Here's what Daniel sent us this week. He writes: "Let's talk about the world of lobbying. We hear a lot about it, but it's also cloaked in secrecy. How big is the industry in the US? What would a typical day in the life of a lobbyist look like? And how regulated is it?" Good questions, and the timing is interesting because the numbers coming out of Washington right now are genuinely staggering.
The Lobbying Disclosure Modernization Act stalled in committee earlier this year, by the way. Bipartisan support, couldn't get out of committee. And the reason that's worth flagging right at the top is that it perfectly captures the central tension here: the people who would write the rules governing lobbyists are the same people lobbyists spend the most time cultivating relationships with.
The system regulating itself. Always goes well. So let's start with the scale, because I think most people have a rough sense that lobbying is a big industry, but the actual number is kind of jaw-dropping.
Federal lobbying hit six billion dollars in reported spending in 2025. That's a thirty-six percent jump from 2024, which itself was around four point four billion. And the drivers behind that surge are exactly what you'd expect: AI regulation debates, trade tariff fights, and healthcare policy. Three of the most contested policy areas of the current moment, all colliding at once.
Six billion. For reference, that's more than the GDP of a few small countries, just... spent on trying to influence one government's legislative agenda.
And that's just the registered, disclosed spending. We'll get to the shadow economy in a bit, but the registered figure alone involves somewhere between twelve thousand and fifteen thousand active lobbyists representing over fifteen thousand five hundred clients.
By the way, today's script is powered by Claude Sonnet four point six, which feels vaguely appropriate for an episode about AI being one of the things lobbyists are most aggressively trying to shape right now.
Ha, full circle. But yes, the tech sector is now the fastest-growing lobbying constituency in Washington. Pharmaceutical and healthcare still hold the top spot in absolute dollars, but tech has been climbing at a rate that's outpacing everything else.
So let's actually define what we're talking about, because I think the word "lobbying" gets used to mean everything from bribery to just... calling your congressman. What is it, professionally speaking?
At its core, lobbying is professional advocacy. You're a paid representative, and your job is to communicate the interests of your client to legislators and regulators in a way that influences policy outcomes. That's it. The legal definition under the Lobbying Disclosure Act is pretty specific: if you spend twenty percent or more of your working time contacting government officials on behalf of a client, and you're being paid to do it, you have to register.
Which means if you spend, say, nineteen percent of your time doing exactly that, you don't have to register. That's not an accident.
Not remotely an accident. But we'll get there. The key thing to understand is that lobbying, as a profession, is a legal, regulated activity. It exists because there's a genuine information problem in Congress. You have five hundred and thirty-five members of Congress trying to legislate across every domain of human activity, from pharmaceutical patent law to semiconductor manufacturing to fisheries management in the Gulf of Mexico. Their staffs are small and overworked.
And that's where lobbyists come in as what some policy researchers call an "information subsidy."
That framing is really useful. A good lobbyist isn't just buying access, they're providing free research. They're handing a legislative assistant a detailed one-pager that explains exactly how a proposed bill would affect the two thousand jobs at a pharmaceutical plant in their district. That's genuinely useful information for an overworked staffer who might be juggling fifteen different issues that week.
The cynical reading of that is: they're providing selectively useful information. The research is accurate, but it's curated to support one conclusion.
The cynical reading is also the accurate reading. But that's not different from what every other participant in the policy process does. Advocacy organizations on the left and right do the same thing. The difference is resources. When pharma can spend three hundred and eighty million dollars on lobbying in a single year, the quality and volume of their "information subsidy" is going to dwarf what a patient advocacy nonprofit can produce.
Three hundred and eighty million. Just pharmaceutical and healthcare.
Just that sector, in 2024. And that number has grown since. The constituent impact model is how they spend a lot of that. You take a piece of legislation, you map out its economic effects at the district level, and you show every relevant member of Congress exactly what it means for their voters. It's sophisticated, data-driven, and it works.
How granular does that modeling actually get? Like, are we talking county-level, or are they going deeper than that?
It can get extremely granular. We're talking specific zip codes, specific employers, sometimes specific job categories within a facility. A well-resourced lobbying operation can walk into a meeting with a representative from a swing district and say: this provision affects four hundred and twelve manufacturing jobs at a specific plant in your district, and here's the projected wage impact over five years. That level of specificity is hard for a member to dismiss, because it's tied directly to their constituents and their electoral interests.
And the member can't easily fact-check it in the room.
Right. They're taking the analysis largely on faith, which is exactly why the "information subsidy" framing cuts both ways. The information is real, but the framing around it is entirely controlled by the party providing it.
Can we talk about what this actually looks like on a day-to-day basis? Because I think the martini lunch image is so baked into how people picture lobbying, and I suspect the reality is considerably more mundane.
Considerably more mundane and considerably more administrative. The day typically starts around seven in the morning with what's essentially an intelligence gathering session. Lobbyists are reading Politico Pro, Punchbowl News, the specialized committee newsletters, looking for any bills or amendments that got filed overnight or any markup schedules that changed. Their clients are paying them to know things before anyone else does.
So it's less "power lunch" and more "frantically reading the political equivalent of financial news at seven AM."
That's a genuinely accurate description. Then from about nine to eleven, you're briefing clients. A lot of what lobbyists do is manage the expectations of the people paying them. A corporate client might think that a particular bill is going to pass in three weeks. The lobbyist's job is to explain why it's actually going to be six months, why two key committee members need to be persuaded first, and what that persuasion campaign looks like.
I'd never really thought about that part of it. They're essentially strategic translators between how Washington actually works and how clients think it works.
That's a big chunk of the value proposition. And it's not glamorous. A significant portion of any senior lobbyist's week is essentially client management — talking someone off a ledge because they read a news story about their bill and panicked, or explaining why a procedural delay isn't the catastrophe it looks like from the outside.
So there's almost a therapist dimension to it.
There really is. Then midday is the "hill run." Moving between the House and Senate office buildings, doing what insiders call drop-bys. And here's something that surprises people: most of those meetings aren't with the actual members of Congress. They're with legislative assistants. The LAs are the ones who do the substantive policy work, who draft the language, who advise the members. Getting your position in front of the right LA is often more valuable than a photo-op with the member themselves.
The staffers are the real infrastructure of the legislative process.
By a significant margin. And they're also the pipeline for the revolving door, which we should talk about. Over five thousand current registered lobbyists are former government officials. That includes former members of Congress, former agency heads, former senior staffers. The institutional knowledge they carry is the product.
So the career path is: work in government, build relationships and expertise, then monetize both on K Street.
The Honest Leadership and Open Government Act of 2007 tried to address this with cooling-off periods. Former members of Congress have to wait two years before they can lobby their former chamber. Senior executive branch officials have a one-year cooling-off. But here's where it gets interesting, and this connects directly to the shadow lobbying problem: the cooling-off period restricts formal lobbying contacts. It doesn't restrict "strategic advising."
Meaning you can still work for a lobbying firm, you just can't technically make the contact yourself.
You sit one desk over from the registered lobbyist, you tell them exactly what to say and who to say it to, and you're perfectly within the law. The relationships you built during your government career are being monetized, just with one degree of separation.
That's a remarkable loophole that is also somehow entirely unsurprising.
What I find genuinely striking about the revolving door data is the directionality. It's not just government officials going to lobby firms. It also runs the other way: former lobbyists become regulators and senior staffers. So you get people who spent years advocating for an industry then overseeing that industry's regulation. That dynamic shapes the institutional culture of agencies in ways that are hard to measure but very real.
There's actually a term for that, right? Regulatory capture?
Regulatory capture, exactly. The classic case that gets cited in economics textbooks is the savings and loan industry in the eighties, where the regulatory agency became so aligned with the industry it was supposed to oversee that it missed — or looked the other way from — the practices that eventually caused a massive financial crisis. The revolving door is one of the main mechanisms through which capture happens. It's not usually explicit. It's more that the people making regulatory decisions share a professional culture, a set of relationships, and sometimes a future career trajectory with the people they're regulating.
Let me push on the economics for a second, because there's a study that gets cited a lot about return on investment for lobbying spending. The numbers are almost absurdly high.
The figure that gets cited most often is around twenty-two thousand percent return on investment for certain targeted tax provisions. The mechanism is straightforward: a company spends, say, one million dollars lobbying for a specific tax treatment, and if they get it, the tax savings over the relevant period are two hundred and twenty million dollars. That math makes lobbying one of the highest-ROI activities a corporation can engage in, which explains why the industry keeps growing.
Which also means that if you're a shareholder of a large corporation, you should arguably be annoyed if they're not lobbying effectively.
It's a rational allocation of capital from a purely financial standpoint. The returns dwarf most other forms of corporate spending. And that's part of why the six billion dollar figure isn't surprising when you look at it from the client side. The question isn't why companies spend that much; the question is why they don't spend more.
Okay, so let's shift to the regulatory side, because Daniel's prompt specifically asks how regulated this industry is. And I think the answer is: more than people think, and also less than people think, depending on which part of the industry you're looking at.
That's the honest framing. The Lobbying Disclosure Act of 1995 and its 2007 amendments created a fairly detailed disclosure regime. Registered lobbyists have to file quarterly reports, the LD-2 filings, that specify how much their firm was paid, which client they were working for, and which specific bills or agencies they were lobbying. That information is publicly searchable.
OpenSecrets aggregates all of it, which is genuinely useful.
It's a remarkable public resource. You can look up any piece of legislation and see which organizations were lobbying on it and roughly how much they spent. The transparency at that level is real. Where it breaks down is at the edges.
The twenty percent threshold being the most obvious edge.
The twenty percent threshold is the structural loophole. But there's also the question of what counts as a "lobbying contact." Under the LDA, certain communications are exempt. Testimony at a public hearing doesn't count. Responses to requests for information from a government official don't count. Providing written comments on proposed regulations doesn't always count. A sophisticated operation can structure its activities to minimize what technically triggers registration while still having substantial influence.
And then there's the gift rules, which I want to talk about because the "toothpick rule" is one of the more delightfully absurd pieces of regulatory specificity I've encountered.
The toothpick rule is real and it is exactly what it sounds like. Lobbyists cannot buy meals for members of Congress. There are strict gift bans. But there's an exception for food at receptions, as long as it's food that can be eaten without a fork or a plate. So a lobbyist can host a lavish reception for a congressman with open bars and expensive hors d'oeuvres, and that's fine, as long as the congressman isn't sitting down to a plated dinner.
The regulatory distinction between standing food and sitting food is doing a lot of work in our democracy.
It's one of those rules that emerged from a genuine attempt to draw a line, and the line ended up being... that specific. The intent was to allow normal networking events while preventing the explicit quid pro quo of a private dinner. Whether it achieves that intent is debatable.
I mean, a lavish open bar and a spread of expensive finger food at a fundraiser venue seems like it's doing roughly the same relationship-building work as a sit-down dinner, just with slightly worse posture.
The functional difference is minimal. What the rule actually does is create a compliance industry around the question of "is this food fork-required?" which is a sentence I genuinely cannot believe describes a real aspect of federal ethics law. But there it is.
What about the international comparison? Because I know the EU has a different approach to this.
The EU's transparency register requires quarterly disclosures from anyone engaging with EU institutions on policy matters, and the threshold for registration is broader than the US twenty percent rule. But here's what's interesting: the EU system covers both the European Parliament and the European Commission, which is roughly analogous to covering both Congress and the executive branch simultaneously. The US system is more fragmented, with different rules for different branches.
And the EU disclosures are more granular in some ways?
The EU requires disclosure of the estimated annual cost of lobbying activity, not just the client relationship. So you get a clearer picture of how much is being spent on specific institutional targets. The US system tells you the total spent by a firm for a client, but it's less precise about where within the government that activity is directed.
Though I'd push back slightly on the idea that more disclosure automatically means more accountability. The EU has its own influence problems, they're just structured differently.
That's fair. The presence of a disclosure regime and the effectiveness of that regime in producing accountability are two different things. You can have very detailed disclosures that no one reads, or very limited disclosures that generate significant public scrutiny. The question is always whether the disclosed information is being used. And in the EU's case, the Qatargate scandal from 2022 is a pretty stark reminder that disclosure requirements don't prevent influence operations, they just potentially make them easier to investigate after the fact.
Right, that was the case where European Parliament members were allegedly receiving cash payments and gifts from Qatari officials to influence votes and statements. A disclosure regime was in place, and it still happened.
Which is the honest ceiling on what transparency reform can achieve. It's a tool for accountability, not a guarantee of it.
Let's talk about the shadow lobbying economy, because I think this is where the real gap is between what the public disclosure numbers show and what's actually happening.
Shadow lobbying is the industry that exists adjacent to the registered lobbying industry and performs essentially the same function while avoiding the registration requirement. The mechanism is usually one of three things. First, keeping individual time under the twenty percent threshold by distributing the work across multiple people at a firm. Second, using titles like "strategic consultant" or "policy advisor" instead of "lobbyist." Third, routing influence activity through trade associations or think tanks that aren't subject to the same disclosure requirements.
The think tank route is particularly interesting because it adds a layer of intellectual legitimacy.
It's very effective. A company funds a think tank, the think tank produces research that supports the company's preferred policy position, the research gets cited in congressional testimony and media coverage, and at no point does anyone have to disclose that the company paid for the underlying analysis. That's not illegal. It's also not fully transparent.
And some estimates suggest the shadow industry is roughly as large as the registered one?
Some analyses put it in that range, though by definition it's hard to measure precisely. What we can say is that the registered six billion dollar figure is almost certainly an undercount of total influence spending in Washington. The real number is probably substantially higher.
There's also the AI dimension that's worth spending some time on, because you mentioned tech is the fastest-growing lobbying sector, and the reason is pretty specific.
The AI regulation push is creating something that policy observers are calling a gold rush moment for D.C. lobbying firms. There are no established federal AI laws in the US yet. That means the rules are being written right now, and whoever shapes the first generation of AI regulation will be shaping a framework that could persist for decades. That's an enormous incentive to invest heavily in lobbying.
And the companies doing this aren't just the obvious names. It's not just the major AI labs. It's the entire ecosystem: cloud providers, enterprise software companies, anyone whose business model is affected by how AI gets regulated.
The lobbying activity around AI has a specific character that's different from, say, pharmaceutical lobbying. Drug pricing debates have decades of established positions and coalitions. AI regulation is being built from scratch, so the lobbying isn't just about defending existing arrangements, it's about defining the regulatory categories themselves. Whether a particular AI system counts as a "high-risk" system, what disclosure requirements apply, what liability frameworks govern AI-generated outputs: these are genuinely open questions, and the answers are being shaped right now through exactly the kind of lobbying activity we're describing.
Which is also a case where the "information subsidy" framing is most powerful, because legislators genuinely don't have the technical background to evaluate these questions independently. They need someone to explain what a large language model actually is before they can regulate it.
And the companies building those models have a strong interest in being the ones doing the explaining. That's not corruption, exactly. But it does mean the technical framing of the regulatory question is being shaped by parties with a financial stake in the outcome. There's a historical parallel here that I think is instructive: when the internet was first being regulated in the mid-nineties, the companies that got in early to shape frameworks like Section 230 of the Communications Decency Act ended up with liability protections that defined their business models for the next three decades. The tech industry learned from that experience. They are not going to be caught flat-footed on AI the way some industries were caught flat-footed on early internet regulation.
So the lesson from Section 230 is essentially: get in the room early, define the terms, and the framework you help build becomes the water everyone swims in.
That's exactly the lesson. And the AI lobbying surge is that lesson being applied in real time.
Let's do the takeaways section, because I think there are a few genuinely practical things listeners can do with this information.
The most immediately useful thing is OpenSecrets. If there's a piece of legislation you care about, whether it's drug pricing, AI policy, energy regulation, anything, you can go to OpenSecrets and look up which organizations were lobbying on that bill and how much they spent. It doesn't tell you what the lobbying said, but it tells you who was paying attention and how much it mattered to them financially.
That's actually a useful signal on its own. If a relatively obscure amendment to a telecommunications bill suddenly shows up with significant lobbying spend from several major companies, that tells you something important about what's actually at stake in that amendment.
Follow the money is a cliché, but the data to actually do it is publicly available and underused. The second thing I'd flag is understanding what cooling-off periods actually mean in practice. When you see a former senior official join a lobbying firm, the formal cooling-off period prevents them from making direct lobbying contacts for one or two years, but it doesn't prevent them from advising the people who do make those contacts. The institutional knowledge and relationships are being deployed from day one.
So the cooling-off period is a disclosure mechanism more than a genuine barrier.
Largely, yes. And the third practical takeaway is about where reform efforts are actually going. The Lobbying Disclosure Modernization Act that stalled in committee this year would have moved disclosures from semi-annual to real-time. That's a meaningful change, because a six-month-old disclosure about lobbying on a bill that passed three months ago is much less useful than a disclosure that shows lobbying activity as it's happening.
Real-time disclosure would also make it much easier to connect lobbying activity to specific votes and amendments, which is where the accountability rubber actually meets the road.
The opposition to real-time disclosure is interesting because it's not primarily ideological. It's operational. Lobbying firms argue that real-time disclosure creates a compliance burden and could reveal proprietary strategic information. There's some legitimacy to those concerns, but there's also a reasonable argument that the compliance burden is a feature rather than a bug, that it should be somewhat costly to conduct influence operations on federal legislation.
The argument that transparency is too burdensome is one of those arguments that tends to benefit whoever is making it.
Generally, yes. The EU's quarterly disclosure requirement hasn't destroyed the lobbying industry in Brussels. It's adjusted to it. The US lobbying industry would adjust to real-time disclosure as well.
What I keep coming back to as we've been talking through this is the structural question: is the lobbying industry a symptom of something, or is it the thing itself? Because a lot of the reform energy focuses on disclosure and cooling-off periods, but the underlying dynamic is that legislation has enormous financial consequences, which means there will always be enormous financial incentives to shape it.
That's the honest framing of the limits of regulatory reform. You can improve transparency, you can extend cooling-off periods, you can close specific loopholes. But as long as a single piece of legislation can be worth hundreds of millions of dollars to a specific company, that company will find ways to try to influence that legislation. The six billion dollar figure is ultimately a function of how much federal policy matters to private economic outcomes.
Which is maybe the most important thing to understand about lobbying as a phenomenon. It's not primarily a story about bad actors. It's a story about incentive structures.
The individual participants are mostly doing legal things. Many of them genuinely believe they're providing valuable services, and in some cases they are. The systemic concern is about the asymmetry: who has the resources to participate effectively in this process, and whose interests get systematically underrepresented as a result.
And that asymmetry is getting bigger, not smaller, because the AI tools being developed for lobbying intelligence and targeting are going to be more accessible to well-resourced organizations than to smaller advocacy groups.
That's the next frontier that I think is going to define the next decade of this industry. AI-driven lobbying tools that can track legislative activity in real time, identify the most persuadable members on a given issue, model the district-level economic impacts of a bill instantly, and generate targeted outreach at scale. Those capabilities will amplify what well-funded lobbying operations can do significantly.
How much of that is already happening versus still on the horizon?
Some of it is already in deployment. There are firms using machine learning to analyze voting records and public statements to score members of Congress on persuadability for specific issues. Constituent impact modeling that used to take weeks of analyst time can now be done in hours. The generation of targeted grassroots-style outreach — what's sometimes called astroturfing when it's done at scale — is increasingly AI-assisted. The more speculative frontier is fully automated legislative tracking and response, where a system flags a regulatory change and generates a draft lobbying memo before a human analyst has even seen the news. That's closer than most people think.
Which brings us back to Daniel's implicit question, which is whether the secrecy framing is even quite right. Because in some ways the registered lobbying industry is remarkably documented. It's the shadow portion and the future AI-augmented portion where the opacity is going to be most significant.
The registered industry is actually one of the more transparent sectors of Washington influence. The problem is that the transparency is partial, the disclosed numbers are underestimates, and the disclosure regime hasn't kept pace with how the industry has evolved. The Lobbying Disclosure Modernization Act stalling in committee is a pretty good symbol of where reform efforts stand: there's nominal bipartisan support for more transparency, but not enough to overcome the institutional inertia.
Will the 2026 election cycle change any of that? The lobbying disclosures are already up significantly this quarter.
My honest read is that election cycles tend to increase lobbying activity without generating much reform momentum. The attention of members is on their campaigns, the attention of lobbyists is on positioning for the next Congress, and the reform conversation tends to get deferred. If anything, the window for meaningful lobbying reform usually opens right after a major scandal rather than during a normal election cycle.
So we're waiting for something to go visibly wrong before the system self-corrects.
Which is a fairly accurate description of how most regulatory reform actually happens. The LDA itself came out of a period of heightened scrutiny in the early nineties. The 2007 amendments came out of the Jack Abramoff scandal. The pattern is: disclosure and reform follow demonstrated abuse, not anticipation of it.
On that cheerful note... this was a genuinely good one. I think what I'll take from this is that the framing of lobbying as "shadowy" is almost backwards. The registered part is actually quite documented. The shadow is in the parts that are specifically structured to avoid documentation.
And understanding that distinction is useful for anyone trying to evaluate policy debates. The question isn't just "who's lobbying on this?" but "who's influencing this that we can't see?"
Thanks as always to our producer Hilbert Flumingtop for keeping this whole operation running. And a genuine thank you to Modal for providing the GPU credits that make this show possible. If you want to support the podcast, a quick review on your podcast app goes a long way in helping new listeners find us. This has been My Weird Prompts. We'll see you next time.
Take care.
