Palantir

Palantir's bespoke Gotham deployments for American intelligence were encoded into platform primitives: ontology, object models, permissions, workflow engines, provenance tracking. Those primitives became Foundry. Jesse Zhang, CEO of Decagon, writing on X on 11 August 2026, treats the sequence as the reference case for a services-to-product transition: once deployments standardised, gross margin climbed into the 80 percent range and Palantir left the Forward Deployed Engineering motion for a named-account sales model. Shyam Sankar, its CTO, supplies the formula, "FDEs eat pain and excrete product." Joe Lonsdale concedes that nearly twenty years of reputation as a consulting firm in disguise rested on an accurate observation.

The name resurfaces because Anthropic and OpenAI have built enterprise deployment arms explicitly modelled on it, which Zhang notes was a motive for criticism until recently, since nothing in the underlying economics changed. Ethan Mollick turned the same pattern into a test of the frontier labs' own claims on 10 May 2026: as long as human engineers are needed to install AI inside client organisations, white-collar work is not imminently replaced.

The second thread is French. Asma Mhalla, writing on 20 September 2026, uses Palantir at the domestic intelligence service as the illustration of a volatile sovereignty doctrine: in place since 2015, presented from the outset as transitional, with a sovereign alternative launched in 2021 whose second tranche was still not finalised in May 2026.