A Board Can Fire Jensen Huang. It Can't Elect Another One.

On November 12, Nvidia announced something that startled the entire technology industry.

Jensen Huang resigned as CEO, president, and director, effective immediately.

The company he co-founded and led for 33 years will be run, for now, by Chief Financial Officer Colette Kress.

According to a filing Nvidia submitted to the U.S. Securities and Exchange Commission, the board accepted his resignation on November 11.

What happened behind closed doors, the public record is not yet enough to establish. We cannot conclude that he was voted out by the board.

But this reminded me of Steve Jobs in 1985, and Sam Altman in 2023.

Three people, three eras, all circling the same question:

When a group of professional directors has the power to decide whether a company's soul stays or goes, who guarantees that their judgment is right?

More interesting still: if they get it wrong, who is responsible?

01Collective decisions are cleverest at one thing: no one has to be responsible

Suppose a company's board has nine members.

Now it must decide whether to remove the founder.

Five agree. Four object.

Under the rules, the resolution passes.

It looks democratic. It fits modern principles of corporate governance.

But suppose the company declines five years later.

The five who voted yes can say it was a collective decision, made by the board on the information available at the time.

The four who voted no can say they were opposed from the start.

Who, in the end, is responsible for the error?

Shareholders absorb the falling share price. Employees face layoffs. Customers bear the decline in product quality. The company may even miss an entire technological era.

And those who made the decision may long since have left the board, compensation in hand, off to the next company.

Of course, directors are not free of legal responsibility. They still owe duties of loyalty and care.

But the problem is this: a decision can be entirely lawful and compliant, and yet wildly wrong commercially.

The law can pursue violations. It struggles to punish a mistaken judgment that followed due process.

This is the most delicate part of collective decision-making.

Power can be concentrated through a vote. Responsibility can be dispersed through the same vote.

When one person decides and things go wrong, everyone knows who to look for.

When nine people decide together and things go wrong, it often comes down to a single sentence:

That was the collective judgment at the time.

It sounds as if everyone took part.

In reality, no one has to bear the full consequence.

02Steve Jobs was pushed out by a group of smart people

In 1985, a famous power struggle took place inside Apple.

On one side was co-founder Steve Jobs; on the other, John Sculley, the CEO Jobs had personally recruited.

In the end, the board chose Sculley.

Jobs lost real management control and soon left Apple.

At the time, the decision was not entirely incomprehensible.

Jobs was hot-tempered, his management style forceful. He clashed with colleagues, and not every product succeeded.

To professional managers, a founder this difficult to manage hardly looked like the ideal partner.

Sculley was different.

He had mature management experience, a talent for business operations, and looked more like what a large company expected in a leader.

But the history that followed, everyone knows.

Apple sank into operational trouble, at one point nearing bankruptcy.

In 1997, Jobs returned at the helm. The iMac, iPod, iPhone, and a series of other products followed.

Apple became a world-changing company again.

Of course, we cannot simply blame every earlier Apple failure on one board decision.

But at least one thing is worth considering:

The board of that era could judge that Jobs was not an easy person to work with, but it could not necessarily judge how rare his creativity really was.

They saw a man who was difficult to manage, but may have underestimated a man who could create the future.

These two judgments are not the same thing at all.

03A board can pick a CEO. It cannot pick a Steve Jobs.

Modern corporate management has a very popular view:

A company cannot depend on any one person; it must depend on systems.

It sounds correct.

But I have always felt the statement is only half right.

Systems matter, of course.

They help a company control costs, reduce risk, standardize processes, and keep tens of thousands of employees working in order.

But can a system create the iPhone?

What would happen if the first-generation iPhone were handed to a committee of finance, marketing, engineering, and sales for repeated discussion?

Someone might worry that without a physical keyboard, consumers would not adapt.

Someone might argue that R&D costs are too high and margins unprotected.

Someone else might suggest launching an improved version of an existing phone first, to see how the market responds.

Everyone's reasoning is reasonable.

The final product, pieced together, might have every feature — and lack precisely the ability to change the world.

This is not to say collective discussion must fail, but that breakthrough innovation often needs someone willing to decide before consensus forms.

Nvidia is the same.

Huang spent years pushing general-purpose GPU computing and the CUDA ecosystem. These strategies could not prove from the outset that they would deliver the returns they do today.

They required continuous investment, and they required withstanding long-term uncertainty.

If a strategy had to prove short-term returns to the board every year, many long-term technology investments might never survive to the day they blossom.

This is where leadership is rarest.

Truly exceptional leaders are not always smarter than everyone else. They can make a judgment while most people still cannot see the direction, and they have the ability to lead an organization in turning that judgment into reality.

A board can vote on who sits in the CEO's office.

But no vote can give a person Jobs's product intuition, or Huang's long-term insight into the computing industry.

Titles can be appointed.

Leadership cannot.

04Why a room full of smart people so easily decides on something mediocre

Because in collective decision-making, being right and being safe are sometimes two different things.

Suppose you are a director.

Two choices lie before you.

The first is to support a highly creative founder who is forceful, hard to manage, and frequently challenges conventional rules.

The second is to replace him with an experienced, steady professional manager who will execute strategy as the board asks.

Which choice is easier to explain to the outside world?

Obviously the second.

Because it conforms to mature corporate-governance logic.

If something goes wrong later, you can say a prudent decision was made according to proper procedure.

The first choice is different.

Once the founder has a major problem, everyone who supported him may be asked why they did not act sooner.

And so a strange phenomenon appears.

The choice most beneficial to the company's long-term development is not necessarily the choice safest for an individual director.

Everyone wants to reduce their own risk.

Everyone wants the decision to be endorsed by the majority.

In the end everyone agrees, but that does not mean the result is correct.

This is also why some organizations, despite gathering large numbers of smart people, find it increasingly hard to produce genuinely breakthrough innovation.

Because breakthrough means controversy, uncertainty, and someone having to bear responsibility for failure.

And what collective decision-making rewards most easily is precisely the choice that is hardest to hold accountable.

When everyone is proving they did nothing wrong, no one may really care whether the company can still create the future.

05A great company's most important asset may be its leader

In 2023, OpenAI's board suddenly announced the removal of Sam Altman as CEO.

The reason given was that the board no longer trusted him to continue leading the company.

What happened next went far beyond the board's expectations.

Employees objected strongly, partners applied pressure, and the whole organization briefly faced fracture.

Within a few short days, Altman was restored to the CEO position.

In 2024, OpenAI published the results of an independent review, affirming trust in Altman's continued leadership.

The most thought-provoking part of this episode is not whether Altman has flaws.

It is that the board had the formal power to dismiss him, yet clearly underestimated his ability to hold the organization together.

What actually establishes a real leader's authority in an enterprise?

Not how large the office is, or how high the title.

It is whether employees are willing to follow, whether partners trust their judgment, whether the whole organization believes this person can lead everyone toward a future worth pursuing.

These things do not automatically transfer to the next CEO because of a board resolution.

This is especially true of companies shaped over time by a founder; the leader is often an important source of corporate culture, strategic direction, and organizational confidence.

This is not a superstition about heroic individualism.

A leader capable of creating the future likewise needs oversight, and must bear responsibility for their own mistakes.

But oversight and replacement are two entirely different things.

A good system should limit a leader's abuse of power, not treat the elimination of leadership as proof that the system has matured.

If, to avoid the risk of one person, the organization's rarest capability is ultimately destroyed, is such governance protecting the company, or protecting the board itself?

06The scariest thing is not failure. It is slow mediocrity.

Back to Nvidia.

After Huang leaves, Nvidia will of course not stop operating immediately.

It still has a massive R&D team, a mature software ecosystem, long-term customers, and a complete supply chain.

These assets will not vanish the day after one person leaves.

The new CEO may well be an excellent manager.

But there is one question no one may be able to answer immediately:

The next time it is necessary to invest ten years and bet on a technology route not yet proven by the market, who makes the call?

The next time all financial indicators say to cut investment, but one technology direction could transform the entire industry, who is willing to persist?

A professional manager may be good at running a company that has already succeeded.

But managing success and creating the next success are two completely different capabilities.

Many companies do not decline suddenly.

Their financial reports still look good. Management becomes more standardized, meetings more rigorous, decisions more democratic.

It is just that products slowly lose surprise, innovation gradually becomes iteration, and strategy starts circling quarterly results.

No obvious mistake appears.

Nor does any truly great breakthrough.

In the end, the company goes from an organization that creates the future to one that is good at managing the past.

This may be more terrifying than one business failure.

Because it is very hard to notice in time.

07Systems set the floor. Leadership sets the ceiling.

I am not opposed to boards, nor do I believe founders should enjoy unconstrained power.

Even the greatest leader may misjudge, and may cross boundaries that should not be crossed.

Board oversight therefore has its reason to exist.

But I oppose equating collective decision-making with correctness by nature, and treating individual leadership as a risk by nature.

Nine people voting are not necessarily more accurate than one person's judgment.

Compliance with procedure does not mean the result serves the company's long-term interests.

More importantly, when nine people jointly decide the fate of a company, do they truly bear responsibility commensurate with that power?

An exceptional founder may spend thirty years building a great company.

A decision that changes the company's direction sometimes requires only one board meeting.

The former requires bearing long uncertainty; the latter may require only proving that the voting procedure was correct.

Are responsibility and power between the two really equal?

After Jobs left Apple, history gave him a chance to prove himself again.

After Altman was dismissed by OpenAI, within a few days he returned to his original position.

After Huang leaves Nvidia, what the future holds cannot yet be judged.

But these events repeatedly remind us:

A great enterprise is not just a collection of systems, capital, and talent.

It also needs someone who sees a direction others cannot see, someone willing to bear risks others will not bear, someone who can make thousands of people believe in the same future.

Systems can keep an enterprise running in order.

Leadership is what keeps it moving forward.

A board can fire a Jensen Huang, but it has no ability to vote a second Jensen Huang into existence.

And the most alarming thing about collective decision-making is precisely this:

A group of people who do not need to bear the company's future alone may hold the ultimate power to decide that future.