A small technology-owning class substantially controls access to reproduction, enhancement, work, or life-sustaining infrastructure by 2126.

Current estimate

Resolution criteria

2035

The full test below is met by 31 December 2035. Nine years is short for a conjunction that currently stands at one economy and zero domains. It would require a second G20 economy to develop, in the SAME domain, both an absent median-income route and an exclusion that persists five years unremedied - and the five-year persistence clause alone consumes more than half the remaining window, so the qualifying episode would have to be underway already. The two live trends run the other way: American insulin is being remedied and India's public coverage is expanding.

2050

The full test below is met by 31 December 2050. Twenty-four years is long enough for a capability now in early clinical or commercial form - a cell therapy, a gene therapy, an enhancement modality - to reach supply at scale and to be priced. It is not obviously long enough for a public backstop to fail to arrive, since the two documented cases of a public body reaching an expensive frontier therapy, Medicare's 1972 entitlement and NICE's Casgevy fund, both arrived within a few years of the capability itself.

2100

The full test below is met by 31 December 2100. Seventy-four years carries most of the probability, and the reason is compositional rather than technological: it is long enough for a domain with NO tradition of public provision anywhere to become important. No G20 health system funds cognitive enhancement, and three-quarters of British fertility patients already self-pay. A century is also long enough for the fiscal and political basis of universal provision to weaken, which is the scenario in which this claim resolves TRUE, and which no current statistic measures.

2126

Resolves TRUE when, by 2126, for at least one of the four named domains and in at least two G20 economies, all three conditions hold together: (1) CONCENTRATION - five or fewer non-state entities supply 50 percent or more of provision, measured by revenue or by units delivered, as reported by a competition authority or national statistics office; (2) NECESSITY - a median-income household cannot obtain the capability by any route, where public financing of privately delivered provision, statutory or social insurance, or any non-market route reaching such a household DEFEATS this condition regardless of who owns the facilities; (3) GATEKEEPING - those entities set eligibility or price terms that exclude a substantial population from access, recorded by a court, a regulator, or a national statistics office, and that exclusion persists or recurs across at least five consecutive years without an effective remedy, or remains unremedied at the resolution date. Resolves FALSE if by 2126 no domain satisfies all three, or provision in every domain remains predominantly public, competitively supplied, or publicly financed.

Probability history

Sources

1. Two firms supply three-quarters of American dialysis, and the five largest supply 87 percent - the concentration condition is met today in one domain

   medpac.gov · 2025 · accessed Sep 12, 2026 · official-advisory-body

   MedPAC's March 2025 Report to Congress, chapter 5, states that the dialysis sector 'is highly consolidated, with two large dialysis organizations (LDOs) - Fresenius Medical Care and DaVita - dominating the industry. In 2023, these LDOs accounted for three-quarters of facilities and FFS Medicare treatments. Moreover, in 2023, the five largest dialysis organizations accounted for roughly 87 percent of facilities and FFS Medicare treatments.' The same chapter's provider table gives the two-LDO share of facilities as 75, 74 and 75 percent across the three years it reports. About 7,700 dialysis facilities served fee-for-service beneficiaries in 2023, and roughly half of Medicare beneficiaries on dialysis were in fee-for-service and half in Medicare Advantage. This clears this forecast's concentration condition - five or fewer non-state entities supplying 50 percent or more of provision, reported by an official body - with room to spare.

   Limit: The denominator is Medicare fee-for-service treatments and facility counts, not the whole national market, and MedPAC is a congressional advisory commission rather than a competition authority or a national statistics office - it is closer to the latter but is not strictly either, which is a weakness in this evidence against the criterion as written. A DISCREPANCY IS REPORTED RATHER THAN RESOLVED: the retrieval lane found combined two-firm shares ranging from 59.1 to 84 percent across secondary sources, differing by year, by segment (hemodialysis only versus all dialysis), and by whether facilities or treatments are counted; none of those were traceable to an authority and none is used here. Most importantly, concentration of PROVISION is not by itself the thing the forecast claims, and hf-evidence-0109 records why this particular domain fails the necessity condition outright.

   Read directly from the primary PDF by the master agent, not by a retrieval lane: the document was fetched with curl and converted with pdftotext after the lane's WebFetch returned unusable font-encoded text for this and four other government PDFs. The quoted sentences are at lines 147-153 of the extracted text and the 75/74/75 series is the 'Two LDOs' row of the provider table. A CORRECTION TO THE LANE IS RECORDED: the lane reported these as 2024 figures; the document says 2023 in both places. CORROBORATED INDEPENDENTLY AT LOWER STRENGTH: the US Renal Data System's provider chapters give the same two firms at approximately 60 percent of dialysis units and patients around 2009 and at 69 percent of patients in 65 percent of units around 2014, a different national data system reporting the same direction and a trajectory consistent with MedPAC's 2023 figures; those USRDS numbers are search-synthesised rather than fetched and their exact report years were not confirmed.

2. The two cleanest competition-authority concentration findings are not about making anything - they are about the layer that decides who gets it

   ftc.gov · 2025 · accessed Sep 12, 2026 · competition-authority

   Two findings come from actual competition authorities rather than trade press or market-research vendors. First, the FTC's July 2024 interim staff report states that 'the top three PBMs processed nearly 80 percent of the approximately 6.6 billion prescriptions dispensed by U.S. pharmacies in 2023' and that the six largest manage 'nearly 95 percent of all prescriptions filled in the United States'. Second, the CMA's cloud services market investigation found that Amazon Web Services and Microsoft each hold between 30 and 40 percent of United Kingdom cloud spend in 2024, against total UK cloud spend of 10.5 billion pounds, up from 9 billion in 2023, and is recommending Strategic Market Status for both. What is notable is what these two are: a pharmacy benefit manager manufactures no drug and a cloud provider treats no patient. The most authority-measured concentration in the economy sits in the layer that decides access rather than the layer that produces the thing.

   Limit: THE COMBINATION IS AN INFERENCE, NOT A CMA FIGURE. The CMA published two individual ranges and did not publish a combined share; adding them is this agent's arithmetic and is flagged as such. The FTC's own 'nearly 95 percent' figure carries no stated year in the source. Neither domain is straightforwardly one of the four this forecast names: prescription benefit management is an intermediary in life-sustaining supply rather than the supply itself, and cloud compute is infrastructure for work rather than work. Both are recorded because the criterion asks about control of ACCESS, and that is precisely what an intermediary layer holds.

   The FTC figures were fetched twice independently: once by the retrieval lane, and once directly by the master agent against the same FTC press release, which returned the sentences verbatim. A CORRECTION TO THE LANE IS RECORDED: the lane reported the 80 percent as a share of insulin prescriptions; the FTC text says all prescriptions dispensed by US pharmacies. The CMA figures are search-synthesised and then corroborated against a secondary article quoting the CMA verbatim; neither the lane nor the master successfully extracted text from the CMA summary PDF, and an earlier CMA provisional-findings page gave the same 'up to 40 percent each' against 2023 data, which is consistent rather than contradictory.

3. The gatekeeping condition has exactly one clean instance anywhere in the world, and it is American insulin

   ftc.gov · 2024 · accessed Sep 12, 2026 · regulator-and-state-enforcement

   On 20 September 2024 the FTC filed an administrative complaint against Caremark Rx, Express Scripts and OptumRx alleging they 'systematically excluded' lower-list-price insulins from formularies in favour of identical high-list-price, highly rebated versions. The complaint states the three 'together administer about 80% of all prescriptions in the United States', that Humalog's average list price rose from 21 dollars in 1999 to more than 274 dollars by 2017 - over 1,200 percent - and that 'by 2019 one out of every four insulin patients was unable to afford their medication'. Separately, the New York Attorney General found in 2023 that manufacturer list prices caused patients to ration insulin or forgo it, settling with Eli Lilly, Sanofi-Aventis and Novo Nordisk to cap cash price at 35 dollars a month for uninsured New Yorkers for five years; Minnesota reached materially identical settlements between February 2024 and January 2025. This is a concentrated private supplier set, setting eligibility and price terms, excluding a stated quarter of the affected population, recorded by a regulator.

   Limit: A filed administrative complaint is an allegation, not an adjudicated finding of liability, and the state settlements are consent agreements in which nothing is admitted. The remedy's shape also narrows what the evidence proves: capping the cash price for the UNINSURED concedes that the excluded population was the uninsured specifically, not a median-income insured household, which is the household the forecast's necessity condition names. Most importantly, this record must be read together with hf-evidence-0107, which records what happened to this case afterwards - the lane's own caveat was that the outcome was unchecked, and checking it changes the meaning of this instance substantially.

   The FTC figures were fetched directly by the master agent from the FTC press release, which returned all three quoted elements verbatim, after the retrieval lane fetched the same page independently. A CORRECTION TO THE LANE IS RECORDED: the lane reported the 80 percent as a share of insulin prescriptions; the FTC text says all prescriptions in the United States. The New York and Minnesota settlements are search-synthesised from the Attorneys-General press releases and were not fetched.

4. The world's single clean instance of gatekeeping is also its single clean instance of reversal, and the clock ran 22 months

   ftc.gov · 2026 · accessed Sep 12, 2026 · regulator

   The insulin case in hf-evidence-0106 did not stay an open allegation. Express Scripts settled with the FTC on 4 February 2026. On 14 July 2026 the FTC announced a settlement with Caremark Rx and Zinc Health Services requiring them to restructure drug pricing and pharmacy compensation, pass negotiated discounts to patients, cap members' out-of-pocket costs on insulin, and stop blocking pharmacies from using third-party patient assistance services; the FTC states the settlement 'locks in place up to 8.5 billion dollars in consumer savings over the next 10 years and unlocks up to 4.5 billion dollars in additional savings for patients over the same 10-year period from point-of-sale rebates'. In June 2026 the remaining matter against the OptumRx respondents was withdrawn from adjudication to consider a proposed consent agreement. Separately, California's CalRx partnership with Civica launches state-branded insulin glargine at about 11 dollars a pen from 1 January 2026, and Civica committed to a 30-dollar vial price nationally from the same date. Twenty-two months separate the complaint from the second settlement.

   Limit: A consent settlement is not an adjudication and admits nothing, so it is not proof the alleged exclusion occurred; symmetrically, it is not proof it did not. The savings figures are the FTC's own forward-looking estimate over ten years, not a measured outcome, and no figure was found for how many patients the caps actually reach. One matter remains unresolved. The CalRx price is a launch commitment for a product that had not been dispensed at volume at the time of writing, and it is one state. What this record establishes is the SPEED and DIRECTION of the institutional response, not that the underlying concentration changed - the same three firms still administer about 80 percent of American prescriptions, which is the point hf-evidence-0110 develops.

   The Caremark settlement page was fetched directly by the master agent from ftc.gov and returned the date, the required conduct and the savings figures verbatim; the Express Scripts date and the Optum withdrawal are corroborated across the FTC's own release and two independent law-firm alerts. THIS RECORD EXISTS BECAUSE A LANE CAVEAT WAS CHECKED RATHER THAN CARRIED: the retrieval lane reported the FTC matter as a filed complaint of unknown outcome and named checking it as the thing that would change its evidentiary weight. It did. The CalRx and Civica launch details are search-synthesised and were not fetched.

5. Government is already the majority financer of health in five of six sampled G20 economies, which is the condition the claim needs to be absent

   api.worldbank.org · 2023 · accessed Sep 12, 2026 · international-statistics-and-official-bodies

   Domestic general government health expenditure as a share of current health expenditure, 2023: Japan 84.83 percent, Germany 79.12, China 57.10, the United States 53.95, Brazil 44.13, India 39.02. Five of the six sampled G20 economies are majority-public financers of health, including the most market-oriented one. Two domain-level cases show the same thing operating: American dialysis, the most concentrated provision found anywhere in this cycle, sits on the Medicare end-stage renal disease entitlement created in 1972, which covers the great majority of Americans with kidney failure on the basis of the diagnosis rather than age or means; and in England, NICE listed Casgevy at 1,651,000 pounds a course and then secured access through the Innovative Medicines Fund for up to 460 eligible patients. In both cases a public body stands between a concentrated private supplier and the patient.

   Limit: A public share of EXPENDITURE is not the same as a public alternative available to a median-income household, and the adversarial lane identified this weakness in its own attack: Medicare is age and disability gated, Medicaid is means-tested below median income, and a median-income working-age American household is covered predominantly by private employer insurance, which this statistic does not isolate. Six countries are not nineteen. And the NICE case is double-edged - a 460-patient quota is a public body rationing access, which defeats the forecast's necessity condition on the letter while illustrating on the substance that access to a 1.65-million-pound therapy is being decided by somebody either way.

   The expenditure table was fetched by the retrieval lane directly from the World Bank API, returning JSON for all six countries at one call, and is the strongest-provenance quantitative block in this cycle. A near-duplicate United States figure of 55.17 percent for 2022 was surfaced separately and is a vintage difference rather than a contradiction; it is reported and not reconciled. The Medicare ESRD coverage share is DISPUTED IN ITS PROVENANCE and is treated carefully here: the lane attributed 'over 90 percent' to a Congressional Research Service product, and the master agent's own search traced the 92-93 percent figure to Institute of Medicine and Health Affairs material rather than to the CRS text, so the qualitative claim of near-universal entitlement is used and the precise percentage is not. The NICE figures are search-synthesised and were not fetched.

6. Antiretrovirals went from 14,000 dollars a year to under 100, and the concentration that made the first number possible lasted about fifteen years

   unaids.org · 2021 · accessed Sep 12, 2026 · international-organisation

   The closest historical analogue to this claim is HIV treatment, and it resolved against durable private control. UNAIDS records first-line triple therapy at roughly 14,000 dollars per patient per year in high-income countries around 1990, still roughly 10,000 to 15,000 around 2000, a median of about 90 to 100 dollars a year for first-line combinations in developing countries by 2007, and under 100 dollars per person per year in sub-Saharan Africa by 2018 for WHO-recommended tenofovir-based fixed-dose combinations, a further 26 percent fall since 2014. UNAIDS names two mechanisms: civil-society activism around TRIPS flexibilities and compulsory licensing, and generic manufacturer entry, with about 80 percent of generic antiretrovirals procured for low- and middle-income countries now made in India. The period of durable, patent-protected, expensive concentration runs from approximately 1987 to the early-to-mid 2000s - on the order of fifteen to eighteen years, well inside this forecast's hundred-year horizon.

   Limit: The mechanism that ended this concentration was generic entry, which requires a substitutable manufacturing path - and the lane flagged, correctly, that the historical base rate cannot settle whether an enhancement or infrastructure technology in 2126 would have an equivalent path. This is the single most important boundary on this evidence: biologics, cell therapies and services such as in-vitro fertilisation do not genericise the way a small molecule does, and a capability delivered as a service or a device may have no generic analogue at all. Procurement volumes through PEPFAR and the Global Fund were named by the lane as a third mechanism but were not verified this cycle and are not relied on here.

   The UNAIDS feature story was fetched directly by the adversarial lane and supplies the 1990, 2000 and 2018 figures and both named mechanisms; the MSF price-reduction series corroborates independently and was searched rather than fetched. The 2007 median is a search-tool summary of UNAIDS and WHO material and was not fetched. The lane recorded no conflict between UNAIDS and MSF on these figures.

7. The regulator's marquee essential-medicine case ended with the merger closing and the conduct restricted - which is the equilibrium the claim's first condition needs

   ftc.gov · 2023 · accessed Sep 12, 2026 · competition-authority

   On 16 May 2023 the FTC sued to block Amgen's 27.8-billion-dollar acquisition of Horizon Therapeutics, arguing it would entrench monopoly pricing on two drugs with no direct approved competition - Tepezza at about 350,000 dollars for a six-month course and Krystexxa at about 650,000 dollars a year. The deal was not blocked and was not unwound. The FTC settled in September 2023 and approved a final order in December 2023 imposing behavioural remedies - fifteen-year restrictions on bundling and on future acquisitions in the two disease areas, a monitor, and annual compliance reports - and the acquisition closed. The adversarial lane recorded this as the point at which its own attack failed: it had searched the 'sues to block' release first and was close to reporting a blocking precedent. The pattern it exposes is that a competition authority can restrain what a concentrated firm DOES while leaving what it OWNS intact.

   Limit: One case is not a base rate, and behavioural remedies are not evidence that structural remedies never happen - the FTC has blocked and unwound mergers in other sectors, which this cycle did not survey. The inference drawn here is deliberately narrow: it shows that an authority documenting harm is compatible with the concentration persisting, which is what makes this forecast's first condition easy to satisfy and its third condition hard to satisfy DURABLY. Read alongside hf-evidence-0107, where the same authority restrained conduct in the insulin matter while the same three firms continued to administer about 80 percent of American prescriptions.

   Both FTC press releases were fetched directly by the adversarial lane, the complaint and the final order separately, which is what produced the correction: the outcome contradicted the lane's working assumption and it recorded the reversal explicitly rather than reporting the first result. This is one enforcement action in one jurisdiction and no base rate across cases was computed; the lane also noted it examined no competition authority outside the United States, which is a stated gap.

8. The top one percent's wealth share was flat for thirty years and then rose nine points - real, and not the variable this claim needs

   ourworldindata.org · 2026 · accessed Sep 12, 2026 · research-database

   United States top-one-percent wealth share, from the World Inequality Database series: 23.86 percent in 1980, 23.79 in 1990, 23.72 in 2000, 23.01 in 2010, 32.30 in 2020, 32.54 in 2024. The shape matters more than the level - roughly flat for three decades, then a rise of about nine points concentrated in the 2010s and early 2020s. A partially inspected multi-country read of the same file gives China rising from 15.8 percent in 1980 to 30.4 in 2024, Belgium falling from 17.7 to 15.5, and France fluctuating to about 27.7. Ownership is concentrating in the large economies, which is a real precondition for the claim even though it is not the claim.

   Limit: This is the weakest support in the cycle and is recorded with its weakness stated, because it is the statistic a reader is most likely to mistake for proof. Aggregate wealth share is a stock measure of who owns assets in general; the claim requires five or fewer firms controlling provision in ONE named domain, plus an access failure. A top percentile that owns more index funds and more real estate is fully consistent with no domain becoming a chokepoint. Forty-four years produced nine points, which is a trend and not a phase change, and the series says nothing about who supplies reproduction, enhancement, work or life-sustaining infrastructure.

   The United States series was fetched directly by the adversarial lane as a country-filtered CSV export from Our World in Data, which cites the World Inequality Database as its source; the lane switched to the filtered query only after whole-dataset fetches failed, and reports reading the US rows directly and the other countries through a tool summary, which is why the non-US figures are marked lower confidence here. A separately surfaced figure of 31.7 percent for the third quarter of 2025 comes from a different underlying dataset, probably the Federal Reserve Distributional Financial Accounts, and is recorded as a distinct non-reconciled series rather than a contradiction. A FIGURE WAS KILLED: the same file's Afghanistan series, flat at 23.6 to 23.8 percent across forty-four years, is not credible as an empirical estimate and is not used.

9. Reproduction and enhancement are the domains this claim is really about, and neither has a concentration figure any authority has published

   hfea.gov.uk · 2024 · accessed Sep 12, 2026 · regulator

   The concentration lane found no authority-grade supplier-concentration measurement for either reproduction or enhancement, and killed every candidate it found. For fertility, the HFEA reports that 66 of 107 licensed United Kingdom clinics, 62 percent, were privately owned in 2022/23 - an ownership split, not a named-supplier share of cycles or revenue, so it cannot support a five-firm fifty-percent test. A directed search for HFEA-reported cycle share by clinic group returned nothing. Private-equity consolidation of fertility clinics is documented commercially but against no regulator-published denominator. For enhancement, every figure located came from market-research vendors and was killed under the authority-grade rule. What the HFEA DOES document is the access side: NHS-funded IVF cycles fell from 35 percent in 2019 to 28 percent in 2024 nationally and 25 percent in England, meaning roughly three-quarters of patients self-pay, at average prices of 6,150 pounds in London and 4,764 in Manchester. Medical gases, insulin manufacture and labour platforms were killed the same way; hospital systems and water utilities were not examined at all.

   Limit: THIS IS ABSENT EVIDENCE, NOT NEGATIVE EVIDENCE, and the distinction decides how it may be used. That no competition authority has measured concentration in fertility services or enhancement does not mean those markets are competitive; it means nobody with the standing this criterion requires has looked, which is itself a finding about the criterion rather than about the world. The UK fertility funding series is genuine regulator data on access, but the rationing it documents is a public commissioner withdrawing its own funding rather than a concentrated private supplier excluding anyone, so it does not satisfy the gatekeeping condition as written - while being, on the substance, the closest thing in this cycle to the future the claim describes.

   The HFEA ownership figure and the NHS-funding series are search-synthesised from HFEA publications and were not fetched; the lane states this explicitly for each. The killed figures are named with their sources in the lane's report - Air Liquide, Linde and Air Products medical-gas shares from three market-research vendors, cancer gene-therapy shares from four more, an insulin figure from a JAMA study reachable only through a trade-press summary, a United Kingdom food-delivery split from a blog whose three shares summed to about 99 percent and which was killed for internal implausibility as well as sourcing, and a European cloud revenue figure of 65 to 70 percent that could not be traced to any Commission document.