Datasets:

Threadbaire
/

research-articles

id string	title string	author string	date timestamp[s]	updated timestamp[s]	word_count int64	license string	filename string	canonical_url string	canonical_md string	arc_position string	core_claim string	tags list	evidence_sources list	related_articles list	related_dataset string
thesis	The Software Collapse Has Already Happened	Lida Liberopoulou	2026-01-30T00:00:00	2026-02-27T00:00:00	10,000	CC BY-SA 4.0	thesis.md	https://threadbaire.com/thesis.html	https://threadbaire.com/thesis.md	Root document. Everything after extends this.	The software industry's value-producing mechanisms have already collapsed. Implementation is no longer scarce. Incumbents deploy defensive measures that delay recognition but cannot reverse the structural shift. Value migrates to the context/integration layer.	[ "software collapse", "value capture", "open infrastructure", "context graphs", "platform capture", "AI industry" ]	[ "OpenClaw", "DeepSeek", "MCP", "mem0", "Klarna", "Salesforce", "Unity", "Reddit", "ChatGPT", "AAIF", "Linux Foundation" ]	[ "the-humans-wont-be-called-back", "who-holds-the-keys-part-1", "who-holds-the-keys-part-2", "who-holds-the-keys-part-3", "two-bridges-to-the-same-place", "tools-multiply-seats-collapse", "the-silicon-industrialists-part-1", "the-silicon-industrialists-part-2", "the-wrong-game", "a-tale-of-two-earn...	null
the-humans-wont-be-called-back	The Humans Won't Be Called Back	Lida Liberopoulou	2026-03-04T00:00:00	null	3,500	CC BY-SA 4.0	the-humans-wont-be-called-back.md	https://threadbaire.com/blog/posts/the-humans-wont-be-called-back.html	https://threadbaire.com/blog/posts/the-humans-wont-be-called-back.md	The human cost layer. What the collapse means for the people inside the organisations.	The prediction that companies will fire humans, discover AI produces bad work, and call the humans back is wrong — not because AI won't produce bad work, but because the institutional knowledge needed to diagnose the failures leaves with the people who are cut.	[ "institutional knowledge", "layoffs", "AI replacement", "skill formation", "human cost" ]	[ "Verizon DBIR", "METR developer productivity study", "Shen & Tamkin skill formation research", "NBER firm data", "Block", "Klarna", "Salesforce" ]	[ "thesis", "tools-multiply-seats-collapse" ]	null
who-holds-the-keys-part-1	Who Holds the Keys to the Agent Web — Part 1: Before the Doors Close	Lida Liberopoulou	2026-03-18T00:00:00	null	4,500	CC BY-SA 4.0	who-holds-the-keys-part-1.md	https://threadbaire.com/blog/posts/who-holds-the-keys-to-the-agent-web-part-1.html	https://threadbaire.com/blog/posts/who-holds-the-keys-to-the-agent-web-part-1.md	First affirmative case — what should be built instead. Part 1 of 2: where the open web closes.	Authentication — not discovery, not execution — is the primary capture point in the emerging AI agent layer. Identity vendors are positioning to become the default delegation layer for agent permissions.	[ "agent authentication", "identity vendors", "credential capture", "local-first", "delegation broker", "MCP" ]	[ "Stytch", "Auth0/Okta", "WorkOS", "Bitwarden MCP server", "Docker MCP Toolkit", "OAuth", "CLAUDE.md", "AGENTS.md", "llms.txt" ]	[ "thesis", "who-holds-the-keys-part-2", "who-holds-the-keys-part-3" ]	null
who-holds-the-keys-part-2	Who Holds the Keys to the Agent Web — Part 2: Who Gets the Receipt	Lida Liberopoulou	2026-03-25T00:00:00	null	6,000	CC BY-SA 4.0	who-holds-the-keys-part-2.md	https://threadbaire.com/blog/posts/who-holds-the-keys-to-the-agent-web-part-2.html	https://threadbaire.com/blog/posts/who-holds-the-keys-to-the-agent-web-part-2.md	Second affirmative case. Part 2 of 2: where the open web gets mapped.	The agent commerce infrastructure is being built now. The commercial graph — transaction history, legitimacy layer, reasoning traces — is the capture mechanism. Whoever holds the canonical record of what was agreed holds the most valuable layer.	[ "agent commerce", "commercial graph", "payment rails", "trust registries", "autonomous wallets", "discovery protocols" ]	[ "Google A2A/UCP", "Stripe MPP", "x402 Foundation", "Olas/Gnosis autonomous agents", "Stanford NegotiationArena", "Microsoft Magentic Marketplace", "Princeton agent manipulation research", "IETF Agent Identity and Discovery draft" ]	[ "thesis", "who-holds-the-keys-part-1", "who-holds-the-keys-part-3" ]	null
two-bridges-to-the-same-place	Two Bridges to the Same Place	Lida Liberopoulou	2026-04-14T00:00:00	null	3,500	CC BY-SA 4.0	two-bridges-to-the-same-place.md	https://threadbaire.com/blog/posts/two-bridges-to-the-same-place.html	https://threadbaire.com/blog/posts/two-bridges-to-the-same-place.md	Bridge post. Connects the January thesis to April 2026 evidence. Proves the model-as-commodity prediction with dated, sourced receipts.	The competition between frontier AI labs is itself the proof that the model layer is commoditising — the same structural pattern as the Netscape/Microsoft browser war, where the winner didn't get to keep the winnings.	[ "model commoditisation", "frontier labs", "convergence", "open-weight", "benchmark gap", "browser war analogy" ]	[ "Chen/Zaharia/Zou (Harvard Data Science Review, 2024)", "Kirgis/Tufekci et al. (Princeton, arXiv 2026)", "Gao/Liang/Guestrin (ICLR 2025)", "Anthropic September 2025 postmortem", "Stanford HAI AI Index 2025", "Nagle/Yue (SSRN, 2025)", "Perplexity enterprise data", "AAIF founding announcement" ]	[ "thesis", "tools-multiply-seats-collapse", "the-wrong-game" ]	null
tools-multiply-seats-collapse	Tools Multiply. Seats Collapse.	Lida Liberopoulou	2026-04-21T00:00:00	null	2,800	CC BY-SA 4.0	tools-multiply-seats-collapse.md	https://threadbaire.com/blog/posts/tools-multiply-seats-collapse.html	https://threadbaire.com/blog/posts/tools-multiply-seats-collapse.md	Third in descending-through-the-stack sequence. Thesis (all layers) → Two Bridges (model layer) → Tools Multiply (software category layer). First article with a practical sorting test.	Per-seat software pricing is a twenty-five-year artifact of the human bottleneck. AI agents sort the software market into two categories: tool-makers whose usage multiplies and coordination surfaces whose seat economics hollow out.	[ "per-seat pricing", "tool-makers", "coordination surfaces", "SaaS collapse", "agent adoption", "vendor sorting" ]	[ "Jensen Huang / Dwarkesh Patel interview", "HubSpot Prospecting Agent", "Adobe Firefly Enterprise", "Salesforce Agentforce", "Reuters SaaS wipeout data", "Atlassian earnings" ]	[ "thesis", "two-bridges-to-the-same-place", "the-humans-wont-be-called-back", "the-wrong-game" ]	null
who-holds-the-keys-part-3	Who Holds the Keys to the Agent Web — Part 3: Who Keeps the Record	Lida Liberopoulou	2026-05-09T00:00:00	null	5,500	CC BY-SA 4.0	who-holds-the-keys-part-3.md	https://threadbaire.com/blog/posts/who-holds-the-keys-to-the-agent-web-part-3.html	https://threadbaire.com/blog/posts/who-holds-the-keys-to-the-agent-web-part-3.md	Third in the triptych. Completes the agent web capture sequence: authentication (security) → commerce (efficiency) → post-transaction (accumulation). First practical procurement framework in the series.	The deepest capture in agent commerce occurs after the session closes, through autopilot services that accumulate operational context as a byproduct of doing the work competently. The risk is not that the vendor trains on your data — it is that the vendor becomes the only party that can explain how the work was done.	[ "post-transaction capture", "autopilot services", "outsourcing wedge", "memory lock-in", "procurement", "agent commerce", "portability" ]	[ "Harvey", "Sierra", "Writer", "Cogent", "Lindy", "Sequoia Capital", "Bessemer Venture Partners", "Cohen & Levinthal (absorptive capacity)", "GDPR/Data Act/DMA portability gaps", "EU AI Act high-risk logging", "Ethical Black Box proposal", "IETF SCITT", "VCAP/ADRP", "Singapore 2026 Model AI...	[ "thesis", "who-holds-the-keys-part-1", "who-holds-the-keys-part-2" ]	null
the-silicon-industrialists-part-1	The Silicon Industrialists — Part 1: AI's Gilded Age	Lida Liberopoulou	2026-05-27T00:00:00	null	7,500	CC BY-SA 4.0	the-silicon-industrialists-part-1.md	https://threadbaire.com/blog/posts/the-silicon-industrialists-part-1.html	https://threadbaire.com/blog/posts/the-silicon-industrialists-part-1.md	First in a new series ('The Silicon Industrialists'). Diagnosis of the economic forces shaping AI. Deepest structural analysis to date — moves below the software layer to explain why the entire ecosystem behaves the way it does.	AI's cost structure is industrial, not software. Its revenue mechanisms are dissolving. The collision between these two forces produces a double bind that pushes the entire AI economy toward regression — through infrastructure dependencies, venture capital, startup absorption, and the capture of open source.	[ "industrial economics", "value dissolution", "double bind", "venture capital", "infrastructure capture", "open source capture", "forward deployment", "model commoditisation", "Gilded Age" ]	[ "Anthropic/Blackstone/Goldman Sachs", "OpenAI/TPG/Bain Capital", "Google FDE", "Microsoft/Accenture", "SpaceX S-1", "Alphabet Q1 FY2026 SEC filing", "Nvidia FY2026 SEC filing", "PitchBook down-round data", "CB Insights mega-round data", "OECD AI funding data", "Dedale Intelligence", "Meritech ...	[ "thesis", "two-bridges-to-the-same-place", "tools-multiply-seats-collapse", "the-humans-wont-be-called-back", "who-holds-the-keys-part-1", "who-holds-the-keys-part-2", "who-holds-the-keys-part-3", "the-silicon-industrialists-part-2", "a-tale-of-two-earnings", "the-silicon-industrialists-part-3" ]	null
the-silicon-industrialists-part-2	The Silicon Industrialists — Part 2: Tech's Great Escape	Lida Liberopoulou	2026-06-19T00:00:00	null	11,000	CC BY-SA 4.0	the-silicon-industrialists-part-2.md	https://threadbaire.com/blog/posts/the-silicon-industrialists-part-2.html	https://threadbaire.com/blog/posts/the-silicon-industrialists-part-2.md	Second in the Silicon Industrialists series. Traces the infrastructure trap through three industries, the escape through three openings, and the institutional conditions that held the ladder in place. Establishes that the software economy was engineered, not natural.	Every major infrastructure technology before software followed the same path: massive capital, consolidation, capture. The personal computer, independent software, and the open internet broke that pattern — not because the technology got cheaper, but because specific institutional decisions opened the substrate before ...	[ "infrastructure trap", "consent decree", "IBM unbundling", "conditioned privatisation", "railroads", "electricity", "telephony", "dot-com crash", "institutional engineering", "double bind", "escape economics" ]	[ "1956 AT&T consent decree", "United States v. IBM", "Walker Report 1938", "Pacific Railway Act 1862", "Panic of 1893", "Samuel Insull", "PUHCA 1935", "Rural Electrification Administration", "Interstate Commerce Act 1887", "Hepburn Act 1906", "Kingsbury Commitment 1913", "Bell Labs transistor 1...	[ "thesis", "the-silicon-industrialists-part-1", "two-bridges-to-the-same-place", "tools-multiply-seats-collapse", "the-silicon-industrialists-part-3" ]	null
the-wrong-game	The Wrong Game - Why Xbox Is Answering Microsoft's Margin Question	Lida Liberopoulou	2026-07-03T00:00:00	null	1,800	CC BY-SA 4.0	the-wrong-game.md	https://threadbaire.com/blog/posts/the-wrong-game.html	https://threadbaire.com/blog/posts/the-wrong-game.md	Fourth standalone applied analysis. Applies the capital allocation stress thesis to a specific company's observable behaviour pattern. Framework operates underneath without being named on surface. Shortest piece in the corpus but complete — one company, one observation, one question, four monitoring points.	Microsoft voluntarily disclosed an internal margin metric for Xbox after twenty-five years of absorbing worse losses quietly — while AI returns remain visible only through aggregated segment-level signals. The question is not whether Xbox deserved a reset but why the only transparency window opened for the business uni...	[ "Microsoft", "Xbox", "capital allocation", "margin disclosure", "selective transparency", "AI infrastructure spending", "corporate patience", "accountability margin", "applied analysis" ]	[ "Xbox Wire Reset memo June 2026", "Xbox Wire console price update June 2026", "Microsoft FY26 Q3 earnings release", "Microsoft FY26 Q3 10-Q", "FTC v. Microsoft trial exhibits 2023", "Meta FY25 10-K Reality Labs segment", "Microsoft FY26 Q3 cash-flow statement" ]	[ "thesis", "two-bridges-to-the-same-place", "tools-multiply-seats-collapse", "the-silicon-industrialists-part-1", "the-silicon-industrialists-part-2", "a-tale-of-two-earnings" ]	null
a-tale-of-two-earnings	A Tale of Two Earnings — Microsoft, Apple, and the Circle Between Them	Lida Liberopoulou	2026-07-27T00:00:00	null	3,800	CC BY-SA 4.0	a-tale-of-two-earnings.md	https://threadbaire.com/blog/posts/a-tale-of-two-earnings.html	https://threadbaire.com/blog/posts/a-tale-of-two-earnings.md	Real-time applied analysis. First article built as a predictive experiment — six falsifiable predictions filed before earnings reports with a public scorecard. Applies the ouroboros and double bind to specific companies in specific quarters.	The AI infrastructure cycle has produced circular financing now visible in earnings reports — the supplier finances its own demand because the infrastructure costs more than the revenue base can fund. Six predictions test what Microsoft, Apple, and Amazon will disclose and what they will leave aggregated.	[ "circular financing", "ouroboros", "earnings analysis", "Microsoft", "Apple", "Amazon", "vendor financing", "capital allocation", "disclosure asymmetry", "falsifiable predictions" ]	[ "Nvidia-OpenAI $250B guarantee (WSJ July 2026)", "Alphabet Q2 2026 10-Q", "Alphabet Q2 2026 earnings release", "Microsoft Q3 FY2026 10-Q", "Microsoft FY2026 Q3 earnings webcast", "Apple Q2 FY2026 10-Q", "Xbox Wire Reset disclosure June 2026", "Amazon Anthropic convertible note/Q1 2026 filing" ]	[ "the-silicon-industrialists-part-1", "the-silicon-industrialists-part-2", "the-wrong-game", "a-tale-of-two-earnings-what-the-filings-cannot-show" ]	null
a-tale-of-two-earnings-what-the-filings-cannot-show	A Tale of Two Earnings — What the Filings Cannot Show	Lida Liberopoulou	2026-07-31T00:00:00	null	3,200	CC BY-SA 4.0	a-tale-of-two-earnings-what-the-filings-cannot-show.md	https://threadbaire.com/blog/posts/a-tale-of-two-earnings-what-the-filings-cannot-show.html	https://threadbaire.com/blog/posts/a-tale-of-two-earnings-what-the-filings-cannot-show.md	Companion and scoring post to 'A Tale of Two Earnings.' First verifiable predictive result with pre-registered pass/fail conditions. Strongest credibility signal the practice has produced.	Six structural claims about hyperscaler earnings disclosure, filed before the event with pass/fail conditions, all confirmed. The score matters less than what it reveals: three separations — investment/operation, investor/customer, revenue/capital — prevent investors from evaluating whether AI infrastructure demand is ...	[ "earnings analysis", "predictive scoring", "three separations", "ecosystem-financed demand", "disclosure gaps", "Microsoft", "Apple", "Amazon", "Anthropic", "OpenAI", "circular financing", "ouroboros" ]	[ "Microsoft FY2026 Q4 earnings release", "Microsoft FY2026 10-K", "Amazon Q2 2026 earnings release", "Apple fiscal Q3 2026 earnings release", "Alphabet Q2 2026 earnings release", "A Tale of Two Earnings (July 27 predictions)" ]	[ "a-tale-of-two-earnings", "the-silicon-industrialists-part-1", "the-silicon-industrialists-part-2", "the-wrong-game" ]	null
the-silicon-industrialists-part-3	The Silicon Industrialists — Part 3: The Return of the Bind	Lida Liberopoulou	2026-09-04T00:00:00	null	14,000	CC BY-SA 4.0	the-silicon-industrialists-part-3.md	https://threadbaire.com/blog/posts/the-silicon-industrialists-part-3.html	https://threadbaire.com/blog/posts/the-silicon-industrialists-part-3.md	Third in the Silicon Industrialists series. Traces how the separations that made computing's escape possible were gradually reversed, establishing that AI entered an already-reconstructed bind rather than creating one.	AI did not create the structure producing its concentration. It inherited it. The three separations that made the PC, independent software, and the open internet different from every previous infrastructure technology were gradually reversed through cloud recentralisation, subscription ownership, externalised technical...	[ "infrastructure reversal", "cloud recentralisation", "subscription ownership", "identity infrastructure", "gated distribution", "institutional decay", "double bind", "AT&T parallel", "SaaS concentration", "capability migration", "possibility space" ]	[ "AWS launch and pricing history", "Microsoft Azure/Active Directory/Entra ID", "Google Cloud/Gmail identity", "Apple App Store/Activation Lock", "Steam/Valve", "NTT DoCoMo i-mode", "Adobe subscription transition", "Autodesk perpetual licence end", "IBM SoftLayer acquisition", "HP Helion shutdown",...	[ "thesis", "the-silicon-industrialists-part-1", "the-silicon-industrialists-part-2", "the-wrong-game", "a-tale-of-two-earnings", "a-tale-of-two-earnings-what-the-filings-cannot-show", "tools-multiply-seats-collapse", "who-holds-the-keys-part-1" ]	null
compute-watch	The Compute Watch: Tracking the Rent on AI Chips	Lida Liberopoulou	2026-10-06T00:00:00	null	2,700	CC BY-SA 4.0	compute-watch.md	https://threadbaire.com/blog/posts/compute-watch.html	https://threadbaire.com/blog/posts/compute-watch.md	Hub for The Compute Watch, the practice's first standing research instrument: a monthly public record rather than a one-off argument. Explains what the watch tracks, why, the prediction it checks, and how readers can contribute. First piece designed to gather data from readers.	Much of the AI buildout is financed with debt that chip rent is meant to repay, while owners' accounts assume chips keep earning for five or six years and a better chip arrives every year or two. The rent that values, secures and will settle bets on these chips is set by a thin leftover market of hourly deals. Once the...	[ "compute futures", "GPU rental prices", "chip-backed lending", "depreciation", "server lifespans", "AI infrastructure financing", "thin markets", "pre-registered prediction", "reader contributions" ]	[ "ICE Futures U.S. notice Sept 2026", "CME Group compute futures announcement Aug 2026", "CFTC request for comment on compute derivatives", "Federal Register notice Aug 2026", "ICE/NATIVX energy-normalised compute futures", "Goldman Sachs AI debt research Aug 2026", "Morgan Stanley data-centre financing ...	[ "compute-watch-00", "compute-watch-01", "a-tale-of-two-earnings", "a-tale-of-two-earnings-what-the-filings-cannot-show", "the-silicon-industrialists-part-1", "follow-the-money-both-ways" ]	https://huggingface.co/datasets/Threadbaire/compute-watch
compute-watch-00	The Compute Watch - No. 0: The Baseline	Lida Liberopoulou	2026-10-06T00:00:00	null	1,800	CC BY-SA 4.0	compute-watch-00.md	https://threadbaire.com/blog/posts/compute-watch-00.html	https://threadbaire.com/blog/posts/compute-watch-00.md	First note of The Compute Watch. Sets the baseline readings that every later monthly note is compared against, and pre-registers the test with symmetric pass/fail conditions. Follows the pre-registration method of A Tale of Two Earnings.	On 2 October 2026 the newest chip (B200) rented for about 2.1 times the older H100 on Silicon Data and about 3.0 times on Ornn: the two indexes behind the planned futures disagree. The readings are consistent with a shortage. The prediction is judged only once a memory-price index shows the shortage easing; it is right...	[ "compute futures", "GPU rental prices", "baseline", "B200", "H100", "index disagreement", "chip-backed lending", "server lifespans", "GDDR6", "pre-registered prediction" ]	[ "Silicon Data GPU rental indices", "Ornn Compute Price Index", "ICE Futures U.S. notice Sept 2026", "CME Group announcement Aug 2026", "CFTC request for comment", "Nscale Form S-1", "Alphabet, Amazon, Meta, Microsoft, Oracle, CoreWeave annual reports", "Nvidia Form 8-K Aug 2026 (residual value guarant...	[ "compute-watch", "compute-watch-01", "a-tale-of-two-earnings-what-the-filings-cannot-show" ]	https://huggingface.co/datasets/Threadbaire/compute-watch
follow-the-money-both-ways	Follow the Money Both Ways - Four Predictions for Anthropic's S-1	Lida Liberopoulou	2026-10-23T00:00:00	null	1,900	CC BY-SA 4.0	follow-the-money-both-ways.md	https://threadbaire.com/blog/posts/follow-the-money-both-ways.html	https://threadbaire.com/blog/posts/follow-the-money-both-ways.md	Pre-registered disclosure test on Anthropic's S-1, filed before the public filing (first posted on X, 23 October 2026). Extends the A Tale of Two Earnings method from the hyperscalers' filings to the lab at the centre of their investments. First piece written in plain language for non-finance readers. Scored in Compute...	The companies Anthropic owes the most to (Google, Amazon, Microsoft, and Broadcom on its chip leases) are also its investors or possible future owners, and the figures reported about its biggest deals measure different things, so they cannot be added up. Four predictions test whether the S-1 lets a reader follow the mo...	[ "Anthropic", "IPO", "S-1", "disclosure", "pre-registered prediction", "AI infrastructure financing", "compute commitments", "supplier-investors", "Broadcom", "Amazon", "Alphabet", "Microsoft" ]	[ "Broadcom Form 10-Q Sept 2026", "Broadcom/Apollo/Blackstone financing announcement June 2026", "Amazon Q2 2026 10-Q and earnings release", "Anthropic-Amazon compute agreement April 2026", "Microsoft/Nvidia/Anthropic partnership Nov 2025", "Reuters on Google investment April 2026", "Reuters on Anthropic'...	[ "a-tale-of-two-earnings-what-the-filings-cannot-show", "a-tale-of-two-earnings", "compute-watch", "compute-watch-01", "income-or-promises" ]	https://huggingface.co/datasets/Threadbaire/compute-watch
income-or-promises	Income or Promises - Five Predictions for Microsoft, Alphabet and Amazon	Lida Liberopoulou	2026-10-27T00:00:00	null	4,400	CC BY-SA 4.0	income-or-promises.md	https://threadbaire.com/blog/posts/income-or-promises.html	https://threadbaire.com/blog/posts/income-or-promises.md	Pre-registered disclosure test on Microsoft, Alphabet and Amazon's July-September 2026 results, filed before the reports of 28 and 29 October. Returns the A Tale of Two Earnings method to the hyperscalers, written in the plain-language format of Follow the Money Both Ways. Meta is used for context only, as it reported ...	The largest AI spenders now spend more cash than their businesses bring in. They cover the gap with borrowing, new shares and bills signed for years ahead, while holding growing backlogs of promised revenue, part of it from AI labs they have invested in. Five predictions and one control test whether any improvement thi...	[ "Microsoft", "Alphabet", "Amazon", "Meta", "earnings analysis", "pre-registered prediction", "free cash flow", "capex", "leases", "backlog", "depreciation", "server lifespans", "AI infrastructure financing", "disclosure" ]	[ "Meta Q3 2026 earnings release, presentation and 10-Q", "Meta 424B5 prospectus supplements Oct 2026", "Meta Q3 2022 earnings release", "Alphabet Q2 2026 earnings release and 10-Q", "Microsoft FY2026 and FY2025 10-K", "Microsoft Q3 FY2026 10-Q", "Microsoft FY2026 Q4 earnings call", "Amazon Q2 2026 earn...	[ "income-or-promises-where-the-bill-went", "follow-the-money-both-ways", "a-tale-of-two-earnings", "a-tale-of-two-earnings-what-the-filings-cannot-show", "compute-watch" ]	https://huggingface.co/datasets/Threadbaire/compute-watch
income-or-promises-where-the-bill-went	Income or Promises - Where the Bill Went	Lida Liberopoulou	2026-10-31T00:00:00	null	4,100	CC BY-SA 4.0	income-or-promises-where-the-bill-went.md	https://threadbaire.com/blog/posts/income-or-promises-where-the-bill-went.html	https://threadbaire.com/blog/posts/income-or-promises-where-the-bill-went.md	Scoring post to 'Income or Promises'. Second scored pre-registration on the hyperscalers, after A Tale of Two Earnings. Four of five predictions held and one could not be tested; one pass is reported with a disclosed research miss and a method change: claims that something is not disclosed now require a search of each ...	The hyperscalers' cash did not improve: Alphabet's and Amazon's free cash flow worsened, and Microsoft's stayed positive while it sold investments to cover payouts and launched its first US bond sale in nine years. The cost of the build is increasingly recorded outside the two figures most readers follow, spending on e...	[ "Microsoft", "Alphabet", "Amazon", "Meta", "earnings analysis", "pre-registered prediction", "prediction scoring", "free cash flow", "capex", "leases", "backlog", "guarantees", "related parties", "Copilot disclosure", "useful life", "AI infrastructure financing", "disclosure" ]	[ "Alphabet Q3 2026 10-Q, earnings release and call", "Amazon Q2 and Q3 2026 10-Q and Q3 earnings release and call", "Microsoft FY27 Q1 10-Q, earnings release and call", "Microsoft FY26 Q3 and Q4 earnings calls", "Meta Q3 2026 earnings release and 10-Q", "Reuters on Microsoft bond sale Oct 2026", "CNBC on...	[ "income-or-promises", "follow-the-money-both-ways", "a-tale-of-two-earnings-what-the-filings-cannot-show", "compute-watch" ]	https://huggingface.co/datasets/Threadbaire/compute-watch
compute-watch-01	The Compute Watch - No. 1: Rents Up, Memory Barely Moved	Lida Liberopoulou	2026-11-02T00:00:00	null	3,100	CC BY-SA 4.0	compute-watch-01.md	https://threadbaire.com/blog/posts/compute-watch-01.html	https://threadbaire.com/blog/posts/compute-watch-01.md	Second note of The Compute Watch and the first monthly reading against the No. 0 baseline. Adds the standing sections promised at launch: change since last month, a loans table typed by what the lender is betting on, the strongest public counter-argument with an answer, and the reader question. Scores the four Anthropi...	In October, AI chip rents rose on both indexes and the newest chips rose fastest, so the B200's premium over the H100 widened (2.12x to 2.22x on Silicon Data, 2.99x to 3.13x on Ornn), while the GDDR6 memory price rose only 1.6%. That is consistent with a tightening shortage; the pre-registered test has not started. Sup...	[ "compute futures", "GPU rental prices", "B200", "H100", "index disagreement", "chip-backed lending", "GDDR6", "power constraints", "pre-registered prediction", "prediction scoring", "Anthropic", "S-1" ]	[ "Silicon Data GPU rental and RAM indices", "Ornn Compute Price Index", "TrendForce DRAMeXchange GDDR6 spot", "ICE OCPI B200 and H100 futures pages", "Anthropic Form S-1 Oct 2026", "SharonAI Form 8-K Oct 2026", "Nebius Form 6-K Oct 2026", "Bloomberg on Volta loan Oct 2026", "WSJ on North Star Oct 202...	[ "compute-watch", "compute-watch-00", "follow-the-money-both-ways", "income-or-promises-where-the-bill-went" ]	https://huggingface.co/datasets/Threadbaire/compute-watch

Threadbaire Research Articles

Independent research on AI economics and infrastructure: where the value in the AI economy is created, who keeps it, and who ends up paying for it. By Lida Liberopoulou (about · threadbaire.com).

Structural analysis of AI industry dynamics, software value collapse, and open infrastructure. Nineteen articles published between January and November 2026, available as raw markdown for analysis, citation, and AI-readable ingestion.

About

This dataset contains the complete text of the Threadbaire thesis and blog — a body of independent research arguing that AI has already collapsed traditional software value capture mechanisms, and mapping the capture infrastructure forming in its wake. All articles are published under CC BY-SA 4.0. Each piece is sourced, dated, and designed to be machine-readable from the ground up — raw markdown endpoints, citation blocks, and structured metadata are part of the publishing philosophy.

File	Title	Date	Words
`thesis.md`	The Software Collapse Has Already Happened	2026-01-30	~10,000
`the-humans-wont-be-called-back.md`	The Humans Won't Be Called Back	2026-03-04	~3,500
`who-holds-the-keys-part-1.md`	Who Holds the Keys to the Agent Web — Part 1: Before the Doors Close	2026-03-18	~4,500
`who-holds-the-keys-part-2.md`	Who Holds the Keys to the Agent Web — Part 2: Who Gets the Receipt	2026-03-25	~6,000
`two-bridges-to-the-same-place.md`	Two Bridges to the Same Place	2026-04-14	~3,500
`tools-multiply-seats-collapse.md`	Tools Multiply. Seats Collapse.	2026-04-21	~2,800
`who-holds-the-keys-part-3.md`	Who Holds the Keys to the Agent Web — Part 3: Who Keeps the Record	2026-05-09	~5,500
`the-silicon-industrialists-part-1.md`	The Silicon Industrialists — Part 1: AI's Gilded Age	2026-05-27	~7,500
`the-silicon-industrialists-part-2.md`	The Silicon Industrialists — Part 2: Tech's Great Escape	2026-06-19	~11,000
`the-wrong-game.md`	The Wrong Game - Why Xbox Is Answering Microsoft's Margin Question	2026-07-03	~1,800
`a-tale-of-two-earnings.md`	A Tale of Two Earnings — Microsoft, Apple, and the Circle Between Them	2026-07-27	~3,800
`a-tale-of-two-earnings-what-the-filings-cannot-show.md`	A Tale of Two Earnings — What the Filings Cannot Show	2026-07-31	~3,200
`the-silicon-industrialists-part-3.md`	The Silicon Industrialists — Part 3: The Return of the Bind	2026-09-04	~14,000
`compute-watch.md`	The Compute Watch: Tracking the Rent on AI Chips	2026-10-06	~2,700
`compute-watch-00.md`	The Compute Watch - No. 0: The Baseline	2026-10-06	~1,800
`follow-the-money-both-ways.md`	Follow the Money Both Ways - Four Predictions for Anthropic's S-1	2026-10-23	~1,900
`income-or-promises.md`	Income or Promises - Five Predictions for Microsoft, Alphabet and Amazon	2026-10-27	~4,400
`income-or-promises-where-the-bill-went.md`	Income or Promises - Where the Bill Went	2026-10-31	~4.300
`compute-watch-01.md`	The Compute Watch - No. 1: Rents Up, Memory Barely Moved	2026-11-02	~3,100

Structured Metadata

articles.json contains per-article metadata: title, date, word count, arc position, core claim, evidence sources, tags, and links to both the dataset files and canonical URLs at threadbaire.com.

from datasets import load_dataset
ds = load_dataset("Threadbaire/research-articles")

Arc

The thesis is the root document. Each blog post extends it by moving one layer deeper through the stack: the human cost of the collapse, then a triptych on the agent web — authentication capture, commerce capture, and post-transaction memory capture — followed by the model commodity layer and the software category sorting mechanism. The latest series, The Silicon Industrialists, moves below the software layer entirely. Part 1 diagnoses the industrial economics — arguing that AI's cost structure is industrial, not software, and that the collision with dissolving revenue models pushes the entire economy toward regression. Part 2 traces the infrastructure trap through railroads, electricity, and telephony, then follows the escape through three institutional openings that made the software economy possible, and asks what held the ladder in place. "The Wrong Game" applies the capital allocation stress thesis to a single company — observing that Microsoft voluntarily disclosed an internal margin metric for Xbox after twenty-five years of absorbing worse losses quietly, while AI returns remain visible only through aggregated segment-level signals. "A Tale of Two Earnings" is the first real-time predictive experiment — six falsifiable predictions filed before Microsoft, Apple, and Amazon report, testing what the circular financing mechanism will make visible in the filings and what it will leave aggregated. "What the Filings Cannot Show" scores all six predictions against the actual filings — all held — and introduces three separations (investment/operation, investor/customer, revenue/capital) as the analytical framework for why the circular financing mechanism remains invisible in standard disclosure, and proposes independently financed versus ecosystem-financed demand as a new analytical category. Part 3 of The Silicon Industrialists traces how the three separations that made the software economy possible — separable hardware, ownable software, and open distribution — were gradually reversed through cloud recentralisation, subscription ownership, externalised capability, gated distribution, joined identity, and the absence of successor institutional protections, establishing that AI entered an already-reconstructed bind rather than creating one. The Compute Watch, launched in October 2026, is a standing instrument rather than an argument: a monthly public record of what AI chips rent for, what lenders charge against them, and how long their owners say they last, with one pre-registered prediction checked against the numbers each month. Its figures and structured context are kept in a separate dataset, Threadbaire/compute-watch. "Follow the Money Both Ways" applies the same pre-registration method to Anthropic's S-1, filed before the document was public: four predictions about whether the filing will let a reader follow the money between Anthropic, the companies that fund it and the companies it pays. It is the first piece written in plain language for readers outside finance, and its predictions were scored in Compute Watch No. 1: all four held. "Income or Promises" returns the method to the hyperscalers in the same plain language: five predictions and one control, filed before Microsoft, Alphabet and Amazon reported on 28 and 29 October. Together they test whether any improvement in their cash is new income, or a cost moved into later bills, investment gains and promised revenue from AI labs they fund. “Where the Bill Went” scores them against the filings: four held and one could not be tested, with one pass reported alongside a research miss it discloses and a change to the method. It then follows the cost of the build into places the headline figures don’t show: guarantees of other companies’ data-centre rent, loans secured on servers, short leases kept off the balance sheet, and backlog from companies the sellers part-own. Compute Watch No. 1, the first monthly reading against the baseline, records rents rising fastest on the newest chips while the memory price barely moved, and supply held back by credit, power and memory at once; the prediction's test has not yet started.

Research Directions

The Silicon Industrialists series identified a dual-force mechanism — the double bind between industrial-scale infrastructure costs and value dissolution — that produces the same economic sequence across every shared infrastructure technology examined (railroads, electricity, telephony, computing). The series also identified the institutional conditions (consent decrees, mandated unbundling, conditioned privatisation) that broke this sequence for the computing industry specifically. Current research is investigating whether this mechanism provides the missing causal explanation for Kondratieff long waves (~50-60 year economic cycles associated with successive infrastructure technologies). The existing long-wave literature (Schumpeter, Perez, Freeman & Louçã) documents the empirical pattern and maps the installation-deployment sequence but has not converged on a structural mechanism that explains why every infrastructure technology follows the same financial shape. The double bind — industrial costs that cannot be avoided combined with value dissolution that erodes the revenue base funding the buildout — may be that mechanism, with the institutional conditions explaining the transition from crisis to deployment. This line of investigation is at an early stage. A full literature review and formal working paper will follow the completion of the Silicon Industrialists series. A preliminary research note establishing this hypothesis is available on SSRN: doi.org/10.2139/ssrn.7026619