The artificial intelligence industry has never spent more to be heard in Washington. Federal lobbying disclosures filed for the second quarter of 2026, made public in the reporting window of July 20-22, show the frontier AI labs collectively pouring record sums into influencing lawmakers and the executive branch — a surge that arrived just as the labs edge toward public offerings and the federal government inches toward its first real rulebook for the technology.
Anthropic led the pack among the pure-play AI developers, disclosing $1.97 million in federal lobbying from April through June, up 26% from the first quarter. OpenAI reported $1.2 million, a jump of nearly 18%. Together the two leading model builders spent $3.17 million in the quarter, up 23% from the first three months of the year, according to the filings submitted under the Lobbying Disclosure Act.
The trajectory is even more striking measured against last year. Anthropic's quarterly spend more than doubled versus the second quarter of 2025, when the company was a comparatively minor presence in the capital. Its first-half 2026 total now exceeds $3.5 million — more than the roughly $3.1 million it reported for all of 2025. OpenAI's spending nearly doubled year over year as well.
Meta remained the single largest tech spender in the sector at $5.99 million for the quarter. But that figure fell 15% from the first quarter, part of a broader pattern in which legacy technology and defense contractors trimmed their Washington budgets even as the AI-native firms accelerated. The crossover is now unmistakable: Anthropic's spend topped Nvidia's and closed in on Oracle's roughly $2 million — a remarkable position for a company that barely registered on lobbying ledgers two years ago.
"When a company's lobbying line item doubles year over year, that is not a vanity expense — it is a bet that decisions with real consequences are about to be made," said one Washington policy analyst who tracks technology influence spending, describing the AI cohort's ramp as the clearest signal yet that the industry expects binding rules rather than voluntary guidance. The observation captures the mood across the disclosures: the labs are staffing up in anticipation of a fight, not a formality.
The filings offer a granular look at what, exactly, the money is chasing. Anthropic reported lobbying on export controls, cybersecurity and AI safety standards, with its representatives meeting lawmakers in both chambers of Congress as well as officials at the White House, the Commerce Department and the Treasury Department. Earlier in the year the company's disclosures had leaned toward AI and defense procurement, supply-chain risk, national security and energy infrastructure — the connective tissue of the buildout powering frontier models. OpenAI and its peers reported activity on overlapping fronts: copyright, cloud computing, cybersecurity and defense contracts.
The spending spike is inseparable from each company's distinct set of Washington entanglements. Anthropic is pursuing a confidential IPO, publicly pushing states toward stronger frontier-AI rules, and is a participant in the White House's frontier-framework talks — a combination that gives it reason to be in the room on standards that could define the competitive landscape. OpenAI, for its part, has floated giving the federal government an equity stake and is simultaneously fighting a lawsuit brought by Apple. Meta, despite its much larger check, is reportedly outside the White House frontier framework — a reminder that spend and access do not always move together.
Why it matters
Lobbying disclosures are a lagging record of money already spent, but they function as a leading indicator of where regulation is heading. Companies do not triple their Washington budgets to influence rules that will never be written. The 23% quarter-over-quarter jump in AI-lab spending, and the doubling of Anthropic's outlay year over year, are the clearest market signal yet that the people closest to the policymaking believe binding federal rules — on safety standards, export controls, procurement and liability — are moving from hypothetical to imminent.
The composition of the spending matters as much as the total. That AI-native firms are climbing the rankings while legacy tech and defense contractors pull back suggests a structural handoff in who sets Washington's technology agenda. Anthropic outspending Nvidia and nearly matching Oracle is a symbolic inflection point: the model developers, not just the chipmakers and cloud incumbents, now see themselves as the parties with the most to gain or lose from the next round of rulemaking. For a company preparing a confidential IPO, that influence spending doubles as regulatory risk management for future investors.
There is also a competitive subtext. Anthropic has positioned itself as the safety-forward lab willing to invite tougher state and federal rules, a stance that — if it becomes law — could disadvantage rivals less prepared to meet stringent standards. Lobbying is one of the instruments through which that strategy gets executed.
What to watch
The immediate flashpoint is the White House frontier framework. Anthropic is inside those talks and Meta reportedly is not, and the disclosures suggest the labs are spending to shape the framework's teeth — voluntary commitments versus enforceable standards. Watch whether the eventual document imposes real obligations on model developers or codifies the lighter-touch, voluntary approach Washington has so far favored.
The second front is the states. Anthropic's public push for stronger frontier rules at the state level sets up a potential clash over federal preemption; if Washington moves to override state AI laws, expect the labs' lobbying to migrate toward that fight. Also worth tracking: whether third-quarter filings show the spend accelerating further into the midterm cycle, and whether OpenAI's government-equity proposal and Apple litigation start showing up more prominently in its disclosed lobbying activity. When the money keeps climbing, the rules are usually not far behind.
“When a company's lobbying line item doubles year over year, that is not a vanity expense — it is a bet that decisions with real consequences are about to be made.”— Policy analyst, technology influence-spending researcher