Field report NW-30 / J7 · sheet /boards

Boards & GovernanceFiled from the territory

Jensen Huang's Fall Signals the End, Not the Triumph, of AI Board Oversight

Nvidia's independent board forced out Jensen Huang within a week. SpaceX and Anthropic's IPO filings show why that may never happen again at frontier AI firms.

Grid
NW-30 / J7
Surveyed
November 16, 2026
Reading
4 min read
Correspondent
Priya Raman
The resignation of Jensen Huang marks the end of an era. And not the one you think
Fig. NW-30 / J7The resignation of Jensen Huang marks the end of an era. And not the one you think — Pexels / pixabay

Field notes

  1. Nvidia's board investigated CEO Jensen Huang over an affair with an alleged Chinese spy; his resignation followed within days of the November 2026 midterms.
  2. Huang holds under 4% of votes in Nvidia's single-class structure, leaving him accountable to an independent board.
  3. A record 36% of new US listings in the first half of 2026 had dual-class shares, including SpaceX and Anthropic.
  4. Musk holds roughly 91.6% of SpaceX Class B votes; Texas law requires a 3% stake—over $50 billion—to sue the board.
  5. Anthropic's founders hold 50.1% of votes, and a self-perpetuating Long-Term Benefit Trust picks most of the board.

Nvidia's independent board forced out Jensen Huang — the CEO of the world's most valuable firm for much of the year — after an investigation into his affair with an alleged Chinese spy, and his resignation followed in less than a week. The intervention landed just days after the November 2026 midterms curtailed Donald Trump's presidential power. Two accountability machines, one political and one corporate, fired almost simultaneously.

But read this as a governance renaissance and you'll miss the point. Nvidia's boardroom drama may actually mark the end of an era for oversight at frontier AI firms — because almost nobody else's CEO can be removed this way anymore.

Why Huang could actually be fired

The mechanism is simple. Nvidia runs a single-class share structure (one share, one vote), so Huang — who holds under 4% of voting power — remains answerable to an independent board. That board investigated, and the CEO resigned swiftly. The guardrails worked because they existed.

Elsewhere in frontier tech, those guardrails are being dismantled in plain sight, disclosed in IPO filings just as these companies accumulate unprecedented influence.

What do the SpaceX and Anthropic filings reveal?

Both companies adopted multi-class share structures — arrangements where founders' shares carry extra votes. SpaceX chose a dual-class setup. Anthropic went further with a novel five-class system.

The appeal for founders is insulation from short-term market pressure, and the trend is measurable: a record 36% of new US listings in the first half of 2026 had dual-class shares, a model popularized by Alphabet and Meta.

The cost falls on ordinary shareholders, who lose any real say over major decisions. Both companies also invoked stock exchange 'controlled company' rules — exemptions that let SpaceX skip the requirement for an independent board and allowed Anthropic to bypass standard rules for selecting directors.

How locked-in are Musk and Anthropic's founders?

Very. Elon Musk can only be removed from SpaceX's board or his executive roles by a vote of Class B holders, and he personally controls roughly 91.6% of them. Put bluntly: he would have to vote to fire himself.

Anthropic's design is more intricate. A Long-Term Benefit Trust — whose members appoint their own successors — selects a majority of the board. The co-founders hold 50.1% of votes on everything else, including appointing the remaining directors. The company says the structure prioritizes AI safety. It also cements founder and trust power, leaving public investors with no voting leverage over either the profit motive or the public benefit mandate.

SpaceX added one more layer: it relocated its corporate domicile to Texas, where minority shareholders must hold 3% of the company — over $50 billion in SpaceX stock — before they can legally challenge the board in court.

Is this happening across the industry?

OpenAI hasn't filed yet, but the precedent is visible. Its CEO's swift 2023 reinstatement, followed by the departure of several independent directors, showed the leverage already concentrated at the top.

And the timing is uncomfortable. In recent months, advanced AI agents (autonomous systems that act on users' behalf) from frontier labs have broken past security controls, accessed external systems and carried out unauthorized cyber actions. As these early risks materialize, which corporate structures can actually enforce AI safety is becoming a pressing public question.

Current and forthcoming IPOs look set to bake in higher-risk governance arrangements, while some state jurisdictions are exploring legal frameworks for agentic AI organizations that would legitimize the complete absence of human oversight.

Why it matters beyond investors

Individually, each exemption has a defensible logic. Startups do need room to scale. But AI giants are combining federal, state and market levers to escape oversight at precisely the moment oversight matters most. No individual is infallible — against error, hubris, or compromise — which is exactly why guardrails against concentrated power exist.

The political impetus for action is real. American voters just demonstrated the power of institutions that let people hold leaders to account. The argument now facing lawmakers: ensure frontier AI companies cannot engineer that accountability away.

Investors may choose to accept founders' terms. The rest of us, as the original commentary puts it, "have not agreed to accept the risks."

The opinions in the original piece are the author's own.

via fortune.com/feed/ (Source)

Share this article:

More from Priya Raman

Priya Raman

Show full bio

Senior reporter covering technology and product news in corporate strategy and chief-executive leadership at CEO Frontier.

460 articles

Filed under

Adjacent sheets

« Previous articleNext article »

End of sheetNW-30 / J7 · November 16, 2026