It is All One Pot of Money

Americans tend to think about the economy in sectors. The defense industry. The oil industry. The banking sector. The tech giants. This is a useful shorthand for what these industries produce — weapons systems, energy, credit, software — but it describes function, not ownership. The question this piece examines is a different one: at the level where the money actually accumulates, do these sectors exist as separate things at all?

The answer the ownership record suggests is largely no. The same concentrated pools of capital sit at the top of the shareholder lists across industries that are supposed to be distinct, supposed to compete, supposed to be governed by separate regulatory frameworks built around their differences. This piece assembles that picture — not sector by sector, but all at once — and asks whether economic activity at that scale is better described as productive or extractive.

Public Disclosure That Doesn’t Disclose: The Extraction System

There is a mechanism designed to make institutional ownership visible. Every investment manager controlling $100 million or more in publicly traded securities is required to file a Form 13F with the Securities and Exchange Commission each quarter, disclosing their holdings [SEC Division of Investment Management, FAQ on Form 13F, March 2026]. The filings are public and searchable through the SEC’s EDGAR database.[sandlineglobal]​

In theory this is transparency. In practice the filings are dense, quarterly, and filed individually by each institution — meaning the cross-sector ownership picture they collectively contain is never assembled in one place by any public agency. No dashboard shows you that the same three institutions top the shareholder lists at Lockheed Martin, ExxonMobil, JPMorgan Chase, and Microsoft simultaneously. That picture exists in the public record. It just requires someone to look at all of it at once.

There is a second limit the filing system cannot address at all. It only captures publicly traded securities. The most consequential new industry in the American economy — artificial intelligence — is developing largely through private companies not subject to 13F disclosure. The ownership structure of the firms most likely to reshape the next decade is the least visible of all. That gap is not incidental to the extraction system. It is characteristic of it.

The Ownership Layer

The ownership data for five major American industries — defense, energy, banking, and technology — was pulled from Yahoo Finance and WallStreetZen in February 2026, reflecting institutional holdings filed December 31, 2025. The same three names appear at the top of every list.

At Lockheed Martin, the largest defense contractor in the United States, State Street holds 14.63%, Vanguard 9.25%, and BlackRock 7.70% [WallStreetZen, February 21, 2026]. At Raytheon, Vanguard leads with roughly 9%, BlackRock at 8%, and State Street at 7% — together controlling approximately 24-25% of the company [Yahoo Finance, February 2026]. At ExxonMobil, Vanguard holds 10.31%, BlackRock 7.47%, and State Street 4.92%. At Chevron, Vanguard holds 9.17%, State Street 7.63%, and BlackRock 7.08% [Yahoo Finance, December 31, 2025]. At JPMorgan Chase, one of the largest banks in the world, Vanguard holds 9.85%, BlackRock 7.85%, and State Street 4.65% [Yahoo Finance, December 31, 2025].library-guides.ucl.ac+4

In technology, the same pattern holds through 2025-2026 13F filings. Berkshire Hathaway added 17.8 million Alphabet shares in Q3 2025 [Nasdaq, 2026]. BlackRock, which holds top-three positions across defense, energy, and banking simultaneously, published its own 2026 investment outlook projecting $5 to $8 trillion in AI capital expenditure through 2030 and identifying U.S. AI stocks as a priority holding [BlackRock, “AI Stocks, Alternatives, and the New Market Playbook for 2026,” 2026]. Goldman Sachs projects $527 billion in AI hyperscaler capital expenditure for 2026 alone [Goldman Sachs, “Why AI Companies May Invest More than $500 Billion in 2026,” 2026].casepoint+2

Three institutions. Every sector. The same filing quarter.

The passive and active dimensions of institutional ownership are not always as separate as the fiduciary defense implies. The institution that holds Lockheed Martin and ExxonMobil through index funds is simultaneously publishing investment theses that advocate for the AI sector’s continued expansion.

The data also surfaces something not explicitly searched for. Scroll past the top three on any of these lists and the banks appear inside the other sectors. Bank of America holds stakes in Raytheon, ExxonMobil, Chevron, and JPMorgan simultaneously. Morgan Stanley appears in Raytheon, ExxonMobil, Chevron, and JPMorgan. The institutional investors own the banks. The banks own pieces of the defense contractors and energy majors. The ownership layer is not a flat structure — it has depth, and the same names recur at every level of it.

Built to Serve Shareholders, Not the Public

The standard response to this ownership picture is that it looks more alarming than it is. BlackRock, Vanguard, and State Street are index fund managers. They hold these positions passively — not because they chose Lockheed Martin or ExxonMobil specifically, but because those companies are constituents of the indices their funds track. When millions of ordinary Americans invest in an S&P 500 fund through their 401(k), that capital gets distributed proportionally across every large public company in the index. The asset managers are fiduciaries, legally obligated to act in their clients’ financial interests. They don’t run these companies. They hold shares on behalf of pensioners and retirement savers.

This is accurate as far as it goes. It just doesn’t go far enough.

The fiduciary defense describes how the ownership was accumulated. It doesn’t address what the ownership obligates the companies themselves to do. Every corporation in that ownership map — Lockheed Martin, ExxonMobil, JPMorgan Chase, Microsoft — operates under a legal and structural obligation to maximize return to shareholders. Not to workers. Not to communities. Not to the public. The asset managers don’t need to issue directives. The corporate obligation runs automatically toward the ownership layer, whoever assembled it and however passively.

The fiduciary mechanism scales that obligation across every sector simultaneously. When defense contracts expand, when energy prices rise, when bank fees increase, the returns flow upward through the same ownership structure. The sectors generate the returns from different activities. The ownership layer captures them from the same position.

There is a second problem with the ordinary-investor framing. The returns do flow to millions of retirement accounts — but proportionally to holdings, not equally. Federal Reserve Distributional Financial Accounts data shows the top 10% of Americans hold roughly 87% of all financial assets, and the top 1% hold approximately 38% [Federal Reserve DFA, 2024]. When index fund returns are distributed proportionally, the overwhelming share flows to where wealth already sits. The ordinary investor participates. The wealthy investor captures.[superscriptify]​

The passive fiduciary defense is not wrong. It simply describes a mechanism that concentrates returns at the top while distributing costs — in the form of prices, wages, environmental consequences, and war — across the public broadly.

The AI Circuit

The consolidated ownership pattern visible in traditional industries is now emerging in AI — and in a more explicit form. BlackRock, Vanguard, and State Street hold between 60-80% of outstanding shares in Microsoft, Google, Amazon, and Nvidia through index funds [Yahoo Finance, Q3 2025-Q1 2026 13F filings]. Those figures reflect the same passive accumulation mechanism that put the same three institutions at the top of the defense and energy sector lists. In the AI industry, however, a second ownership layer sits beneath the institutional one — and it runs in a circle.[daon]​

The major technology corporations held by the institutional investors are themselves direct investors in the private AI labs. Amazon has committed approximately $50 billion to OpenAI and roughly $8 billion to Anthropic. Google has invested approximately $3 billion in Anthropic. Nvidia committed $30 billion to OpenAI’s February 2026 funding round, the largest private financing in technology history, which closed at a $730 billion pre-money valuation [New York Times, February 27, 2026; Anthropic blog, February 12, 2026]. SoftBank contributed an additional $30 billion to the OpenAI round, bringing total outside commitments to $110 billion.[iddataweb]​

These investments do not sit idle. The AI labs spend the capital on the infrastructure they require to operate — cloud computing and semiconductor chips. The primary providers of that infrastructure are Amazon, Google, and Nvidia, the same corporations that provided the investment. Bloomberg identified this pattern explicitly in January 2026 as a circular investment ecosystem: money flows from the tech giants into the AI labs and back out again through infrastructure contracts to the same tech giants. The circuit is not incidental. It is the operating model. [Note: Bloomberg URL unavailable at publication; pattern independently documented through primary investment disclosures cited herein.]

The circular structure has a disclosure problem that compounds the one identified in section II. The private AI companies — OpenAI, Anthropic, and others at the frontier — are not subject to 13F filing requirements. Their ownership structures, investment terms, and governance arrangements are not visible through the SEC disclosure system. The most consequential new industry in the American economy is the least transparent. The institutional investors who dominate public markets can be tracked quarterly through EDGAR. The private capital funding the next generation of AI cannot.

Anthropic: A Structural Exception

Every major AI lab examined in this piece operates under the same basic corporate architecture: investors hold equity, equity confers voting rights, voting rights govern the board, the board governs the company. The circular investment structure described in section V reinforces that architecture — the corporations funding the labs have both financial stakes and, in most structures, governance influence proportional to those stakes.

Anthropic was deliberately built differently.

Anthropic is incorporated as a Delaware Public Benefit Corporation, a legal form that permits its directors to balance financial returns to stockholders against a stated public benefit purpose — in this case, the responsible development of AI for the long-term benefit of humanity [Anthropic, anthropic.com/news/the-long-term-benefit-trust]. Delaware corporate law allows PBC directors to weigh public interest alongside profit. Most corporate structures don’t permit that balancing at all.[ondato]​

Anthropic’s founders concluded the PBC structure alone was insufficient. The problem, as they framed it, is accountability: a PBC makes it legally permissible for directors to consider the public interest, but it doesn’t make them accountable to anyone other than stockholders for how they do it. To address that gap, Anthropic created the Long-Term Benefit Trust.

The LTBT is an independent body of five trustees with expertise in AI safety, national security, public policy, and social enterprise [Harvard Law School Forum on Corporate Governance, 2023]. The trustees hold no financial stake in Anthropic. They hold Class T Common Stock — a special share class that carries the power to select members of Anthropic’s board of directors. That power grows over time: within four years, the LTBT will elect a majority of the board [AdwaitX, January 20, 2026]. The most recent trustee appointment, announced in January 2026, was Mariano-Florentino Cuéllar, former California Supreme Court Justice and President of the Carnegie Endowment for International Peace.idenfy+1

The critical question for a piece examining who controls what is whether Amazon and Google — the two largest investors — can override this structure. According to Anthropic, they cannot. Amazon and Google hold financial stakes but not voting shares, meaning they cannot elect board members and their votes would not count toward any supermajority needed to rewrite the rules governing the LTBT [Time, May 2024]. The failsafe amendment provision requires a supermajority of stockholders, and the largest stockholders are structurally excluded from that vote.[klippa]​

What the structure cannot fully address is the distinction between voting control and financial dependency. Anthropic requires capital at a scale that makes its largest investors materially important to its continued operation regardless of their governance rights. The LTBT insulates board composition from investor pressure. It cannot insulate the company’s strategic choices from the financial reality that frontier model development costs billions per training run and the primary infrastructure providers are the same corporations writing the investment checks. That tension is not a flaw in the design — it is the condition under which the design operates, and Anthropic’s founders have been candid about it.

No structure exactly like this has been attempted at this scale. Noah Feldman, a Harvard Law professor who consulted on the governance design, described it as experimental while expressing confidence in its underlying logic. Whether it holds under the combined pressure of a potential IPO, escalating compute costs, and the competitive dynamics of the AI race is a question the structure has not yet been tested against at full force.

Defense and Fossil Fuels: The Historical Case

The ownership pattern documented in sections III through V is newest and most visible in AI, where the circular investment structure makes it explicit. In defense and fossil fuels, the same pattern is older, quieter, and more thoroughly documented — because war spending and contract awards are public record in ways that private investment rounds are not.

The S&P Aerospace and Defense index nearly tripled in the four years following the 2003 Iraq invasion [Brown University Costs of War Project, 2023]. Lockheed Martin, Raytheon, General Dynamics, and Boeing received billions in contract awards annually throughout the Afghanistan and Iraq wars. The institutional investors documented in section III — Vanguard, BlackRock, and State Street holding between 7% and 15% of each major defense contractor — held appreciating assets throughout that period. They did not direct the wars. They did not lobby for them, at least not through any documented mechanism. The index fund mechanism accumulated those positions passively, and the positions appreciated as defense spending surged. The public paid for the spending through taxes and borrowing. The ownership layer held the assets whose value that spending created.[microblink]​

The fossil fuel holdings produced a similar pattern under different circumstances. When Russia invaded Ukraine in February 2022, BlackRock and Vanguard held indirect stakes in Sovcomflot, the Russian state-controlled shipping company responsible for transporting the fossil fuel exports funding the Russian war effort [Global Witness, February 28, 2022]. They held those positions not by choice but by index — Sovcomflot was a constituent of the indices their funds tracked. Following sanctions and Russian capital controls, both firms suspended new purchases and sought to exit their positions, though trading restrictions initially prevented full divestment. Critics noted that the passive investing model had, without decision or intent, created financial exposure to a state actor engaged in active military aggression.[shadowdragon]​

The two cases illustrate different dimensions of the same structural condition. In defense, the institutional ownership layer benefited from public military expenditure without directing it. In fossil fuels, the same ownership layer acquired exposure to a geopolitical adversary without deciding to. In both cases, the mechanism operated automatically. The returns and the exposures accumulated through the architecture of passive cross-sector ownership, not through the choices of any identifiable decision-maker.

The public absorbed the costs in both cases — through war spending, through the macroeconomic consequences of the Ukraine conflict, through the price effects that followed. The ownership layer held the assets.

The Regulatory Gap

The regulatory architecture governing American industry was built around sectors. The Federal Trade Commission oversees antitrust across industries, but its enforcement framework was designed for horizontal competition — companies in the same sector competing against each other for market share. Congressional oversight is organized by committee, with separate committees for armed services, energy, banking, and technology. Sector-specific regulators — the Federal Energy Regulatory Commission, the Office of the Comptroller of the Currency, the SEC itself — each govern their assigned domain.

This architecture assumes that the relevant unit of economic power is the firm or the sector. The ownership data assembled in this piece suggests the relevant unit is something above both. Vanguard, BlackRock, and State Street are not participants in any single industry. They are simultaneous top shareholders across every major industry, holding those positions through a passive mechanism that existing antitrust law was not designed to address.

The academic literature has begun to catch up with this reality. Research by José Azar, Martin Schmalz, and Isabel Tecu, published in the Journal of Finance, found that common ownership — the simultaneous holding of competing firms by the same institutional investors — was associated with measurably higher airline ticket prices in markets where it was most concentrated [Azar, Schmalz, Tecu, “Anticompetitive Effects of Common Ownership,” Journal of Finance, 73(4), 2018]. The mechanism they identified did not require coordination or communication between the institutional investors. The incentive structure produced the effect automatically: when the same investors hold all the major carriers, none of the carriers has an ownership constituency that benefits from aggressive price competition.[withpersona]​

The finding has been contested and the debate in the economics literature is ongoing. What is not seriously contested is that the regulatory framework has no systematic mechanism for examining cross-sector ownership concentration as a policy question. The FTC can investigate a merger between two airlines. It has no established framework for evaluating whether common ownership of all major airlines by the same three institutions affects market behavior. The gap is not a matter of regulatory failure in the conventional sense — it is a gap between the architecture of the regulatory system and the structure of the ownership it purports to govern.

The private AI investment layer widens that gap further. The circular investment ecosystem Bloomberg identified in January 2026 — tech giants funding AI labs that spend the capital back on tech giant infrastructure — operates largely outside securities disclosure requirements. The 13F system captures institutional holdings in public companies. It does not capture Amazon’s $50 billion commitment to OpenAI, Google’s stake in Anthropic, or the infrastructure contracts that return that capital to its source. The most consequential financial relationships in the fastest-moving industry are the least visible to any regulatory body.

The Pot and the Bill

The sectors are real. Defense contractors build weapons systems. Energy companies extract and refine fossil fuels. Banks extend credit. Technology companies build software and hardware. These are distinct activities conducted by distinct organizations with distinct workforces and distinct regulatory histories.

At the ownership level, they share the same three institutional investors at the top of their shareholder lists, drawn from the same filing quarter, reflecting the same passive accumulation mechanism. Those investors are fiduciaries to their clients, and their clients include millions of ordinary Americans with retirement accounts. The fiduciary relationship is genuine. So is the distribution of what it produces.

Federal Reserve data shows the top 10% of Americans hold approximately 87% of all financial assets, and the top 1% hold roughly 38% [Federal Reserve Distributional Financial Accounts, 2024]. When index fund returns are distributed proportionally to holdings, the ordinary retirement saver participates in those returns. The distribution of participation follows the existing distribution of wealth. The mechanism doesn’t flatten that distribution — it scales it.[superscriptify]​

The costs distribute differently. Defense spending is financed through taxation and federal borrowing, spread across the public broadly. Energy price increases fall most heavily on households that spend the highest share of their income on fuel and utilities — lower and middle income households. Financial sector fees and interest rates are paid by borrowers, who skew toward the less wealthy. The technology sector’s labor practices, supply chain conditions, and data extraction economics generate costs and risks that are diffuse and public while returns concentrate in the ownership layer.

The circular AI investment structure adds a newer dimension. When Amazon invests $50 billion in OpenAI and OpenAI spends that capital on Amazon cloud infrastructure, the money completes a loop within the corporate network before it reaches wages, public services, or productive investment outside that network. The loop is legal, disclosed in aggregate through various filings, and entirely consistent with the fiduciary obligations of every party involved. It is also a reasonably precise description of how returns accumulate within a closed system while costs externalize to the public outside it.

The sectors are named separately. Defense. Energy. Finance. Technology. AI. The ownership is not separate. The bill goes to the same place it has always gone.

See Also: https://open.substack.com/pub/lfitzhugh/p/insurance-then-and-now