The Innovators Caucus: A Blockchain Auditor's Reading of Washington's Newest Bipartisan Signal

The email arrived on a Tuesday afternoon in Nairobi, during one of those long rains that turn the city's red clay roads into slow rivers. I was sitting with three of my teammates in the co-working space we had managed to keep through the worst of the 2022 winter, revising a curriculum module on risk management—one of the forty percent of lessons I had rewritten by hand when our donations collapsed. The subject line was unremarkable, the kind that usually ends up buried under a dozen Telegram alerts about gas fees and governance votes: "Bipartisan Innovators Caucus Launched to Support Small AI Businesses." What caught me, and held me for the better part of an hour, was not the announcement itself but what it revealed about the vocabulary our industry has quietly adopted in the eight years I have been auditing code and teaching strangers how to read a smart contract. Here was a group of American lawmakers—a Republican from the South Carolina coast and a Democratic newcomer from Virginia's technology corridor—choosing to frame an entire policy agenda around the words "small" and "innovators." Not "decentralized." Not "open." Not "public." Small. As though the moral weight of an entire technological revolution could be carried by a single adjective borrowed from the vocabulary of craft brewing and artisanal bakeries. I read the brief three times. It was thin—the kind of industry note that Crypto Briefing has been publishing more of lately, a paragraph dressed up as a policy event. But thinness, in my experience, is often where the most interesting signals hide. The silence between the blocks, as I have come to think of it, carries more information than the noise. Let me set the stage properly, because the announcement makes little sense without the landscape that produced it. The Innovators Caucus, as reported, is a new bipartisan congressional group co-chaired by Representative Russell Fry, a South Carolina Republican whose district includes the port city of Charleston and its surrounding tech and financial services corridor, and Representative Suhas Subramanyam, a Virginia Democrat elected in 2024 to represent a district that hugs the edge of the Northern Virginia technology belt. Its stated purpose is to support small AI businesses—a phrase that, in the current Washington dialect, functions as a kind of political solvent: it dissolves into whatever the listener wants to hear. For the venture capitalist, it means reduced compliance burdens. For the civil libertarian, it means resistance to surveillance-driven AI. For the labor advocate, it means jobs. For the decentralization evangelist, it might mean—if one reads generously—a foothold for alternative AI architectures that do not depend on the hyperscale data centers of four American companies. And for the crypto industry, which is the audience Crypto Briefing primarily serves, it may signal something else entirely: the possibility that AI and blockchain, two technologies that have been circling each other warily for a decade, are finally being discussed in the same policy breath. I have watched that conversation from a strange vantage point. I was thirty-four in 2017 when I served as a senior smart contract auditor for the ZEIP-20 standardization working group here in Nairobi—the same year, incidentally, that the Congressional AI Caucus was first formed in the United States. For six months I reviewed more than one hundred and fifty proposal drafts, cataloging forty-two critical edge cases in token transfer logic that quietly privileged centralized validators. Fifteen of my pull requests eventually made it into the Ethereum Improvement Proposal repository. At the time I believed, with the particular fervor of the recently converted, that the stakes were purely technical—gas optimization, reentrancy protection, the tidy horror of integer overflow. By the time I finished, I understood something different. Every edge case was a moral case in disguise. Every function signature was an argument about who gets to decide what a token can do, and who gets to bear the cost when it does it wrong. Decentralization, I wrote in one of the pull request descriptions that no one read, is not a feature—it is an ethical position about who holds power when the system is stressed. That was eight years ago. The Innovators Caucus, whatever its eventual legislative output, arrives at a moment when the same question has migrated from token contracts to model weights. Who holds power when the AI system is stressed? Who gets to audit the thing that decides whether a loan is granted, a resume is read, a village is connected to a grid? The answers being offered in Washington—and here I must be careful, because I am an auditor and auditors learn to be careful—are being shaped by a coalition that has not yet shown its full hand. Let me be clear about the limits of what we know. The brief does not tell us who else sits on the caucus. It does not tell us whether the small businesses it champions will be defined by revenue, headcount, or some measure designed to include subsidiaries of the very giants the caucus might ostensibly counterbalance. It does not tell us whether "AI" here means foundation models, applied machine learning, or—as I suspect, given Crypto Briefing's interest—the layered intersection of AI and decentralized systems that projects like Render, Akash, and Fetch have been building for years. The article's silence on these points is not an accident. It is the silence of a policy placeholder, and placeholders in Washington are filled by whoever arrives first with a staff and a budget. So here is what I want to do—not with the certainty of a policy analyst, whose trade is the confident pronouncement, but with the more modest tools of someone who has spent a career reading code and contracts for hidden assumptions. I want to trace the moral code behind this particular token of political currency. I want to ask what it would mean, in practice, for small AI businesses to be genuinely supported rather than merely name-checked. And I want to consider, against the backdrop of the current market's euphoria, whether the deeper significance of this caucus lies not in what it will do but in what it reveals about the priorities of the industry that has finally captured Washington's attention. Let me begin where any honest audit begins: with the observable record. The United States already has an AI Caucus. It was founded in 2019, co-chaired by Representatives on the House side and a rotating cast of members in the Senate. It has held hearings, issued letters, and generally served as the legislative vehicle through which members with an interest in artificial intelligence can signal that interest to donors and constituents. It has no statutory authority, no budget of its own, and no capacity to compel anything. Its power—such as it is—lies in agenda-setting: the ability to decide which questions get asked in which rooms, and which witnesses get invited to answer them. The Innovators Caucus, on the record as reported, appears to be a similar instrument wearing a different label. But the label matters, and the label is what I want to examine first, because the words "small" and "innovators" are doing a great deal of work. Consider the term "small AI business." In the American policy tradition, small business support is one of the few categories that commands near-universal rhetorical assent. You can be a progressive who wants to break up monopolies or a conservative who wants to deregulate everything, and both of you can comfortably describe yourself as pro-small-business. The phrase is a kind of universal donor in the bloodstream of political language. It carries the moral halo of the artisan, the inventor, the risk-taker—all those figures the American imagination has been trained to admire since at least the time of Franklin. But smallness, in a technical field, is not a neutral descriptor. It is a statement about a position in a supply chain. A small AI business in 2026 is not the equivalent of a small bookstore or a family farm. It is a company that depends, almost without exception, on infrastructure controlled by a handful of firms: the compute of NVIDIA, the cloud of Amazon, Microsoft, and Google, the model weights of OpenAI and Anthropic, the data pipelines of the open web. Supporting a small AI business, under these conditions, is not like supporting a small bakery. It is like supporting a small boat in a harbor whose depth is controlled by four locks. You can paint the boat however you like. You can give it a tax credit or a grant. But if the locks are closed, the boat does not sail. I learned this lesson, oddly enough, in the NFT market of 2021—not in the imagery but in the mechanics. I helped facilitate the launch of a collection called Savanna Voices, a collaboration with ten Kenyan digital artists. We structured a DAO-governed royalty system that returned seventy percent of secondary sales directly to the artists. The collection sold twelve hundred items in forty-eight hours and raised one hundred and fifty thousand dollars. On paper, this was a triumph of small creators against a system that had long ignored them. And then, three months later, the secondary market collapsed, the royalties were quietly ignored by marketplaces that had decided royalties were no longer enforceable, and the community that had formed around the collection dispersed. The artists had been supported on the way up and abandoned on the way down. The mechanism we built was technically sound—I had reviewed the contracts myself, twice, line by line. But technical soundness is not the same as structural power. The real leverage in the NFT ecosystem, as in the AI ecosystem, sat with the platforms, and the platforms could change the rules whenever the market turned. That is the trap I want the Innovators Caucus to avoid—and the trap I fear it will walk directly into. If "supporting small AI businesses" means tax credits and ribbon-cuttings without addressing the compute supply chain, the cloud pricing structures, the model access protocols, and the data acquisition bottlenecks, then it will be a program for making small businesses more comfortable while they remain small. It will be patronage dressed as policy. Now, before I am accused of cynicism—a charge I take seriously, since one of the core commitments of my writing is that cynicism is the enemy of meaningful action—let me acknowledge what the caucus might genuinely accomplish. The history of American small-business policy offers two models. The first is the Small Business Administration, which provides loans, counseling, and contracting preferences. The second is the Small Business Innovation Research program, which funds early-stage technology development across federal agencies. Both have real limitations. But both have, at certain historical moments, opened doors that would have otherwise remained closed. A caucus focused on small AI businesses could, in principle, push for three things that would matter. First, a computation credit or voucher program that gives small AI firms access to federated compute at fair prices—an idea that has been floated in various forms by the National AI Research Resource initiative. Second, procurement preferences that set aside a meaningful percentage of federal AI contracts for firms below a strict size threshold, with aggressive anti-shell-company enforcement. Third, a safe harbor for open-weight model development that shields small developers from liability regimes designed with the largest labs in mind. I would support all three. I would support them not because I believe small is inherently virtuous—it is not; some small businesses are badly run and deserve to fail—but because the concentration of AI capabilities into a handful of firms represents a structural risk to the kind of decentralized future I have spent my adult life advocating. If the AI systems that mediate human life are built, owned, and governed by four companies, then the promise of the internet as a decentralized commons will have been reversed. The blockchain world knows something about this reversal: we have watched our own ideals of decentralization become concentrated in the hands of staking cartels, exchange gatekeepers, and venture-funded oligopolies. Watching something similar happen to AI, at a larger scale and with higher stakes, is a grief I would like to spare the next generation. Here my earlier work in standards becomes relevant in a way I did not anticipate at the time. When I audited the ZEIP-20 proposals, I found that the most dangerous edge cases were not the ones that caused immediate failures. They were the ones that shifted power quietly: a default behavior in a transfer function that meant tokens could be frozen by an upstream contract; a rounding convention that transferred dust—fractions of a cent—from every transaction to the protocol's treasury rather than returning it to the user. Each of these was technically minor. Each was ethically significant. The person who wrote the convention may not have intended a power grab, but power grabs rarely require intent. They require only that no one be watching. The AI policy debate is at risk of a similar quiet drift. If the Innovators Caucus drafts legislation, the critical passages will not be the ones that name the money. They will be the definitions: what counts as a "small AI business," what counts as "AI," who counts as a "developer," and—most importantly—what counts as "compliance." Get those definitions wrong, and you can write a bill that sounds like support for the little guy while functioning as a moat around the big four. This is not speculation. It is the pattern of every technology policy cycle I have observed in my twenty-seven years around this industry, from the early internet to crypto to AI. I want to be fair to the two representatives named in the brief. Russell Fry, the South Carolina Republican, has a record that is mostly conventional for a member of his party, but his district includes the growing technology and logistics sectors of Charleston, and he has shown some willingness to engage with industry on practical terms. Suhas Subramanyam, the Virginia Democrat, was born to Indian immigrant parents, served in the Virginia legislature, and represents a district that is home to a meaningful chunk of the federal AI workforce. He has spoken publicly about the need for AI governance that is proportional and outcomes-focused. Both, in other words, are plausible partners for something more substantive than a press release. The question is whether the caucus will attract the staff and the coalition needed to convert those individual instincts into real legislative machinery. It is worth noting—since we have been talking about the crypto audience this brief was written for—that Subramanyam's district also contains a number of blockchain and fintech firms, and that Fry's state has been moving toward a more permissive regulatory posture on digital assets. The Innovators Caucus, on the face of the brief, is an AI story. But the reason Crypto Briefing covered it first is that the underlying coalition may be broader: a convergence of AI and crypto interests that have, in previous cycles, had little to say to one another. If that convergence is real, it matters enormously—not because the two technologies are the same, but because they face a common structural obstacle in the form of platform concentration, and a common policy adversary in the form of regulatory frameworks designed to protect incumbents under the banner of consumer protection. I want to pause here, because this is the point at which I want to be most careful. The bullish reading of this news is that a bipartisan caucus marks the beginning of a real political counterweight to Big Tech's dominance in AI. The bearish reading is that it is a branding exercise, and that the small businesses it champions are being used as human shields for the interests of the same large firms, only now with better public relations. The honest answer is that we cannot know yet. The brief does not contain enough information to resolve the question. What it does contain—what any such document contains, if you read it slowly—is a set of assumptions about who deserves to be supported, and those assumptions deserve our scrutiny precisely because they are quietly shaping the terms of the debate. Let me make the case for a third reading, which I find more plausible and more interesting than either of the first two. The Innovators Caucus is likely to operate primarily as a signal, not a machine. Its immediate legislative impact will be small—party caucuses without statutory authority rarely produce major legislation in their first cycle, especially under divided government and with an election on the horizon. But signals matter, because they shape what becomes sayable. Once a bipartisan group of members can stand in front of a microphone and say "small AI businesses need support," a certain set of ideas enters the Overton window. Once those ideas have entered, they can be refined, challenged, and eventually converted into actual policy—not this year, perhaps, but in the cycle after. I have seen this pattern before, and I trust it because I have lived it. When the Ethereum Improvement Proposal process began in earnest, the idea that a token contract should be auditable, transferable, and free of privileged admin keys was fringe. The first pull requests arguing for these properties were ignored or rejected. Six years later, after enough failure modes had been observed in the wild, the same properties were baseline expectations. The Overton window in computer science is slower and more stubborn than in politics, but it moves. Policy windows move too, and a caucus is a tool for prying them open. Here is where I think the crypto and AI communities have something to learn from each other—something the caucus, if it is well-staffed, might help transmit. The crypto community's great insight, the one that has survived two brutal winters and the collapse of multiple billion-dollar protocols, is that decentralization is a spectrum, not a binary, and that the crucial question is always "decentralized from whom, for what purpose?" This is a more sophisticated question than the slogans of the 2017 boom permitted, but it is the question the market has finally begun to ask after the failures of the intervening years. Chainlink's own history illustrates the tension: a project whose stated purpose was to bring decentralized data to smart contracts, and whose actual architecture has long relied on a permissioned set of node operators—the kind of helpful concession that eventually becomes an unexamined assumption. If the AI policy conversation takes the same path—first slogans, then scandals, then sophistication—the Innovators Caucus could be a modest force for good, provided it lasts long enough to reach the third stage. The risk is that it does not last, or that it becomes captured before it can mature. This is where the structure of small-business coalitions in Washington matters: they are almost always under-resourced relative to the large-firm coalitions they nominally counterbalance, because the whole point of being small is that you cannot afford the same lobbying operation. A caucus formed to give small AI businesses a voice faces the same asymmetry it exists to address. The solution, if there is one, is not more money but more clarity. Small AI businesses are not a coherent interest group. A two-person generative AI startup and a hundred-person industrial AI firm have almost nothing in common except their size category on a tax form. If the Innovators Caucus is to be effective, it will need to find the specific issue—probably compute access, possibly model liability, definitely data rights—on which the small-business coalition can stay united long enough to move legislation. That is a hard thing to do, and it is the kind of thing that usually requires either a crisis or a decade of patient organizing. I do not know whether the caucus's founders have that decade in mind. I find myself returning, as I often do when I think about these questions, to a story from my own work with the Open Ledger, the non-profit I founded in 2020 to translate DeFi mechanics into Swahili and English. We launched with twelve whitepapers and a conviction that accessibility was the truest form of decentralization. Within a quarter, we had reached five thousand unique readers—a number that, by the standards of global finance, is almost nothing, but that felt like the beginning of something real. I personally mentored twenty young developers over the following two years, many from underserved communities, and I watched them build small things—payment rails, savings models, a token for a farming cooperative—that had more genuine human value than half the protocols that raised nine figures in the same period. But I also watched them hit walls. Not technical walls, mostly, but structural ones: wallet onboarding that assumed a bank account, exchanges that required identification from countries where identification is expensive and sometimes dangerous, cloud costs that were trivial for a venture-funded team and catastrophic for a self-funded one. The small builders I mentored were not failing for lack of skill. They were failing because the infrastructure around them had been designed by and for people with different resources. I suspect the same will be true of small AI businesses if the policy conversation does not address structural access, not just size categories. This is why I read the Innovators Caucus's framing with some ambivalence. The word "innovators" is doing something clever. It is not a neutral description of people who make new things. In American political discourse it has become a kind of honorific—a title reserved for founders, tinkerers, and garage dreamers, and quietly denied to workers, researchers without patents, and the communities that bear the external costs of the technologies those innovators build. Innovation is celebrated. The costs of innovation—displacement, concentration, environmental load—are usually someone else's problem. A caucus that takes "innovators" as its mascot has already chosen whose voices it will amplify, and whose it will leave in the room next door. I want to insist, though, that this is not a reason for despair. The thirty stakeholders I worked with in 2026 to construct the African AI-Blockchain Ethics Charter taught me that the most unlikely coalitions can hold together if they are given enough time and enough honesty. Farmers, technologists, policymakers, a handful of lawyers who knew both regulatory systems—for eight months we argued and revised and argued again, and what we eventually produced was adopted by two East African regulatory bodies. Not because it was perfect—it was not—but because it was specific, honest about tradeoffs, and legible to the people it would affect. The charter's centerpiece was a mandatory transparency audit for AI-driven smart contracts, a provision we managed to include only after the farmers in the room—who had been burned before by opaque digital systems—refused to accept a voluntary framework. That experience is the one I carry with me into any conversation about AI policy in the United States, because it convinced me that the essential question is not "how do we get innovation?" but "who gets to decide what innovation means?" The Innovators Caucus has answered that question in a certain way by choosing its name. It remains to be seen whether the substance will follow the name or complicate it. Let me now do what I promised at the beginning—step back from the story and ask what the story is not telling us. The most contrarian observation I can make about the Innovators Caucus is that its most likely long-term effect is not to support small AI businesses at all. It is to provide political cover for two other projects that are much closer to the hearts of the firms that will end up funding it: a friendly regulatory posture for foundation model developers, and a preemption of state-level AI regulations that have begun to proliferate across the United States. The first project follows from the logic of the word "innovation." In American policy, "innovation" is frequently a synonym for "leave us alone." When lawmakers say that a new technology should be allowed to flourish, they are usually saying that the incumbents deploying it should not be slowed by oversight until the technology has reached a scale at which oversight becomes politically difficult. This is how the early internet was governed, and how algorithmic platforms were governed, and how the AI industry has so far largely escaped the regulatory fate of, say, pharmaceutical companies. A caucus named for innovators is unlikely to be the body that reverses that pattern. It is more likely to be the body that insists the pattern is not a problem. The second project—preemption—is less discussed and more important. In the absence of federal AI regulation, a number of American states have begun passing their own laws: disclosure requirements for AI-generated political content, restrictions on algorithmic decision-making in credit and housing, and other provisions designed to protect consumers in areas where the federal government has been slow to act. The large AI firms have consistently advocated for a national framework—arguably because a single national framework is easier to shape than a patchwork of state frameworks, and because the national framework can, if written correctly, override the state ones. A small-business-focused caucus is a natural vehicle for the preemption argument, because the argument goes: small businesses cannot afford to comply with fifty different state AI regimes, so we need a uniform standard. The uniform standard, once written, is likely to be the one the largest firms can best afford to comply with. I do not want to overstate my confidence in this reading. It is a pattern inference, not an evidence-based claim, and I would change my mind in an instant if the caucus's first substantive actions pointed in a different direction. But I have seen this movie before. The crypto industry spent a decade fighting state-level money transmission rules, and while small crypto businesses genuinely suffered under fifty different licensing regimes, the beneficiaries of the eventual federal framework were not the small businesses. They were the exchanges that could afford the compliance departments. The same dynamic is visible in the NFT royalty collapse I mentioned earlier. Marketplaces argued that enforcing royalties was technically infeasible and that creators should bear the cost of programmability. Small creators were told the new rules were for their benefit—greater liquidity, more volume, more access. The beneficiaries, when the dust settled, were the marketplaces and the traders. The artists I worked with in Kenya have largely stopped issuing NFTs. The narrative of empowerment was used to justify a redistribution of power away from the people the narrative claimed to serve. Is the Innovators Caucus going to do the same thing with AI? I hope not. But I have learned, over twenty-seven years, to be suspicious of any political initiative whose stated beneficiaries and actual funders are not the same people. The small AI businesses the caucus champions will not, in most cases, be the firms that write its talking points. The writing will be done by staffers, and the staffers will be advised by lobbyists, and the lobbyists will be paid by—well, we will see. That is not cynicism. That is the straightforward machinery of Washington, and pretending otherwise does no one any favors. My advice to anyone who wants to take this caucus seriously, on either the optimistic or the pessimistic side, is to watch three things. First, the money—who is donating, and does the caucus accept corporate members? Second, the definitions—when the caucus produces its first substantive document, how does it define "small," "AI," and "compliance"? Third, the votes—when AI legislation reaches the floor, do the caucus's members vote together, and in whose interest? Those three signals will tell us more than any press release ever could. Walking away from the hype to find the soul—that is the work I have set for myself over the years, and it is the work this small news brief demands of anyone who wants to understand it honestly. The Innovators Caucus may yet become something real: a serious, well-staffed, cross-aisle effort to ensure that the next generation of AI is not written entirely by and for the firms that already control the compute, the models, and the data. That outcome is possible. It is not likely, but it is possible, and possibility is the only thing any of us in this industry has ever been able to work with. What I would ask of the two representatives who launched it—and of the industry that is watching them so closely—is a specific discipline. Do not let the word "innovators" become what the word "disruptors" became in the crypto boom: a license to concentrate power while speaking the language of its dissolution. If the caucus's first substantive act is to define "small AI business" in a way that cannot be gamed by a subsidiary of a hyperscaler, I will be the first to applaud. If it is to draft a computation access program with teeth, I will write about it with the same care I have brought to this brief—and, for once, with a great deal more to praise. Because I do believe, despite everything I have seen, that the interesting question is not whether the caucus will fail—most new political vehicles fail, and the failures are usually quiet—but whether the industry it is meant to serve will stop waiting for Washington to save it and start building the institutions it needs on its own. The libraries where I have chosen to work—the Open Ledger, the Swahili curriculum, the ethics charter, the young developers I have mentored—are not waiting for a congressional caucus. They are small, they are slow, and they are theirs. That, more than any bill, is what I consider the real innovation. Building libraries where others build empires. Preserving the human story in digital ledgers. Listening to the silence between the blocks. If the Innovators Caucus wants to support that kind of work, I will be glad to be wrong about my skepticism. But I will not confuse its announcements for its achievements. And I would ask you not to either.