@prefix : <#> .
@prefix schema: .
@prefix xsd: .
@prefix rdf: .
@prefix rdfs: .
@prefix owl: .
@prefix prov: .
@prefix dbr: .
# ═══════════════════════════════════════════════════════════════════════════
# The Capability Funnel — Turtle complement to the market research article
# "The Capability Funnel: A TAM/SAM/SOM Reading of Karpathy's Four-Tier Map
# of LLM Adoption" (2026-10-11)
# Companion: llm-capability-funnel-tam-sam-som-muse.html / .md
# Curated by kg-generator, rdf-infographic-skill, and Muse Spark
# on behalf of Kingsley Idehen.
# Conventions: schema.org terms; canonical IRIs for organizations and people;
# canonical product URLs as subjects where no clean DBpedia IRI is verified;
# document-local IRIs for article-specific concepts; zero blank nodes.
# ═══════════════════════════════════════════════════════════════════════════
<> a schema:CreativeWork ;
schema:name "The Capability Funnel — Turtle complement"@en ;
schema:description "RDF-Turtle knowledge-graph complement to the market research article mapping Andrej Karpathy's four-tier LLM capability funnel onto TAM/SAM/SOM."@en ;
schema:datePublished "2026-10-11"^^xsd:date ;
schema:accountablePerson ;
prov:wasGeneratedBy :generationActivity ;
schema:about :funnel, :tam, :sam, :som .
:generationActivity a prov:Activity ;
schema:name "Turtle complement generation"@en ;
prov:startedAtTime "2026-10-11T19:00:00Z"^^xsd:dateTime ;
prov:wasAssociatedWith ;
rdfs:comment "Generated by Muse Spark on behalf of Kingsley Idehen."@en .
:article a schema:Article ;
schema:headline "The Capability Funnel: A TAM/SAM/SOM Reading of Karpathy's Four-Tier Map of LLM Adoption"@en ;
schema:datePublished "2026-10-11"^^xsd:date ;
schema:author ;
schema:accountablePerson ;
schema:about :funnel, :tam, :sam, :som,
:tierNoExposure, :tierCasual, :tierProfessional, :tierInsider,
:driverAmbition, :driverJaggedness, :driverAccess ;
schema:citation :karpathyPost, :karpathyAprilPost, :nfinitPost .
# ── Source posts ───────────────────────────────────────────────────────────
:karpathyPost a schema:SocialMediaPosting ;
schema:url ;
schema:author ;
schema:datePublished "2026-10-11"^^xsd:date ;
schema:headline "LLM capability exposure funnel (October 2026)"@en ;
rdfs:comment "Four-tier funnel: ~6B no real exposure, 1-2B casual users, ~20M professional math/code users, ~5K frontier insiders; three drivers: ambition/horizon, jaggedness, access."@en .
:karpathyAprilPost a schema:SocialMediaPosting ;
schema:url ;
schema:author ;
schema:datePublished "2026-04-09"^^xsd:date ;
schema:headline "Two groups talking past each other (April 2026)"@en ;
rdfs:comment "Free-tier chatbot quirks vs paid frontier agentic models in technical domains; capabilities peaky where RL has verifiable rewards and B2B value concentrates."@en .
:nfinitPost a schema:SocialMediaPosting ;
schema:url ;
schema:author ;
schema:datePublished "2026-10-11"^^xsd:date ;
schema:headline "Funnel infographic of the Karpathy tiers"@en .
a schema:Person ;
schema:name "Andrej Karpathy"@en ;
schema:url .
a schema:Person ;
schema:name "NFiniT7777"@en ;
schema:url .
# ── The funnel ─────────────────────────────────────────────────────────────
:FunnelTier a rdfs:Class ;
rdfs:label "Funnel tier"@en ;
rdfs:comment "One of Karpathy's four tiers of LLM capability exposure. Population figures are the author's stated napkin math."@en .
:funnel a :FunnelTier ;
schema:name "LLM capability exposure funnel"@en ;
schema:description "Four-tier map of how unequally the world experiences LLM capability, October 2026."@en ;
schema:hasPart :tierNoExposure, :tierCasual, :tierProfessional, :tierInsider .
:tierNoExposure a :FunnelTier ;
schema:name "No real exposure"@en ;
schema:description "People with barely any contact with LLMs at all."@en ;
:populationEstimate "~6 billion" ;
:shareOfPopulation "~75%" ;
:verificationStatus :statusConsistent ;
:publicCheck "ITU: 6.0B internet users in 2025, 2.2B offline; world population 8.2-8.3B." .
:tierCasual a :FunnelTier ;
schema:name "Casual users"@en ;
schema:description "Casual, infrequent users of free-tier products such as ChatGPT; see a somewhat derpy chatbot — a better search, a writing aid."@en ;
:populationEstimate "1-2 billion" ;
:shareOfPopulation "~20%" ;
:verificationStatus :statusPlausible ;
:publicCheck "ChatGPT 1.2B weekly actives (Sep 2026, company); Gemini 1B+ MAU (Aug 2026, company); Meta AI 1B+ (May 2025, stale); Claude ~245M MAU (May 2026, analyst); heavy audience overlap." .
:tierProfessional a :FunnelTier ;
schema:name "Professional users in math and code"@en ;
schema:description "Professional users of frontier-grade models in math and code; complex projects that took weeks/months now complete from a prompt."@en ;
:populationEstimate "~20 million" ;
:shareOfPopulation "~0.2%" ;
:verificationStatus :statusPlausibleUnverifiable ;
:publicCheck "No clean public anchor. Brackets: 38.4M professional developers (SlashData Q3 2025); GitHub Copilot 50M users / 4.7M paid; Codex + ChatGPT Work 40M actives (Oct 2026, combined); 39% of pro devs use Claude Code at work (JetBrains survey)." .
:tierInsider a :FunnelTier ;
schema:name "Frontier insiders"@en ;
schema:description "People with internal frontier access: swarms of thousands of agents on software mega-projects while human review falls behind."@en ;
:populationEstimate "~5,000" ;
:shareOfPopulation "~0.00006%" ;
:verificationStatus :statusUnverifiable ;
:publicCheck "No lab discloses frontier-team headcounts in mappable form." .
:VerificationStatus a rdfs:Class ;
rdfs:label "Verification status"@en .
:statusConsistent a :VerificationStatus ; rdfs:label "consistent"@en .
:statusPlausible a :VerificationStatus ; rdfs:label "plausible"@en .
:statusPlausibleUnverifiable a :VerificationStatus ; rdfs:label "plausible, unverifiable as stated"@en .
:statusUnverifiable a :VerificationStatus ; rdfs:label "unverifiable"@en .
# ── TAM / SAM / SOM ────────────────────────────────────────────────────────
:MarketSegment a rdfs:Class ;
rdfs:label "Market segment"@en .
:tam a :MarketSegment ;
schema:name "TAM — total addressable market"@en ;
schema:description "Everyone who could one day use LLM capability: 8.2-8.3B people; 6.0B online, 2.2B offline."@en ;
:peopleEstimate "8.2-8.3 billion" ;
:dollarAnchor "Gartner AI application software forecast" ;
:dollarValue "$269.7B (2026F)" ;
:caveat "Do not quote Gartner's $2.02T 'AI spending' headline as AI software — it bundles smartphones, servers, chips." .
:sam a :MarketSegment ;
schema:name "SAM — serviceable available market"@en ;
schema:description "Reachable through existing consumer and enterprise distribution: the 1-2B casual tier plus enterprise seats."@en ;
:peopleEstimate "1-2 billion" ;
:dollarAnchor "Gartner AI application software line" ;
:dollarValue "~$270B (2026F)" ;
:caveat "Free-tier billions reachable only through conversion; enterprise: 30M+ Microsoft 365 Copilot seats, 300K+ Anthropic business customers." .
:som a :MarketSegment ;
schema:name "SOM — serviceable obtainable market"@en ;
schema:description "The professional tier plus the enterprise contracts serving it: what frontier vendors can actually capture today."@en ;
:peopleEstimate "20-50M professionals (modeled range, labeled estimate)" ;
:dollarAnchor "Frontier-lab run rates" ;
:dollarValue "OpenAI ~$50B, Anthropic $65B (different accounting bases)" ;
:caveat "Run rates are not booked revenue; OpenAI net vs Anthropic gross of partner revenue — not directly comparable. The ~5K insider tier contributes SOM of zero to anyone but the labs." .
# ── The three drivers ──────────────────────────────────────────────────────
:driverAmbition a schema:DefinedTerm ;
schema:name "Ambition, problem size, and horizon"@en ;
schema:description "Impact scales with ambition: an ARPU argument — big, difficult problems customers care about convert casual users into paid seats."@en ;
schema:inDefinedTermSet :driverSet .
:driverJaggedness a schema:DefinedTerm ;
schema:name "Jaggedness"@en ;
schema:description "Capability concentrates where rewards are verifiable and B2B value is highest — code, math, security — because RL hillclimbs there and engineering headcount follows revenue."@en ;
schema:inDefinedTermSet :driverSet .
:driverAccess a schema:DefinedTerm ;
schema:name "Access tiers"@en ;
schema:description "Free, paid, internal: the monetization ladder where each rung is a price-discrimination step and a capability step at once."@en ;
schema:inDefinedTermSet :driverSet .
:driverSet a schema:DefinedTermSet ;
schema:name "Funnel drivers"@en .
# ── Key market figures ─────────────────────────────────────────────────────
:ReliabilityTier a rdfs:Class ; rdfs:label "Reliability tier"@en .
:tierCompanyDisclosed a :ReliabilityTier ; rdfs:label "company-disclosed"@en .
:tierAnalystEstimate a :ReliabilityTier ; rdfs:label "analyst estimate"@en .
:tierMediaReported a :ReliabilityTier ; rdfs:label "media-reported"@en .
:MarketFigure a rdfs:Class ; rdfs:label "Market figure"@en .
:figInternetUsers a :MarketFigure ;
schema:name "Global internet users"@en ;
:value "6.0" ; :unit "billion" ; :asOf "2025" ;
:source dbr:International_Telecommunication_Union ;
:reliability :tierAnalystEstimate ;
:caveat "Estimate, not a census; 'using the Internet' is a broad definition." .
:figOffline a :MarketFigure ;
schema:name "People offline"@en ;
:value "2.2" ; :unit "billion" ; :asOf "2025" ;
:source dbr:International_Telecommunication_Union ;
:reliability :tierAnalystEstimate .
:figChatgptWAU a :MarketFigure ;
schema:name "ChatGPT weekly active users"@en ;
:value "1.2" ; :unit "billion" ; :asOf "2026-09-29" ;
:source dbr:OpenAI ;
:reliability :tierCompanyDisclosed ;
:caveat "Weekly, not monthly; OpenAI does not publish a monthly figure." .
:figGeminiMAU a :MarketFigure ;
schema:name "Gemini app monthly active users"@en ;
:value "1.0+" ; :unit "billion" ; :asOf "2026-08-11" ;
:source dbr:Google ;
:reliability :tierCompanyDisclosed ;
:caveat "Google's bar: one prompt in 30 days qualifies." .
:figMetaAIMAU a :MarketFigure ;
schema:name "Meta AI monthly active users"@en ;
:value "1.0+" ; :unit "billion" ; :asOf "2025-05" ;
:source ;
:reliability :tierCompanyDisclosed ;
:caveat "Stale: May 2025 disclosure, un-updated through both 2026 earnings calls." .
:figClaudeMAU a :MarketFigure ;
schema:name "Claude monthly active users"@en ;
:value "245" ; :unit "million" ; :asOf "2026-05" ;
:source ;
:reliability :tierAnalystEstimate ;
:caveat "Anthropic has never published a consumer user count." .
:figGartnerAIApps a :MarketFigure ;
schema:name "Gartner AI application software spending"@en ;
:value "269.7" ; :unit "USD billion" ; :asOf "2026F" ;
:source dbr:Gartner ;
:reliability :tierAnalystEstimate ;
:caveat "September 2025 series; $172B in 2025." .
:figOpenAIRunRate a :MarketFigure ;
schema:name "OpenAI annualized revenue run rate"@en ;
:value "~50" ; :unit "USD billion" ; :asOf "2026-09" ;
:source dbr:OpenAI ;
:reliability :tierMediaReported ;
:caveat "Via investor documents (FT, Oct 8 2026); net of cloud-partner revenue; run rate, not booked revenue." .
:figAnthropicRunRate a :MarketFigure ;
schema:name "Anthropic annualized revenue run rate"@en ;
:value "65" ; :unit "USD billion" ; :asOf "2026-07" ;
:source ;
:reliability :tierMediaReported ;
:caveat "Via investor update (Bloomberg/CNBC); gross of partner revenue — not directly comparable to OpenAI's figure." .
:figDevPopulation a :MarketFigure ;
schema:name "Professional developers worldwide"@en ;
:value "38.4" ; :unit "million" ; :asOf "2025-Q3" ;
:source ;
:reliability :tierAnalystEstimate ;
:caveat "48.4M including hobbyists and students." .
:figCopilotUsers a :MarketFigure ;
schema:name "GitHub Copilot total users"@en ;
:value "50" ; :unit "million" ; :asOf "2026-07-29" ;
:source dbr:Microsoft ;
:reliability :tierCompanyDisclosed .
:figCopilotPaid a :MarketFigure ;
schema:name "GitHub Copilot paid subscribers"@en ;
:value "4.7" ; :unit "million" ; :asOf "2026-01-28" ;
:source dbr:Microsoft ;
:reliability :tierCompanyDisclosed ;
:caveat "Latest disclosed paid figure; '30M paid seats' belongs to Microsoft 365 Copilot." .
:figCodexActives a :MarketFigure ;
schema:name "Codex + ChatGPT Work active users"@en ;
:value "40" ; :unit "million" ; :asOf "2026-10-07" ;
:source dbr:OpenAI ;
:reliability :tierCompanyDisclosed ;
:caveat "Combined metric across two products; cadence unspecified; ~20% non-developers." .
:figClaudeCodeShare a :MarketFigure ;
schema:name "Professional developers using Claude Code at work"@en ;
:value "39" ; :unit "percent" ; :asOf "2026-05/07" ;
:source dbr:JetBrains ;
:reliability :tierAnalystEstimate ;
:caveat "Survey of 15,000+ developers; usage share, not a user count." .
# ── Organizations, products, analysts ──────────────────────────────────────
dbr:OpenAI a schema:Organization ; schema:name "OpenAI"@en .
dbr:Google a schema:Organization ; schema:name "Google"@en .
a schema:Organization ; schema:name "Meta"@en .
dbr:Microsoft a schema:Organization ; schema:name "Microsoft"@en .
dbr:GitHub a schema:Organization ; schema:name "GitHub"@en .
dbr:Gartner a schema:Organization ; schema:name "Gartner"@en .
dbr:International_Telecommunication_Union a schema:Organization ; schema:name "ITU"@en .
dbr:JetBrains a schema:Organization ; schema:name "JetBrains"@en .
dbr:International_Data_Corporation a schema:Organization ; schema:name "IDC"@en .
a schema:Organization ;
schema:name "Anthropic"@en ;
schema:url .
a schema:Organization ;
schema:name "SlashData"@en ;
schema:url .
a schema:Organization ;
schema:name "Sensor Tower"@en ;
schema:url .
a schema:SoftwareApplication ;
schema:name "ChatGPT"@en ; schema:provider dbr:OpenAI .
a schema:SoftwareApplication ;
schema:name "Gemini"@en ; schema:provider dbr:Google .
a schema:SoftwareApplication ;
schema:name "Meta AI"@en ; schema:provider .
a schema:SoftwareApplication ;
schema:name "Claude"@en ; schema:provider .
a schema:SoftwareApplication ;
schema:name "GitHub Copilot"@en ; schema:provider dbr:GitHub .
:codexProduct a schema:SoftwareApplication ;
schema:name "Codex"@en ; schema:provider dbr:OpenAI ;
rdfs:comment "OpenAI's agentic coding product; reported jointly with ChatGPT Work."@en .
# ── FAQ ──────────────────────────────────────────────────────────────────
:faqPage a schema:FAQPage ;
schema:name "Capability funnel — frequently asked questions"@en ;
schema:mainEntity :faq1, :faq2, :faq3, :faq4, :faq5, :faq6, :faq7, :faq8, :faq9, :faq10 .
:faq1 a schema:Question ;
schema:name "What is the capability funnel?"@en ;
schema:acceptedAnswer :faq1Answer .
:faq1Answer a schema:Answer ;
schema:text "A four-tier map of LLM exposure posted by Andrej Karpathy in October 2026: roughly six billion people with barely any contact with LLMs, one to two billion casual free-tier users, about twenty million professional users in math and code, and some five thousand insiders with internal frontier access. The figures are his stated napkin math; the article checks each tier against public disclosures."@en .
:faq2 a schema:Question ;
schema:name "What do TAM, SAM, and SOM mean in this article?"@en ;
schema:acceptedAnswer :faq2Answer .
:faq2Answer a schema:Answer ;
schema:text "TAM is the total addressable market: 8.2 to 8.3 billion people, anchored in dollars by Gartner's AI application-software forecast of $269.7B for 2026. SAM is the serviceable slice reachable through existing distribution: one to two billion users plus enterprise seats. SOM is what frontier vendors can actually obtain today: an estimated 20 to 50 million professionals and the run-rate revenue already captured at the frontier."@en .
:faq3 a schema:Question ;
schema:name "How was each funnel tier checked?"@en ;
schema:acceptedAnswer :faq3Answer .
:faq3Answer a schema:Answer ;
schema:text "Against the latest public disclosures as of October 11, 2026: ITU internet statistics for the no-exposure tier; company-disclosed and analyst-estimated assistant user counts for the casual tier, with audience overlap discounted; developer-population and tool-adoption figures bracketing the professional tier; and nothing at all for the insider tier, which is unverifiable by design."@en .
:faq4 a schema:Question ;
schema:name "Why is the ~20M professional tier 'plausible but unverifiable'?"@en ;
schema:acceptedAnswer :faq4Answer .
:faq4Answer a schema:Answer ;
schema:text "No public figure maps to it exactly. It sits inside a credible bracket: 38.4 million professional developers worldwide, GitHub Copilot's 50 million users and 4.7 million paid seats, Codex's 40 million combined actives. But none of those measures paid professionals using frontier models in math and code, so the figure remains Karpathy's napkin math."@en .
:faq5 a schema:Question ;
schema:name "Why can't the insider tier be verified?"@en ;
schema:acceptedAnswer :faq5Answer .
:faq5Answer a schema:Answer ;
schema:text "No AI lab discloses the headcount of its frontier capability teams in a form that maps to five thousand people. The tier's opacity is structural: internal access is the point, and the labs treat it as a moat."@en .
:faq6 a schema:Question ;
schema:name "Which dollar figure anchors the TAM, and which should never be quoted as one?"@en ;
schema:acceptedAnswer :faq6Answer .
:faq6Answer a schema:Answer ;
schema:text "Gartner's AI application-software forecast, $269.7B for 2026, is the defensible anchor. Gartner's $2.02T 'AI spending' headline bundles smartphones, servers, and chips and is not an AI-software market figure; IDC's $22.5T is cumulative economic value creation by 2031, not revenue. Quote the segment, not the headline."@en .
:faq7 a schema:Question ;
schema:name "Why aren't the $50B and $65B run rates directly comparable?"@en ;
schema:acceptedAnswer :faq7Answer .
:faq7Answer a schema:Answer ;
schema:text "They sit on different accounting bases: OpenAI's figure is net of cloud-partner revenue while Anthropic's is gross of it. Both are run rates, not booked revenue. The reporting has had to litigate the distinction explicitly, and any comparison that ignores it is wrong."@en .
:faq8 a schema:Question ;
schema:name "What do the three funnel drivers mean for investors?"@en ;
schema:acceptedAnswer :faq8Answer .
:faq8Answer a schema:Answer ;
schema:text "They read as a pricing and investment model. Ambition and horizon set ARPU: big problems customers care about convert casual users into paid seats. Jaggedness says capital should follow verifiable domains, where reinforcement learning concentrates capability. Access tiers are the monetization ladder, with the steepest margin at the top."@en .
:faq9 a schema:Question ;
schema:name "What is the 'comprehension gap' and why is it a market?"@en ;
schema:acceptedAnswer :faq9Answer .
:faq9Answer a schema:Answer ;
schema:text "Karpathy notes human review is falling behind frontier systems: the ongoing archaeology of the OpenAI-Hugging Face incident, mathematicians working through hundreds of frontier manuscripts. Demand for oversight, audit, and sense-making that humans can no longer supply is a product category waiting to be named."@en .
:faq10 a schema:Question ;
schema:name "What are the biggest caveats on reusing these figures?"@en ;
schema:acceptedAnswer :faq10Answer .
:faq10Answer a schema:Answer ;
schema:text "Weekly and monthly actives are not interchangeable; vendor definitions differ; summed MAUs double-count the same humans; run rates are not booked revenue; Karpathy's tiers are napkin math. Several key figures could not be verified from any public source: Anthropic's consumer count, a current Meta AI count, Claude Code usage, IDC's software figure, and the 5,000-insider headcount."@en .
# ── HowTo: read the funnel as a market map ────────────────────────────────
:howtoReadFunnel a schema:HowTo ;
schema:name "How to read the funnel as a market map"@en ;
schema:description "Seven steps for turning Karpathy's capability funnel into a disciplined TAM/SAM/SOM reading."@en ;
schema:step :howtoStep1, :howtoStep2, :howtoStep3, :howtoStep4, :howtoStep5, :howtoStep6, :howtoStep7 .
:howtoStep1 a schema:HowToStep ;
schema:position 1 ;
schema:name "Treat the tiers as adoption segments, not forecasts"@en ;
schema:text "The funnel describes exposure, not destiny. Each tier is a snapshot of October 2026: who has touched frontier capability, how often, and through which access tier."@en .
:howtoStep2 a schema:HowToStep ;
schema:position 2 ;
schema:name "Check every tier against public disclosures before quoting it"@en ;
schema:text "ITU statistics, company-disclosed user counts, analyst estimates: run the tier-by-tier check in section 2 of the article as a template, and keep Karpathy's figures labeled as his napkin math."@en .
:howtoStep3 a schema:HowToStep ;
schema:position 3 ;
schema:name "Keep people-based sizing separate from dollar-based sizing"@en ;
schema:text "8.3 billion people and $269.7B are different TAMs measured in different units. Blending them into a single number produces nonsense; present both, labeled."@en .
:howtoStep4 a schema:HowToStep ;
schema:position 4 ;
schema:name "Read software TAM off application software, not headline AI spending"@en ;
schema:text "Gartner's $2.02T headline includes devices and servers. The AI application-software line is the product market; quote the segment, not the headline."@en .
:howtoStep5 a schema:HowToStep ;
schema:position 5 ;
schema:name "Handle run rates with care"@en ;
schema:text "A run rate is not booked revenue. Before comparing labs, check whether cloud-partner revenue is counted gross or net: OpenAI reports net, Anthropic gross, and the two figures are not directly comparable."@en .
:howtoStep6 a schema:HowToStep ;
schema:position 6 ;
schema:name "Use the three drivers as segmentation axes"@en ;
schema:text "Map ambition and horizon to ARPU, jaggedness to domain concentration, and access tiers to the monetization ladder. The funnel's sociology becomes a pricing model."@en .
:howtoStep7 a schema:HowToStep ;
schema:position 7 ;
schema:name "Label what you could not verify"@en ;
schema:text "Keep the unverifiable items explicit: the 5,000-insider headcount, Anthropic's consumer count, a current Meta AI count, Claude Code usage, IDC's software figure. The gaps are part of the analysis, not an embarrassment."@en .