Own or Be Owned

A Tale of Two Tomorrows

The year is 2032. The machines have arrived—not with the clang of a factory revolution, but with the quiet hum of server farms the size of small cities. AI systems designed and operated by a handful of trillion-dollar corporations now handle most of the world’s cognitive and physical labor. Code writes itself. Factories run themselves. Medical diagnoses arrive before the patient finishes describing symptoms. Warehouses are staffed by robots that never sleep, never unionize, and never ask for a raise. There is no shortage of goods or services. The machines can produce almost anything at a marginal cost approaching zero.

So far so good, but the question is, who owns it?

In the optimistic telling everyone benefits. Productivity explodes, costs collapse, and abundance spreads outward. But abundance does not distribute itself. Corporations are not charities. Governments are already structurally broke and are not built to socialize gains they do not control. 

If the AI revolution makes millions of people less economically necessary while leaving the productive assets in a few private hands, most people will not experience liberation. They will experience dependence.

That is the real fork in the road. In one future, AI creates broad prosperity because ownership of the productive infrastructure is broad. In the other, a small cluster of companies owns the systems that perform an ever-larger share of economically valuable work, and everyone else lives downstream of their incentives. Every autonomous taxi ride, AI-generated legal brief, robotic surgery, contactless grocery purchase, and drone-delivered order sends revenue back to the same small set of firms and their shareholders. The people who own the machines get rich. The people displaced by them are told to admire the efficiency and be thankful for whatever handouts those in power choose to provide.

The bitter irony here is that most of the institutions that are poised to define the AI age began by speaking in the language of public benefit. OpenAI was founded as a nonprofit research lab with a charter focused on benefiting humanity. Anthropic arrived as a public benefit corporation focused on safety. Yet the gravitational pull of capital did what it always does. Models became more expensive. Competition intensified. Billions arrived with strings attached, and those strings pulled toward control, commercialization and shareholder value. The organizations that were supposed to restrain the corporate logic of AI were absorbed by it.

The pattern is old. The scale is new. Railroads, oil, and telecommunications concentrated wealth by changing specific industries. AI threatens to concentrate wealth by changing the status of labor itself. If machines can perform a very large share of today’s work, and a handful of companies own the energy, compute and machines that perform that work, those companies do not merely dominate a sector. They dominate the productive base of the economy.

It’s like a game of Monopoly where one player doesn’t just own all the properties—they also own the dice, the board and the bank. The other players are still technically in the game, but only because the owner lets them keep rolling.

The Empirical Case for Concern

This is not speculative dystopia. A 2024 study in Technology in Society, using cross-country measures of AI capital stock across the United States, Europe, and Japan, found a statistically significant relationship between AI capital accumulation and wealth disparity. What makes that result striking is that the study used data from 1995-2020 where AI capital referred to industrial robots, enterprise software, and early compute infrastructure rather than the generative systems now remaking white-collar work. If narrow AI was already widening the gap across three of the world's largest economies, generative AI is unlikely to reverse the sign.

You can see the gap widening in current equity and labor markets. From 2016 through 2025, the Magnificent Seven massively outperformed the broader index, with AI optimism acting as an accelerant. It goes without saying that gains accrued disproportionately to people who already owned those firms. Meanwhile, even the IMF has warned that generative AI threatens not only routine work, but high-skill roles once thought insulated from automation. The first cuts are already appearing where prestige used to offer protection: junior bankers, accountants, lawyers, HR staff, software workers. Next will come the supervisors, then the organizations built around managing them. Who needs an accounting partner when transactions are digital and feed directly into software that handles the books in real time? Who needs a junior lawyer when Harvey drafts the contract? US firms announced nearly 950,000 job cuts through September 2025—the highest since COVID—and for the first time, companies are not planning to hire these workers back. As one mortgage broker put it: "There's no exit ramp. When you saw jobs disappear during COVID, we knew those jobs were coming back. With AI, they're not." And that's before the robots. This matters because upper-middle-class labor is not a sideshow to modern consumption; it is the tax base, the mortgage base, and the spending base of advanced economies. AI is dismantling that model from the inside out.

The geopolitical dimension makes the picture still harder. AI capability is far from an American monopoly, and the race is not being run on a single track. China has paired state-directed industrial policy with brutal domestic competition, producing scale advantages in electric vehicles, autonomous driving and increasingly in AI research and deployment. By 2023, China accounted for 22.6% of all AI research citations globally, compared to 13% from the US, according to Stanford University's AI Index Report 2025.  If the AI economy becomes both more automated and more geopolitically competitive, ownership concentration will not merely be a domestic distribution problem. It will become an international one, dividing not only firms from workers but AI-producing societies from AI-dependent ones.

Neither Companies Nor Governments Will Close the Gap

The comforting answers are the least convincing ones. Start with corporations. The same firms now talking about “benefiting humanity” are under enormous pressure to capture the returns from the most consequential technology in generations. Expecting them voluntarily to socialize those gains is wishful thinking dressed up as strategy. Capital does not build world-changing systems only to surrender the upside at the moment of maximum leverage.

Governments are not a clean answer either. Universal basic income may become politically necessary in some places, and it may provide a floor. But a floor is not ownership. A transfer payment funded by taxing concentrated machine-generated wealth leaves the underlying power structure untouched. Worse, if AI compresses labor income and narrows the taxable base before states have built durable redistribution mechanisms, the fiscal capacity to close the gap may weaken precisely when the need becomes acute. A society in which a small number of entities own the productive base and everyone else receives an allowance is more stable than anarchy, but it is far from free.

Owning the Means of Production in the AI Era

The phrase ‘means of production’ deserves to be brought back into the spotlight. For two centuries, the central economic question has been who owns the tools, factories, and systems that generate wealth. The industrial era concentrated ownership in factory owners. The digital era concentrated it in platform companies. The AI era threatens to concentrate it in infrastructure that powers intelligence itself: energy, compute, connectivity, data, robotics and the systems through which autonomous agents transact. This concentration will happen faster than any prior technology wave, and the window to build alternative ownership structures is correspondingly narrow.

If AI-powered machines are going to do most of the world’s economically valuable work, people need ways to own the machines—or more precisely, the infrastructure layer on which those machines run. Not symbolically. Not only through wages, subsidies, or taxes levied after the fact. Directly.

This is where tokenized infrastructure becomes interesting. Strip away the jargon and the speculative debris, and the underlying idea is straightforward: productive assets that would normally sit on a corporate balance sheet can instead be owned fractionally by a broad base of participants, with revenue rights and governance rights embedded directly into the system that operates them.

Decentralized Physical Infrastructure Network (DePIN) is one of the clearest emerging expressions of this concept, but the proposition is broader than the label DePIN. In its best form, it is not “crypto for infrastructure.” It is an ownership architecture for a world in which infrastructure matters more than firms and execution is increasingly automated.

Why Tokenization Matters 

• The first reason tokenization matters is control. Buying shares in a centralized AI company may give investors economic exposure, but it does not materially decentralize power. Decision-making remains at the board table. Token-based systems can, at least in principle, distribute governance rights more directly across a network’s ownership base: fee structures, access rules, upgrade paths, validator requirements, operating mandates. That does not guarantee democracy, but it does create a different starting architecture from a Delaware corporation.

• The second reason is execution. Traditional equity assumes that returns depend on the efforts of managers and employees. In an AI economy, more of the work that generates returns will be performed by software, robots, and automated protocols. Ownership structures designed for a world of human managerial intermediation may become a less natural fit. Tokenized claims can attach directly to the infrastructure that produces value, rather than to a company whose human staff sits between the asset and the cash flow.

• The third reason is machine-native settlement. If autonomous agents increasingly buy compute, bandwidth, energy, storage and mobility from one another, economic infrastructure that can be settled programmatically and globally starts to matter. Here the point is practical rather than ideological. A machine economy should not be tied to the fiscal policy of any particular country or its fiat currency, nor will it wait politely for paper-heavy systems designed for a slower, more human pace of exchange.

None of this means tokenization is a magic wand. Most token structures have been badly designed, many governance systems drift toward plutocracy, and much of crypto has confused trading volume with value creation. Those are real objections, not rhetorical obstacles to be waved away. A token that carries no enforceable claim, no meaningful governance right, and no relationship to a productive asset is not decentralization. It is branding.

But the skeptic’s critique, properly framed, is not an argument against the ownership model. It is an argument for building better versions of it. The relevant comparison is not between an idealized token system and a flawed public company. It is between two imperfect worlds: one in which AI infrastructure is owned almost entirely by large firms and states, and another in which at least some of that infrastructure is widely ownable, governable, and machine-native from the start.

Seen this way, tokenization is not primarily about access or liquidity, though both matter. It is about designing ownership for an economy in which productive assets become more autonomous, more networked, and more important than traditional labor relationships.

The challenge is to make those claims real: legally legible, economically enforceable and resistant to simple capture by early insiders. The hard part is not minting a token. It is tying a token to something in the world that actually matters.

That “something” can take many forms: compute capacity, battery storage, bandwidth, fleet revenue, data-center output, toll roads, shipping containers, municipal cash flows, even rights to consume the infrastructure itself. What unifies these cases is not the technology stack but the ownership logic. Instead of leaving economically valuable infrastructure accessible only to institutional capital, the asset becomes fractionally ownable and its returns flow directly to a distributed base of holders.

The internet offers the clearest analogy. The base layer was built as a commons. The great fortunes of the digital era were made not by owning the internet as such, but by building centralized platforms on top of it and capturing the value created by participants. A decentralized ownership model aims to change that pattern. The point is not decentralization as an aesthetic preference. The point is making sure the value generated by a network accrues to the people who build, supply, and use it rather than only to the company that encloses it.

Where the Logic is Strongest

Not every sector is equally compelling. The best cases are the ones where three things converge: the asset is economically central, the cash flow is real, and the ownership structure could plausibly shape power rather than merely create another tradable wrapper. Three areas stand out.

• Autonomous vehicle fleets. The robotaxi is the most vivid near-term example of what AI-driven displacement looks like in physical form: a machine that does the job, owned by a corporation, generating revenue that flows entirely to that corporation and its shareholders. The people who once drove those routes earn nothing. The people who live in the cities those vehicles operate in earn nothing. Several projects are beginning to explore tokenized fleet ownership, but the model as currently practiced tends toward investment exposure — fractional financial participation in fleet revenue — rather than genuine governance. The further step is a community-owned autonomous fleet, where the residents of a city hold token-weighted governance rights over fleet deployment, pricing, route prioritization, and the terms on which the fleet operator accesses public road infrastructure. The roads are public. The license to operate on them is granted by the public. There is no structural reason the returns should not flow back to the public — only the absence of a protocol capable of enforcing that logic.

• Distributed grid storage and energy infrastructure. Peer-to-peer energy trading has been tokenized in several markets, allowing prosumers to sell surplus generation to neighbors. What has not been built is a protocol for distributed ownership of the storage layer — the batteries, substations, and grid-balancing infrastructure that determine whether renewable energy is usable at scale. Home battery systems already exist by the millions. Their owners participate in grid balancing programs run by utilities, contributing storage capacity in exchange for modest rate credits. The economic value of that aggregated storage capacity is captured almost entirely by the utility. A protocol that tokenizes the storage capacity itself — allowing anyone who contributes a battery node to hold a governance-bearing ownership stake in the virtual power plant formed by the aggregate network, with returns flowing via smart contract rather than through a utility billing cycle — would transform distributed energy from a grid service into a commons. The hardware exists. The coordination layer does not.

• AI compute infrastructure. Decentralized GPU networks have made progress in aggregating idle compute and routing it toward AI workloads, and contributors earn tokens for providing capacity. But earning tokens for compute time is not the same as owning a governance stake in the systems that train models, serve inference, and define who gets access under what constraints. If AI is the strategic industry of the next decade, then connecting compute ownership to data and model governance may be the highest-leverage place to experiment with broader ownership before the stack hardens around a few incumbents.

Other sectors may follow: wireless connectivity, municipal infrastructure, logistics, industrial equipment. But the principle remains the same. The opportunity is greatest wherever a critical asset is becoming more automated, more indispensable, and more likely to be consolidated unless an alternative ownership structure is built early.

That last point matters. Once an ownership pattern calcifies, it becomes extraordinarily difficult to unwind. The public usually enters after the infrastructure is built, the rents are assigned, and the legal architecture protects the incumbents.

* * *

This post-AGI world order debate is often framed as a question about jobs. It is really a question about ownership. In an economy where machines do most of the work, wages become a weaker claim on prosperity than ownership.

Retraining, redistribution, and welfare may all remain necessary, but they are downstream conversations in a negotiation about how to distribute the proceeds from a productive base that someone else controls. It is a negotiation that has historically gone poorly for the people without ownership stakes. The real question on which to focus is who owns the productive assets generating the surplus in the first place.

Tokenized, fractional, protocol-governed ownership is not a complete answer. Legal enforceability across jurisdictions remains unresolved. Governance can be captured. Regulation may lag for years. Many projects will fail, and some deserve to. But none of those problems change the underlying direction of travel. If AI continues to concentrate productive power, broad ownership of infrastructure moves from a niche crypto idea to a serious political-economic imperative.

The industrial era concentrated wealth in factories. The digital era concentrated it in platforms. The AI era is on track to concentrate it in machines, models, and the infrastructure beneath them—faster than either of the previous two waves and at far greater scale. The window to build a different ownership structure before that pattern hardens is narrow.

Own the infrastructure or accept a future in which someone else does—and everything else in the economy is simply arguing about the allowance they are willing to give you.

Disclaimer: The views expressed herein are solely those of the author and are provided for informational purposes only. They do not constitute investment, legal, tax, or other professional advice, and should not be relied upon as the basis for any investment decision. References to specific companies, protocols, tokens, or sectors are illustrative only and do not constitute a recommendation or solicitation to buy or sell any security, token, or other asset. The author is a founding partner of No Limit Holdings, a venture capital firm that invests in blockchain and digital asset-related businesses, and No Limit Holdings and/or its affiliates may hold positions in, or have other economic interests in, certain companies, protocols, tokens, or sectors discussed herein.