# OpenAI Contracts Government-contract and partnership ledger for [[OpenAI]] — the concrete evidence of the defense pivot. Unlike [[Palantir]], whose business is overwhelmingly government, OpenAI's revenue is primarily consumer (ChatGPT) and enterprise. But its government contracts are the leading indicator of where it is positioning itself in the defense-tech ecosystem and how deeply it is embedding in federal infrastructure. > OpenAI is private and does not disclose contract revenue. Values below are reported ceilings/awards from public sources. When the S-1 becomes public, cross-check government revenue concentration. ## U.S. federal contracts and partnerships | Program / contract | Agency | Value | Date | Notes | |---|---|---|---|---| | Pentagon classified AI contract | DoD | $200M ceiling | Feb 2026 | Classified environments; healthcare, acquisitions, cyber defense. Announced hours after [[Anthropic]] banned. Three red lines: no mass domestic surveillance, no autonomous weapons direction, no social credit. Accepted "all lawful use" standard | | GenAI.mil classified network clearance | DoD (CDAO) | N/A (platform access) | May 2026 | One of seven companies cleared for IL-6/IL-7; [[Anthropic]] excluded. [[Gavin Kliger]] ran clearance | | GSA ChatGPT Enterprise | GSA → all federal agencies | $1/year | Aug 2025 | Loss-leader to embed OpenAI in government workflows across every agency | | Stargate Project infrastructure | DoD-adjacent / multi-agency | portion of $500B JV | Jan 2025 | JV with [[SoftBank]], Oracle; Trump-announced; data center buildout for AI compute | ## Stated red lines OpenAI published three "red lines" for its Pentagon work: 1. No use of OpenAI technology for mass domestic surveillance. 2. No use for directing autonomous weapons systems. 3. No use for high-stakes automated decisions such as "social credit" systems. These red lines are narrower than [[Anthropic]]'s former restrictions (which included broader prohibitions on military applications). The "all lawful use" standard OpenAI accepted means any military use that doesn't cross these three lines is permitted — a permissive framework that aligns with the Trump administration's expectations and contrasts sharply with the safety-first posture OpenAI claimed at founding. ## The government staffing pipeline OpenAI's government-contract strategy is supported by a deliberate hiring pipeline from defense-tech and government: - **Joseph Larson** — Head of Government (July 2025). Former [[Anduril]] executive and former Deputy Chief Digital and AI Officer in the Office of the Secretary of Defense. "Wash100 award" recipient for advancing AI adoption in national security. - **Reginald Babin** — lobbyist. Former legal counsel to Senate Majority Leader Chuck Schumer. This mirrors the [[Revolving Door]] pattern the vault tracks across the defense-tech ecosystem — except OpenAI's pipeline runs from the very agencies it now contracts with. ## Lobbying spend | Year | Lobbying spend | YoY change | |---|---|---| | 2023 | $380K | — | | 2024 | $1.8M | +374% | | 2025 | $2.99M | +66% | The 7x increase over two years tracks the company's shift from "we just build tools" to active policy advocacy. OpenAI's policy positions — federal preemption of state AI regulation, "freedom to learn" on copyright, light-touch voluntary partnership over binding regulation — all serve its commercial interests. The "New Deal for AI" industrial-policy framework, published April 2026, is the most comprehensive articulation of this posture. ## Procurement pattern OpenAI's government entry strategy differs from [[Palantir]]'s (sole-source emergency procurement → integration → dependency). OpenAI's pattern is: 1. **Loss-leader access** — $1/year GSA partnership gives every agency free ChatGPT Enterprise, building familiarity and dependency. 2. **Classified contract** — once embedded in workflows, transition to classified environments with higher-value contracts ($200M Pentagon deal). 3. **Platform positioning** — GenAI.mil clearance positions OpenAI as a default AI provider on classified networks, alongside [[Google]], [[Microsoft]], and Nvidia. 4. **Infrastructure lock-in** — [[Stargate Project]] creates physical infrastructure dependency. The trajectory mirrors the classic government IT vendor strategy: give it away, make it indispensable, then charge. ## Open questions to track - Total government revenue as percentage of overall revenue (will be disclosed in S-1). - New contracts or extensions beyond the $200M Pentagon ceiling. - Which agencies adopt ChatGPT Enterprise through the GSA channel, and at what volume. - Any congressional scrutiny of the [[Anthropic]]-exclusion / OpenAI-inclusion pattern. - Whether the three "red lines" are tested or expanded. ## See Also - [[OpenAI]] — company page - [[OpenAI Dossier]] — dossier front door - [[OpenAI Watch Log]] - [[OpenAI Leadership]] — government staffing pipeline - [[Anthropic Government Ban]] — the exclusion that enabled OpenAI's contract - [[Defense Tech Ecosystem]] — the broader context - [[Dual-Use AI]] - [[Stargate Project]] - [[Deals Hub]]