Training a frontier model costs more than most companies can fund from product revenue, so a distinctive set of financing structures grew up around the leading labs. This concept collects the recurring patterns, each documented in the labs’ own words.
The first pattern is the capped-profit company. In March 2019 OpenAI announced OpenAI LP, a structure in which investor returns were capped (OpenAI said returns for the first round of investors were capped at 100 times their investment) with any excess owned by the original nonprofit. The aim, in OpenAI’s telling, was to raise the billions needed for compute and talent while keeping a mission-governed nonprofit on top. Anthropic took a related but distinct path: it describes itself as “a Delaware Public Benefit Corporation, or PBC,” whose directors may “balance the financial interests of the stockholders with the public benefit purpose,” paired with a Long-Term Benefit Trust - “an independent body comprising five Trustees” that holds authority, phased in over time, to “elect a majority of the board within 4 years.” Both structures try to bolt a mission-protecting governance layer onto an entity that still needs outside capital.
The second pattern is the compute partnership. In July 2019 Microsoft invested 1 billion dollars in OpenAI, and OpenAI agreed to “port its services to run on Microsoft Azure” while Microsoft became its “preferred partner for commercializing new AI technologies.” The deal exchanged capital and cloud capacity for a commercial position - a template repeated, in varying forms, across the industry.
The third pattern is mega-scale infrastructure. By January 2025 the numbers had grown by orders of magnitude: the Stargate Project was announced as “a new company which intends to invest 500 billion dollars over the next four years” in US AI infrastructure, with SoftBank “having financial responsibility and OpenAI having operational responsibility.” Frontier economics had become an infrastructure story - data centers, power, and chips - as much as a research one.
The fourth pattern is the acqui-hire, in which a well-funded lab’s people and direction move inside a big-tech company without a conventional acquisition. The clearest example is Microsoft’s March 2024 absorption of Inflection AI’s founders and many staff (see the Inflection acqui-hire entry). Together these four mechanisms - capped profit and trust governance, compute-for-equity partnerships, infrastructure mega-ventures, and acqui-hires - describe how frontier AI is paid for.
Note on sourcing: the OpenAI LP details could not be fetched directly (OpenAI pages return HTTP 403 to automated fetchers) and were corroborated via search against the canonical openai.com URL. Microsoft’s announcement, Anthropic’s Long-Term Benefit Trust page, and SoftBank’s Stargate release were fetched and verified live.