Meta’s $7.1 Million Quarter Is Reshaping Defense AI Rules

Follow the Money: AI Lobbying's Explosion

Meta spent $7.1 million on federal lobbying in the first quarter of 2026. That single quarterly figure captures the new reality: AI companies are spending record sums on federal lobbying, creating a new influence ecosystem that is already reshaping defense and space policy and forcing frontier tech startups to find new ways to compete for Washington's attention.

The growth has no modern precedent. Total corporate spending on lobbying that mentioned AI was $1.8 million in 2013, per The AI Lobby's data. By 2022, the year ChatGPT launched, that figure hit $12 million. In 2025 it exceeded $50 million, a 400 percent increase in three years, faster than any other tech category. Quarterly spending climbed from $18.2 million in Q1 2024 to $35.2 million in Q1 2026, putting the industry on pace for $70 million-plus this year.

The player count exploded too. OpenSecrets records 648 companies reporting AI lobbying in 2024 versus 458 in 2023. By 2025, lobbyists represented 774 organizations and filed over 3,500 reports mentioning AI. Congressional hearings on AI quadrupled between 2022 and 2023. AI-related bills introduced in Congress jumped from 88 in the 117th Congress to over 200 in the 118th.

The federal purse is expanding just as fast. Brookings found the value of funds obligated on federal AI contracts grew from $261 million in 2022 to $675 million in 2024 (150 percent) to $7.2 billion in 2026 (966 percent from 2024). Potential awards jumped from $355 million to $4.6 billion to $91.8 billion over the same periods. The Department of Defense dominates: its contract count rose from 254 in 2022 to 657 in 2024 to 1,319 in 2026. DoD's share of potential contract value went from $269 million (76 percent of federal AI spending) in 2022 to $4.3 billion (95 percent) in 2024 to $90.7 billion (98.9 percent) in 2026. Twenty-eight federal agencies now hold AI contracts, up from 17 in 2022.

For comparison, the pharmaceutical industry, long a top lobbying spender, has seen annual expenditures hold relatively stable around $350–380 million. The AI industry is still smaller in absolute terms, but its growth rate signals a sector that perceives existential regulatory risk and is investing heavily to manage it. The lobbying infrastructure built in three years rivals what pharmaceuticals, energy, and finance constructed over decades. The rules governing AI procurement, export controls, copyright, and safety standards are being written now — shaped by the companies spending the most to be in the room.

Defense and Space: Procurement Rewritten

Defense AI contractors spent $43 million on federal lobbying in 2025. The Department of Defense's AI budget request hit $1.8 billion. Yet there is still no binding policy on autonomous weapons targeting. That gap — between spending power and regulatory clarity — is where the lobbying money goes.

Palantir Technologies leads the pack with over $8 million in total federal lobbying spend, heavily focused on defense and intelligence AI applications. The company employs lobbyists from six firms, including former senior Pentagon officials and ex-congressional defense committee staffers. Its disclosures reference AI-related provisions in the National Defense Authorization Act, autonomous systems policy, and intelligence community AI procurement. Combined, defense-focused AI companies spent over $31 million on federal lobbying in 2025.

Incumbents are bundling AI into existing platforms. Lockheed Martin's recent lobbying for the Department of Defense Appropriations Act grouped artificial intelligence into the same line item as hypersonic missiles, munitions, and the F-35 aircraft program. Huntington Ingalls reported lobbying on "FORTIS AI" alongside core maritime projects like the Columbia Class submarine, the DDG-51 destroyer, and the Carrier Replacement Program. Lockheed Martin and Microsoft reported total annual lobbying expenditures of $15.7 million and $10.1 million respectively, using their scale to bundle AI into multibillion-dollar defense and infrastructure contracts.

Newer entrants are targeting specific appropriations. Applied Intuition saw its lobbying spend jump from $550,000 in 2024 to $910,000 in 2025 as it targeted Defense appropriations bills and support for deploying autonomous systems for the Navy and Air Force. Scale AI and Shield AI reported programs of $1.2 million and $1.4 million respectively. Booz Allen Hamilton reported $650,000 in Q3 2025 lobbying spending through its in-house team of five lobbyists, focused on the NDAA, Intelligence Authorization Act, Veterans Affairs IT modernization, and the VET Artificial Intelligence Act, seeking to influence how billions in defense and intelligence resources are allocated.

The policy debate centers on autonomy and accountability. DoD Directive 3000.09 requires a human in the loop for lethal autonomous weapons, but critics argue the directive is outdated and its definitions too vague to constrain current AI capabilities. Congress has held hearings on updating the directive, with testimony from military officials, AI researchers, and, notably, the companies building the systems. Campaign contributions from defense AI companies flow disproportionately to members of the Armed Services and Intelligence committees.

Procurement rules are being rewritten in real time. The Protecting AI and Cloud Competition in Defense Act of 2025 mandates competitive procurement for DoD cloud, data, or AI contracts exceeding $50 million annually, creating favorable conditions for emerging vendors. The Trump administration's America's AI Action Plan directs the creation of an AI procurement toolbox managed by GSA in coordination with OMB, updates federal procurement guidelines to ensure the government only contracts with frontier LLM developers whose systems are "objective and free from top-down ideological bias," and prioritizes DOD-led agreements with cloud providers to codify priority access to computing resources in a national emergency.

A $200 million dispute between Anthropic and the Department of Defense over an AI contract for national security, warfighting, and enterprise uses (Axios first reported) suggests separating commercial labs from defense work may be difficult under the current administration. In August 2025, OpenAI and Anthropic announced agreements allowing government agencies to use their models for $1. Elon Musk's xAI offered a similar deal for 42 cents.

Space policy is moving too. Executive orders require the FAA to deploy AI tools for UAS waiver applications by October 2025, directly relevant to autonomous systems operating in contested airspace. Federal agencies struggle with AI deployment barriers, including data access, IT infrastructure, and workforce skills, creating immediate demand for AI service providers who can manage the compliance environment their own lobbying helped shape.

Defense AI spending is projected to exceed $18 billion in fiscal year 2027. The companies competing for those contracts are spending heavily to shape the rules. As governmental responsiveness increases to industry lobbying pressures, procurement strategies will increasingly reflect these transformations, ultimately shaping contract requirements and compliance standards crucial for success in government contracts.

Inside the Giants' Playbook

Meta's lobbying machine operates on a different scale. The company has sustained its pace: $6.5 million in Q4 2025, $5.8 million in Q3, and $5.99 million in Q2 2026, according to CNBC. Its 2025 total hit $26.29 million, the highest of any company in any industry. Issue One's analysis shows Meta now employs roughly one registered lobbyist for every six members of Congress.

Anthropic's trajectory is steeper. The company spent $360,000 on federal lobbying in Q1 2025. One year later that figure hit $1.6 million, a 344 percent increase. By the second quarter of 2026 Anthropic had already spent $3.5 million in the first half of the year, exceeding its full-year 2025 total. The spending reflects a strategic pivot. Anthropic hired Ballard Partners to pursue Defense Department and Pentagon AI procurement contracts, moving beyond its safety-research origins. Its disclosures now reference the Healthcare AI Accountability Act (S. 4178) and Energy Department Genesis Mission funding. In mid-2026, Commerce Department export controls forced Anthropic's models offline, triggering an emergency lobbying scramble. The company also donated $20 million to Public First Action, a bipartisan network advocating state-level AI authority and whistleblower protections, positions that invert the preemption agenda Meta and Google are funding.

OpenAI's lobbying follows its compute demands. The Stargate joint venture with SoftBank and Oracle requires data-center capacity measured in gigawatts. OpenAI's filings target energy permitting reform and the Data Center Energy Efficiency Act (H.R. 4425). CNBC's data shows its Q2 2026 spending reached $1.2 million, CNBC found up 18 percent from the prior quarter, with disclosed work on copyright, cloud infrastructure, cybersecurity, and privacy. The company also faced government delays rolling out its latest model, a friction point that likely accelerated its Washington investment. OpenAI president Greg Brockman personally contributed $12.5 million to Leading the Future, a super PAC that has raised roughly $140 million for the 2026 cycle and already targeted a New York Democratic primary candidate who co-sponsored a state AI safety bill.

The three companies share a playbook: hire the regulators. In the past 18 months OpenAI recruited a former White House technology policy director. Anthropic brought on a former NIST official who helped draft AI safety standards. Google added three former Senate Commerce Committee staffers. Microsoft poached a senior FTC technologist. The revolving door runs both ways — former tech AI ethics researchers have moved into NIST, the AI Safety Institute, and Congressional Research Service roles. OpenLobby tracks over 5,000 former government officials now registered as lobbyists, with AI-sector demand accelerating the premium for government experience.

Coalition building amplifies each company's voice. The AI Alliance (co-founded by Meta and IBM), the Information Technology Industry Council, TechNet, and the Chamber of Commerce's Technology Engagement Center all lobby on AI with funding traceable to the same handful of firms. When five organizations testify with identical talking points, the appearance of consensus is manufactured. Google funds ITI, TechNet, and the AI Alliance. Microsoft funds ITI, BSA, and the Partnership on AI. The "broad coalition" supporting industry-friendly regulation is often five companies speaking through different megaphones.

This infrastructure creates a structural disadvantage for any company without a nine-figure lobbying budget. Defense contractors (CNBC reported the eight largest spent $19.68 million in Q2 2026) are outgunned on AI-specific policy. Startups in space, robotics, and applied AI face a Washington where the rules for procurement, export controls, and liability are being written by the same firms that will bid on the contracts. The midterm elections add another layer: AI industry PACs have pledged over $200 million for the 2026 cycle. The message to lawmakers is explicit — support the preferred framework or face funded opposition.

Can Startups Crash the Party?

The giants' spending spree has created a structural problem: when Meta drops $7.1 million in a quarter and the top seven tech firms push $50 million in nine months, the price of admission for a House Energy and Commerce hearing or a DoD procurement briefing effectively locks out everyone else. That vacuum is where the counterattack is forming — not from traditional boutique firms, but from startups betting that the lobbying trade itself is ripe for the vertical AI playbook.

Locke, a Y Combinator S26 company that launched six weeks before its August 2026 demo day, is the clearest signal of the new model. Founder Parth Badhwar arrived at the problem from both sides of the table: policy roles in the Biden-Harris White House Office of National Drug Control Policy and on Deloitte's government relations team, then go-to-market leadership at a startup selling into dozens of government offices. He watched legacy firms charge incumbent prices for a fragmented product: one shop for the Hill, another for agencies, a third for proposals, with no single entity accountable for the outcome. "You cannot simply deploy software around them," Badhwar wrote in an August essay on government process. His strategy: use AI to compete inside the existing system before attempting to replace parts of it.

Locke's pitch is an "AI-native government affairs firm" — agents paired with in-house lobbyists who own the full path from regulation monitoring through first meeting to funded contract. The agents track legislation, regulations, hearings, RFPs, and market-moving government signals in real time, analyze legislation, map stakeholder engagement, and draft complex contract proposals. The policy staff ("Members of Policy Staff" in Locke's terminology) own client outcomes that depend on agency priorities, legislative calendars, and human relationships. The company targets defense, robotics, healthcare, energy, space, advanced manufacturing, and critical infrastructure: sectors where the $10 trillion in annual government spending gets reallocated each cycle through budgets, appropriations, and contracts.

Early traction suggests the model has legs. Since its quiet launch, Locke has blown past a $500,000 annual revenue run-rate serving clients across defense, healthcare, and GovTech. That breadth makes it closer to a software-enabled professional services firm than a conventional SaaS subscription. Agents compress enough research, monitoring, and document work to let a small policy team serve more clients than a traditional government affairs shop, faster and cheaper than the legacy firms guarding the front door.

Locke isn't alone. Fed10's platform monitors every bill across all 50 states and Congress, drafts amendments, and tracks legislative intelligence in real time. The pattern is consistent: former insiders who know the manual grind are building the tools to eliminate it.

Federal lobbying hit a record $5.08 billion in 2025, and one in four lobbyists now works on AI. The startups are selling AI back to the influence trade. If the thesis holds, government affairs becomes as systematic and data-driven as modern software sales. If it fails, it will be because the problem was never just information asymmetry — it was trust and relationships, things AI has not yet learned to build.

Reading the Disclosures: What This Means for Frontier Tech

The lobbying surge is already rewriting the rules of engagement for anyone building or selling technology to the federal government. Procurement officers at the Pentagon and intelligence agencies are drafting requirements shaped by the same companies bidding on the contracts. When Palantir spends $8 million lobbying on NDAA provisions for autonomous systems, or when Anduril and Shield AI pour millions into shaping AI-specific program language, the resulting solicitations reflect their architectures.

The talent market tells the same story. OpenAI added 42 roles in the past week alone, with salary bands running $185,000 to $500,000 and a median of $366,000. Anthropic's 42 new postings show a band of $210,000 to $554,000, median $400,000.

New compliance burdens compound the pressure. The FY2025 NDAA included a provision — Section 851 — that bars DoD contractors from retaining consultants, law firms, or lobbying shops that also represent companies on the Pentagon's 1260H list of Chinese military-linked entities. The Defense Department has yet to publish the DFARS clause defining "reasonable inquiry," but the certification requirement takes effect July 1. Contractors are already auditing their vendor rosters. Several Washington lobbying firms have cut ties with Chinese clients rather than risk their defense revenue.

Data-center infrastructure adds another layer. The Trump administration's July 2025 executive order designates any facility requiring more than 100 megawatts for AI inference, training, or simulation as a "Qualifying Project" eligible for expedited permitting, federal loans, and categorical NEPA exclusions. Amazon, already spending $4.35 million on lobbying in Q3 2025 and on track for $125 billion in annual capex, is positioning its AWS infrastructure as critical national security assets. The JWCC program ($9 billion in potential task orders shared among Amazon, Google, Oracle, and Microsoft) will favor cloud providers who can demonstrate both security clearance readiness and permitting speed. Companies building space or defense AI workloads on smaller clouds face an uphill certification path.

The military right-to-repair fight illustrates how lobbying alters operational reality. John Deere spent more than $700,000 and Philips over $1 million lobbying the NDAA to block provisions requiring contractors to provide repair data. The final bill stripped both the Warren-Sheehy Warrior Right to Repair Act and a contractor-backed "data-as-a-service" alternative. For maintainers at depots and forward operating bases, that means continued dependence on OEM technicians for diagnostic access — a logistics tail that lobbyists protected.

For engineers and operators, the takeaway is practical: read the lobbying disclosures alongside the RFPs. The companies shaping the requirements are filing quarterly reports naming the exact bills, agencies, and provisions they target. Those filings reveal which technical standards will harden into contract clauses, which autonomy definitions will gate funding, and which data-rights clauses will determine whether your team can repair its own hardware in theater. When Meta spends $7.1 million in a quarter to shape the rules, and a six-week-old startup builds AI agents to read those same rules faster than the incumbents can write them, the arms race has already moved past lobbying into something that looks like algorithmic warfare. The disclosures are public. The RFPs are drafting. The only question is who gets to the next clause first.

Working in frontier tech? Zero G Talent tracks the openings: see every open OpenAI role, browse frontier tech jobs, openings at Anthropic and Arena, and the people building the field.