Jake Lawrence · STS / Political Economy of AI · Invisible Infrastructure theme
OpenAI's counterparty concentration as an interactive dependency graph. The systemic risk is the graph, and no single filing composes it, so no category system renders the risk. Filter by disclosure source and watch the rest go dark.
An interactive radial dependency graph of OpenAI's counterparty concentration, and an Invisible Infrastructure reading of it. OpenAI sits pinned at the center; twenty-six entities and thirty relationships fan out across three rings by tier, each edge encoding two independent axes (stroke color for how sure the audit is, dash for what the deal actually is) plus a width that grows with the dollars. Every edge carries a disclosureSource, the single filing, ratings action, or testimony where that one relationship is visible, and each of the thirty edges resolves to a distinct source. The thesis, after Bowker and Star: Oracle's 10-K shows the OpenAI backlog but not Ellison's margin pledges; the Warner Bros proxy shows the guarantee but not the S&P rating that undermines it; the graph is the only place these compose, which is exactly the invisible infrastructure the series is about. The signature interaction is the source filter: choose one document and watch the rest of the graph go dark, because no other filing is obliged to show them. A mobile relationship ledger renders the same thirty edges below 760px. Derived from an independent claim-by-claim fact-check; figures are time-stamped and age.
Seeing Like an AI Company reads one firm's proposal to classify the AI industry from the inside. One Load-Bearing Company reads the counterparty graph that same firm sits at the center of, and finds the concentration no single filing is obliged to show.
Classification as Infrastructure names the pattern: each category is locally honest and globally blind. One Load-Bearing Company makes it literal, every edge traces to the single filing where that relationship is visible, and no document composes the whole graph.
Two field-data readings of a legibility gap. The Legibility Gap measures what Ukraine's registry cannot see; One Load-Bearing Company measures what no counterparty's disclosure can see, because the systemic risk lives only in the graph between the filings.
What the State Keeps reads the federal state as a portfolio of held options evaluated from above. One Load-Bearing Company reads a private portfolio of counterparty exposures that no single participant can evaluate, because each one sees only its own slice.
Unratified shows infrastructure acquiring force with no ratifying moment. One Load-Bearing Company shows systemic risk acquiring mass with no composing document, an exposure that exists across the filings and inside none of them.
Back to the release table · See it on the network · Machine-readable catalog
I design and ship AI tools, full-stack apps, and data pipelines — end to end, to production. Tell me the problem in a sentence; I'll give you an honest read on fit within a day.
Work with me →