Good night, good governance?

The provocation

Look at what the market rewarded in 2026, and the thought is hard to avoid: perhaps good corporate governance no longer counts.

SpaceX: one man, three titles, 82 percent of the votes.

In June 2026 SpaceX completed the largest IPO in history, raising 75 billion dollars at a valuation of roughly 1.77 trillion. Elon Musk holds 82.4 percent of the voting power on roughly 42 percent of the equity, and serves as CEO, CTO and Chair at once. The company qualifies as a controlled company, which lets it opt out of the requirement that a majority of the board be independent. The oversight, in effect, reports to the person it is meant to oversee.

Meta: 13 percent ownership, 61 percent of the power.

Mark Zuckerberg owns about 13 percent of Meta’s economics and controls 61 percent of the votes through super-voting shares, while chairing the board he runs. He can carry any decision on his own.

Oracle: the overseer who oversees himself.

Larry Ellison, Oracle’s founder, holds around 40 percent of the company and sits as both Chairman and CTO. The day-to-day chief executives are two insiders promoted from within. So the founder chairs the board that is meant to supervise management, runs the technology inside management as CTO, and owns 40 percent of the stock. The supervisor and the supervised are the same person.

These are not fringe cases. They are among the most valuable and most admired companies in the world. If they ignore the rules of good governance and prosper, the question follows on its own: is governance still relevant, or was the whole apparatus an expensive illusion?

Is it true?

First, why these companies matter. The Magnificent Seven, worth a combined 22 trillion dollars, now make up 32.7 percent of the S&P 500 by value, produced about 42 percent of the index’s return in 2025, and delivered 44 percent of its earnings growth over the year to September 2025. Whatever governance looks like at this small group, it looks like that where most of the market, and much of the economy, now sits.

So the question deserves a proper test. Take the thirteen largest US value creators of the past three and a half years, restricted to companies listed over the full period, together responsible for about 20.9 trillion dollars of added market value. Score each on the classic independence standard: independent chair or dual role, one share one vote or super-voting, controlling shareholder or none. Then plot value created against governance quality.

Ordered by absolute market-cap added since end-2022; the percentages show growth over the same period. Restricting the sample to companies listed throughout keeps SpaceX, public only since June 2026, out of the comparison. A note on scoring: Broadcom separates the CEO and Chair roles, but the Chair is co-founder Henry Samueli, so the company scores Mixed rather than Strong.

The result is thin. The coefficient of determination is 0.25, and with thirteen companies that is statistically indistinguishable from zero. Remove the single outlier, NVIDIA, and it falls to 0.05. There is no reliable relationship.

The finding holds whichever way value creation is measured. Switch from dollars added to percentage growth and the coefficient stays low, then collapses toward zero once NVIDIA is removed. The absence of a relationship is not an artefact of the metric.

Two limits of method belong on the table. With thirteen companies the exercise is illustrative, not a statistical proof; the point is that no pattern is visible, not the decimals. And market-cap growth mixes share-price performance with buybacks and stock issuance, so total shareholder return would be the cleaner measure. A relationship this weak would not survive the switch either.

That disposes of both easy stories. The provocative reading, that the best performers are the worst governed, does not hold: NVIDIA and Apple are cleanly governed and sit at the top. The reformer’s reading, that good governance drives value, does not hold either: Alphabet, Meta, Oracle, Walmart and Berkshire are controlled and create enormous value. Governance quality and value creation are decoupled.

Why good governance still matters

If governance does not explain returns, the temptation is to conclude that it does not matter. That conclusion is wrong, and five honest caveats show why.

1. Survivorship. The sample is a winners’ list. The founder-control disasters are missing, because they collapsed (FTX), imploded at the IPO (WeWork) or ran a fraud in the dark (Theranos). Judge control by the survivors and it looks harmless. It is not.
2. Reward, not cause. A founder who delivers earns the right to rule. On that reading, concentration is the consequence of success, and the score measures a prize rather than a driver.
3. Time. Three years is short. The cost of weak governance shows up over decades, in botched succession and entrenchment, not inside a bull market. The window may be too small to see the bill.
4. Form, not quality. A first-rate controlling owner, Warren Buffett being the model, can serve shareholders better than an independent but passive board. The scale reads structure, not judgment.
5. Value for whom. Market value for shareholders, yes. Governance also protects others: minority investors, creditors, employees. The chart captures one dimension of a wider question.

Put those caveats together and the conclusion is clear: returns cannot justify governance. What governance buys is something else.

Governance is insurance against tail risk.

It is the brake for the day the founder is wrong. Meta shows the mechanism live. Zuckerberg’s voting control has let him commit more than 80 billion dollars of cumulative losses to Reality Labs since late 2020, through years of open investor objection. Whether that bet eventually pays off is unknown, and that is the whole point. No one can know in advance which large bet is the ruinous one, and at Meta no one could have overruled him either way. When a single person can carry a commitment that size alone, the board cannot check it. It can only watch. That is what governance insures against: the ruinous decision that no one is able to stop.

The measure shifts to accountability and minority protection.

Classic governance rests on one mechanism: an independent board holding management to account. When one shareholder controls the votes, that mechanism is broken by construction. The owner elects the board that is meant to check him. And the classic independence test cannot even see the problem: Alphabet separates the roles and appoints an independent chair, everything the test asks for, and the founders still hold about 51 percent of the votes. The form is immaculate and the power sits elsewhere.

None of this is a scandal. Investors buy into these companies knowing the structure, and the market can price the discount. A majority owner also has a legitimate claim to more say than a small shareholder; that is how ownership works. The open question is where that power ends: what may he do at the expense of the minority, the outside investors who hold stock but few votes, while staying within the rules?

SpaceX shows the pattern live. Its IPO filing disclosed that the company bought 131 million dollars of Cybertrucks from Tesla in 2025, between 6 and 9 percent of every one sold that year, and it paid full list price. Ordinary buyers did not. 2025 was the year Cybertruck sales collapsed to half the prior year’s level, and Tesla was moving the trucks with discounts that ran to 10,000 dollars a vehicle. In the final quarter alone, SpaceX took more than 18 percent of all Cybertruck deliveries. So the minority investors of one Musk company paid top dollar to prop up the weakest product of another, alongside roughly 700 million dollars of Megapack batteries bought across 2024 and 2025, most of them by xAI, which SpaceX absorbed in February 2026. Every dollar stayed inside the same empire, and the outside shareholders carried the cost.

That is the situation minority protection has to handle: transactions that are legal, disclosed, and still tilted toward the man at the top. A board elected by the owner will not stop him. So the safeguards have to sit in the structure itself: related-party deals approved by independent directors or by a vote of the minority, sunset clauses that let super-voting rights expire, and full disclosure of every dealing with the controlling shareholder. Independent directors cannot hold a majority owner to account, but they can still vet the transactions where he sits on both sides of the table. That is the job that remains, and it is the one worth measuring.

So, is it dead?

US corporate governance is not dead. It has split into two worlds. For the many listed companies with dispersed ownership, the independence model still does its job, and it is slowly tightening: combined CEO and chair roles fell from 47 to 42 percent of the S&P 500 between 2020 and 2025, and independent chairs rose from 33 to 39 percent over the same years. For the small group of founder- and family-controlled giants that now dominate the index, the model barely applies, and, as the value-creation figures show, the market has rewarded them all the same.

So the standard governance checklist needs a new purpose. Proxy advisers and rating firms still score companies on independent chairs, split roles and board independence. For a controlled company those measures ask the wrong question, because a board the owner elects cannot hold the owner to account. The features worth scoring are different, and they are just as observable. On resilience: is there a credible successor, and a board that can function without the founder, so the company survives the day his judgment fails? On minority protection: are related-party deals approved by independent directors or by a vote of the minority, do the super-voting rights carry a sunset, and is every dealing of the controlling owner disclosed in full? Those are answerable questions, and they measure whether power is checked rather than how the chart is drawn. The title on the org chart tells you what the governance looks like. The voting structure, and what protects the minority under it, tells you whether it works.

Sources

The Conference Board via Harvard Law School Forum on Corporate Governance, CEO/Chair Leadership, May 2026. Spencer Stuart U.S. Board Index 2025. Forbes Investor Hub and Motley Fool, S&P 500 weight and combined market value of the Magnificent Seven, June and July 2026. First Trust Advisors via Proactive Advisor Magazine, Magnificent Seven share of S&P 500 earnings growth, data to 30 September 2025. SpaceX Form S-1 and CNBC/CNN, IPO terms and Musk post-IPO voting control (82.4 percent), June 2026. Bloomberg and Business Insider, SpaceX related-party purchases from Tesla, May 2026. Carscoops, SpaceX share of Q4 2025 Cybertruck deliveries, May 2026. Forbes, Electrek and New York Post, Cybertruck discounts and inventory build-up, 2025. Kelley Blue Book, Cybertruck sales 2024 and 2025. CNBC, Meta Reality Labs cumulative losses, January 2026. CNBC, Berkshire Hathaway leadership transition, January 2026. AInvest, Walmart Walton-family voting control, 2026. Forbes, Larry Ellison and Oracle. Market capitalisations: companiesmarketcap.com (end-2022 and end-June 2026 figures).