin and another 12-24 months after warnings become consensus -- could not be located anywhere in its cited source (Quinn & Turner's *Boom and Bust*) across two independent search passes, including three secondary treatments of the book and a broad open search; it is dropped rather than restated on an author-name citation alone. What holds without it: "it's a bubble" and "get out now" are different claims, and exiting years early has historically cost real money. * New sceptical voices entered the window this steelman has to answer: Michael Burry cited a real, independently retrievable paper -- "Intelligence per Watt" -- to argue frontier GPU demand is overstated relative to what small local models can already handle,[^burry-intelligence-per-watt-arxiv] and Jim Chanos put the moment "closer to a '99-type moment than a '97" in a 31 July 2026 appearance.[^chanos-99-not-97-benzinga] Neither is a filed fact; both are named, dated positions this page's counterarguments have to hold up against ([the Ed Zitron record](ed-zitron-record.md) tracks them alongside its own subject for balance). * A bull-case voice also entered the window this steelman can draw on: Goldman Sachs Research's Joseph Briggs modeled global AI-related investment at roughly $1 trillion in 2026, including $581 billion in the US -- a derived estimate, not a disclosed figure -- calling it "consistent with the 2%-5% of GDP peak investment impulses observed in prior general-purpose technology buildouts."[^goldman-ai-investment-1tn-aug2026] The same note flags the widely cited $794 billion hyperscaler-capex figure as itself mismeasured, by around $200 billion in each direction, globally and for the US -- Goldman's own reconciliation, not an audited one. This is a bank's model, not a filed fact, and it is weighed here on the same terms as the sceptical voices above. [^fortune-hyperscaler-cashflow]: Fortune/BofA on hyperscaler cash flow (forecast, Dec 2025) [^msft-10k-fy2025-cashflow]: Microsoft 10-K (FY2025) [^googl-10k-fy2025-cashflow]: Alphabet 10-K (FY2025) [^meta-10k-fy2025-cashflow]: Meta 10-K (FY2025) [^amzn-10k-fy2025-cashflow]: Amazon 10-K (FY2025) [^orcl-10k-fy2025-cashflow]: Oracle 10-K (FY2025) [^hyperscaler-ocf-most-recent-filing]: SEC filings, most recent 10-K for each hyperscaler (Microsoft and Oracle FY2026, Alphabet, Amazon and Meta FY2025): combined operating cash flow $634.9B against $464.6B capex, spanning January 2025 to June 2026 across five different fiscal calendars, not one period [^goldman-ai-buildout-scale-may2026]: Goldman Sachs, Tracking Trillions (AI build-out scale) [^goldman-genai-gdp-2023]: Goldman Sachs, Generative AI could raise global GDP by 7% [^innovapath-systematic-review]: InnovaPath systematic review [^shiller-brookings-housing-bubble-2005]: Shiller, Bubble, Bubble, Where's the Housing Bubble? [^burry-intelligence-per-watt-arxiv]: Saad-Falcon et al., Intelligence per Watt (arXiv:2511.07885) [^chanos-99-not-97-benzinga]: Benzinga on Chanos, '99-type moment' [^goldman-ai-investment-1tn-aug2026]: Global AI Investment Is Forecast to Exceed $1 Trillion in 2026