AI Lobbying Hits Record: Anthropic, OpenAI Q2 Spending

The two most valuable companies in American artificial intelligence spent more money influencing Washington in the second quarter of 2026 than they ever have. According to federal lobbying disclosures reported on July 21, 2026, Anthropic spent $1.97 million and OpenAI spent $1.2 million in the quarter—a combined $3.17 million, up roughly 23% from the first quarter and a record for both companies.

The figures land at a pointed moment. Both companies are preparing to go public at valuations approaching $1 trillion, both are fighting over the shape of AI regulation at the state and federal level, and both spent the spring absorbing direct policy shocks—most sharply an export-control order that briefly took Anthropic’s most advanced models offline. The spending is the clearest signal yet that the AI industry now treats Washington as a first-order business risk, not a background concern.

The numbers

The disclosures cover the April–June quarter and reflect sharp year-over-year increases:

• Anthropic: $1.97 million, more than double what it spent in the same quarter a year earlier. That total put Anthropic ahead of Nvidia, which reported roughly $1.25 million, and nearly level with Oracle’s ~$2 million.
• OpenAI: $1.2 million, up nearly 18% from the prior quarter and close to double its year-earlier figure. OpenAI also opened its first Washington office as its footprint in the capital grew.

The reference points matter. A pair of AI labs—neither a decade old, neither yet public—now out-lobbies or matches long-established chip and enterprise-software giants. Anthropic outspending Nvidia, the most strategically important hardware company in the industry, is the detail that captures the shift: influence spending has decoupled from company age and revenue and reattached to regulatory exposure.

Reporting on the quarter framed the AI surge against a backdrop of legacy tech and defense lobbying slipping, underscoring that the money is not just rising in absolute terms but rotating toward AI as the sector where policy is most in flux.

What they were lobbying on

Anthropic’s disclosures point to export controls, cybersecurity, and AI safety standards among its top issues, with lobbyists meeting lawmakers in both chambers of Congress as well as officials at the White House, the Commerce Department, and the Treasury Department. That agenda maps directly onto the quarter’s biggest shock to the company.

In June, Commerce imposed export controls on Anthropic that led to two of its most advanced models going dark, forcing a scramble to negotiate terms before access was partially restored. Living through a regulatory action that can switch off your product is the kind of event that reprices lobbying from optional to existential—and the Q2 total reflects a company that had just learned the lesson firsthand. The episode sits inside a wider tightening of AI and chip trade policy, the same current running through debates over chip export licenses.

The two companies also diverge sharply on state-level AI regulation, and that split shapes where their money goes. Anthropic has urged states to pass tougher AI safety laws rather than converge on a single national rulebook; OpenAI has pushed the opposite—a nationwide common framework that would preempt a patchwork of state rules. The disagreement is not academic. A federal proposal to freeze state AI laws for a decade drew public opposition from Anthropic’s leadership, and the outcome determines whether the industry answers to one regulator or fifty.

Why now: IPOs and midterms

Two forces are pulling spending up at once.

The first is the IPO calendar. Both companies are moving toward public offerings at historic valuations, and the Anthropic IPO filing has already put the economics of these businesses under a public microscope. Anthropic’s shares have changed hands on the secondary market at levels implying a valuation around $1.2 trillion. Valuations that large rest on assumptions about future profit margins, and those margins are hostage to policy—export rules that determine which markets a model can serve, safety mandates that raise the cost of shipping, and liability regimes that set the downside. A company about to sell equity to public investors has every incentive to shape the rules that bound its addressable market before the prospectus is priced.

The second is the political calendar. With midterm elections approaching, AI policy is becoming a campaign issue, and the industry is spending to be at the table as candidates and committees stake out positions. The combination—an unusually consequential regulatory year running straight into an election—turns lobbying from a steady line item into a surge.

What it means

The record spending is best read as a leading indicator: it measures how much regulatory outcomes are worth to these companies, and the number is climbing fast. When a firm doubles its lobbying year over year, it is pricing the probability that a policy decision materially changes its business. For AI at the frontier, that probability is now high enough that $2 million a quarter is a rational hedge.

Who this matters for. For investors weighing the coming IPOs, lobbying spend is a signal worth reading alongside the financials. It quantifies political risk from the inside—these companies are telling you, in dollars, how exposed they believe they are to Washington. A business whose path to public markets depends on export licenses and safety mandates carries a risk profile that a pure revenue model misses, and the DeepSeek and Anthropic filings will be read for exactly this kind of exposure.

The competitive subtext. Anthropic and OpenAI are not just lobbying against regulation—they are lobbying against each other’s preferred version of it. Anthropic’s push for strict state laws and OpenAI’s push for a single federal framework are competing bids to shape a rulebook each believes it can live with better than its rival can. Whoever wins that argument gains a structural advantage, which is why both are willing to spend records to influence it. Standards fights like the EU’s GPAI enforcement regime show how much a favorable rule can be worth once it hardens into law.

The broader tell. AI labs now out-lobby the chipmakers and cloud incumbents that supply them. That is a marker of where power and money are concentrating in the technology stack—and a preview of a policy fight that will intensify as these companies go public and their decisions ripple further into the economy. The government’s own posture is still forming, from federal AI task forces to the export regime, and both sides are spending to shape it before it sets.

What to watch next: whether Q3 spending climbs again as IPO roadshows begin, whether Congress advances or kills a preemptive freeze on state AI laws, and whether the export-control framework that hit Anthropic hardens into a durable regime or gets negotiated down.