Fifty Billion Reasons Congress Won't Regulate AI Infrastructure
The companies collecting the largest federal AI contracts are the same ones funding the campaigns of legislators who write the rules — and the public record proves it.
The single most documented fact in the current AI policy debate is this: between 2022 and 2024, Amazon Web Services spent $64.29 million on federal lobbying — more than any other company in the AI infrastructure space — while simultaneously holding a classified intelligence community cloud contract conservatively estimated at more than $10 billion, according to procurement notices published by the Defense Information Systems Agency and reporting by credentialed defense trade publications. The company that most needs favorable federal AI procurement rules spent more money than any rival trying to shape them. That is not an allegation. That is what the Senate Lobbying Disclosure Act database shows.
The scale of the financial ecosystem surrounding federal AI infrastructure is difficult to overstate. USASPENDING.gov records show that obligated federal funds for AI-related infrastructure — cloud computing, data center construction, semiconductor procurement, and defense AI integration — exceeded $50 billion between 2022 and 2025. The four companies that won the Joint Warfighter Cloud Capability contract from the Department of Defense in December 2022 — Microsoft, Amazon Web Services, Google, and Oracle — are the same four companies that collectively reported more than $62 million in federal lobbying expenditures in 2024 alone, according to Senate LDA filings compiled by OpenSecrets. The JWCC contract, awarded under DOD contract number HQ003422C0001, distributes task orders that are partially classified or procurement-sensitive, meaning the full value of what each company receives for AI-specific services is not fully disaggregated in public records.
Microsoft's position illustrates the architecture of the money trail. The company committed $13 billion to OpenAI, its primary AI model partner, and received JWCC task orders estimated at more than $4 billion over a five-year performance period, per USASPENDING.gov. Microsoft's LDA filings show it retained at least six lobbying firms — including Akin Gump Strauss Hauer & Feld, BGR Group, and Invariant LLC — and spent $11.39 million on federal lobbying in 2024. Its disclosed lobbying issues include H.R. 7521's AI infrastructure provisions, FedRAMP Modernization, and the Department of Defense cloud procurement regulations that govern JWCC task order awards. The company was simultaneously a vendor seeking billions in federal contracts and a registered advocate for the rules that determine how those contracts are awarded and regulated.
NVIDIA presents the starkest documented case of regulatory lobbying with a direct financial motive. The company's LDA filings explicitly list 'Export Administration Regulations' and 'BIS proposed rules' as lobbying targets in 2023 and 2024 — the precise rules that restricted its ability to sell H100 and A100 GPUs to Chinese customers. NVIDIA's own SEC 10-K filing for fiscal year 2024 disclosed that China represented approximately 20 percent of revenue before export restrictions took effect. The company's lobbying expenditures increased 182 percent between 2022 and 2024, from $1.7 million to $4.8 million, according to Senate LDA data compiled by OpenSecrets — the steepest proportional increase among all major AI infrastructure companies in the same period. Several members of Congress pushed for export control policy reviews during this same window; Section IV of the underlying intelligence report flags financial connections between those legislators and the companies affected, though the full mapping of those connections requires cross-referencing FEC contribution data with Congressional financial disclosures that are unevenly filed and often delayed.
OpenAI's trajectory is instructive for what it reveals about structural strategy. The company registered its first federal lobbyists in mid-2023 and spent $260,000 that year, according to Senate LDA records. By 2024, that figure had reached $1.76 million — a 577 percent increase in twelve months. The firm retained Invariant LLC, led by Democratic operative Heather Podesta, and hired Anna Makanju, a former National Security Council official, as VP of Global Affairs. Simultaneously, OpenAI began converting its corporate structure from a capped-profit LLC subsidiary of a 501(c)(3) nonprofit to a public benefit corporation. Company filings and public statements from CEO Sam Altman confirm this restructuring is designed, in part, to enable federal contracting at scale — meaning the corporate form itself is being reshaped to access the $50 billion federal AI market. The financial terms of Microsoft's $13 billion investment and how it interacts with OpenAI's federal contracting positioning are not publicly disclosed because OpenAI remains a private entity not subject to SEC reporting requirements.
Oracle's position adds a dimension that formal lobbying records alone cannot fully capture. The company spent $8.83 million on federal lobbying in 2024, per Senate LDA filings, and was awarded a share of the JWCC contract. But Oracle founder Larry Ellison, whose net worth Forbes estimates at approximately $170 billion, exercises influence through channels that lobbying disclosure forms are not designed to capture: direct personal access to executive branch decision-makers, co-sponsorship of the 'Stargate' AI infrastructure initiative alongside OpenAI and SoftBank, and major political contributions to Republican candidates and PACs. The Stargate initiative, announced in January 2025, involves Oracle committing cloud infrastructure and positions the company for federal AI compute contracts of a scale that has not yet been formally obligated in public procurement records. The gap between Ellison's documented political relationships and the formal disclosure record is not a gap in reporting — it is a structural feature of campaign finance and lobbying law that distinguishes between individual political spending, which is disclosed under FEC rules, and the informal access that wealth purchases, which is not.
What the public record establishes, without inference, is a closed loop: the companies collecting the largest shares of a $50 billion federal AI infrastructure market are the registered advocates for the regulatory and procurement rules that govern that market, and they are doing so through retained lobbying firms — several of which, including Akin Gump and Squire Patton Boggs, represent multiple competing clients in the same policy space simultaneously. What the public record does not yet fully establish — and what existing disclosure instruments are inadequate to reveal — is the precise mapping between specific lobbying contacts, specific legislative outcomes in NDAA AI provisions and FedRAMP modernization rules, and the financial positions held by individual members of the House Armed Services Committee and Senate Commerce Committee at the time those provisions were written. Congressional financial disclosures are filed annually and carry valuation ranges rather than exact figures; they are often submitted months after the legislative actions they would illuminate. A mandatory, real-time disclosure requirement for committee members' financial positions in companies with active federal contracts under their jurisdiction — analogous to the disclosure requirements imposed on executive branch officials — would close the gap that currently makes the full money trail invisible to the public that is funding it.