Jensen Huang Is Out. Nvidia's Crisis Is Just Beginning.

On November 12, Nvidia issued an unusual announcement.

No successor had been arranged in advance. There was no long handover period.

Jensen Huang, who had run the company for 33 years, resigned as chief executive, president, and director, effective immediately.

His replacement: Colette Kress, who had served as chief financial officer for years. She would serve as interim CEO until the board found a permanent successor.

After the news broke, Nvidia's share price swung violently. The entire AI industry was forced to confront a question it had rarely taken seriously:

Without Jensen Huang, is Nvidia still the same Nvidia?

More notable still, the upheaval landed in a sensitive window.

The U.S. midterm elections had just ended. Democrats had won control of both chambers of Congress, and a new round of AI regulation was taking shape.

Dario Amodei, the Anthropic founder who had publicly disagreed with Huang on technology policy, was preparing to take the company behind Claude toward a trillion-dollar IPO.

And behind Nvidia lay an enormous supply chain dependent on GPU orders, data-center investment, and capital-market financing.

Several seemingly unrelated things were converging in the same window.

Was Huang's departure merely a personal crisis, or could it change the future of the entire AI industry?

01Why did a private scandal ultimately force Jensen Huang out of Nvidia?

The story began earlier than November 12.

On October 29, The New York Times reported that Huang had been in an intimate relationship with a businesswoman named Alice Huang, whom U.S. intelligence agencies suspected of ties to Chinese intelligence.

Alice Huang is not related to Jensen Huang. Her associated businesses involve data centers and AI computing, and had drawn scrutiny over advanced chip transactions.

What had been a private matter quickly escalated into an issue involving national security and export controls.

On November 8, Huang publicly apologized to employees, admitting he had been unfaithful to his wife.

The same day, Nvidia's board announced an independent investigation.

Later, The New York Times and other outlets reported that Huang had also been accused of an intimate relationship with a company executive who reported to him.

Once private relationships involve subordinates, business partners, or potential intelligence-security concerns, the board can no longer think only about public image.

It must also consider corporate governance, conflicts of interest, and whether the top leadership remains fit to serve.

On November 10, Nvidia's independent directors retained an outside legal team to investigate.

On November 11, the board accepted Huang's resignation. The next day, the company made the announcement public.

To be clear: suspicion by U.S. intelligence is not a judicial finding of espionage, and the companies involved have denied wrongdoing. Nvidia's announcement did not disclose the full conclusions of the investigation.

But whatever the investigation ultimately finds, one fact cannot be changed.

Jensen Huang, who founded Nvidia and steered it for 33 years, had suddenly left the company's highest decision-making body.

02What Nvidia lost may be more than a CEO

The technology industry has a particular kind of company.

The founder does not merely build it; he turns his personality, his taste, and his judgment about the future into the company's character.

Steve Jobs at Apple was that kind of person.

Jensen Huang at Nvidia is too.

Apple did not decline after Jobs died.

On the contrary, under Tim Cook, Apple built a more efficient supply chain, launched the Apple Watch, AirPods, in-house chips, and other important products, and kept growing revenue and profit.

Yet to this day, people still argue:

Why is it so hard for Apple to deliver another shock that redefines the industry the way the first iPhone did?

This does not mean Apple stopped innovating after Jobs.

It means that a company's continued commercial success and its retention of the founder's willingness to overturn the status quo are two different things.

Uber went through a similar change.

In 2017, founder Travis Kalanick stepped down as CEO. His successor, Dara Khosrowshahi, improved operating discipline and moved Uber toward profitability.

By commercial measures, the transition worked.

But the Uber that expanded at any cost and challenged the boundaries of a traditional industry gradually became a more mature company, one that cared more about operating efficiency.

It is not simply that the later Uber was worse. It is that it was no longer the same Uber.

Huang's case may be even more distinctive than these two examples.

In 1993, he co-founded Nvidia.

For more than three decades, he was deeply involved in setting the company's technical direction.

In 2006, Nvidia introduced CUDA.

At the time, GPUs were still used mainly for graphics. Making them handle general-purpose computing meant sustained R&D spending with little prospect of a clear short-term return.

Huang persisted.

When deep learning later took off, Nvidia's years of software investment gradually turned into an advantage that was difficult to replicate.

If a professional manager focused on short-term profit had been in charge then, would CUDA have received the same steadfast support?

No one can answer with certainty.

But that is precisely the point: a founder's value is sometimes not visible on the current income statement, but a decade later.

Today's Nvidia already has a mature engineering organization, a powerful software ecosystem, and a global supply chain.

Those advantages will not disappear because Huang leaves.

The new management might even cut costs, improve margins, and push the share price to a new high.

But the next time the company needs to commit tens of billions of dollars to a technology the market has not yet validated, who else has Huang's judgment and willingness to take the bet?

A company's commercial value can be measured by its financial statements. Its soul is often not understood until the next technological shift reveals what was lost.

03The Democrats just won the midterms. The AI industry is already hitting the brakes.

If Huang's departure is understood only as a private scandal, it is easy to miss the changing political environment behind it.

On November 3, the U.S. midterm elections ended.

Democrats won control of both chambers of Congress and will hold a majority in the new Congress beginning in January 2027.

On November 6, Reuters reported that Democratic lawmakers were preparing a wave of new regulatory bills aimed at the AI industry.

From data-center power consumption to independent safety testing of advanced AI systems to the legal liability of technology companies, the AI industry that had sprinted for years might now face stricter oversight.

There is real pressure behind this.

Large AI data centers consume enormous amounts of electricity and water, which can affect local residents' power bills and infrastructure planning.

As AI companies pour in more money, the public is beginning to ask: who bears these costs? How much responsibility should technology companies take for technological risk?

Huang happens to sit at the center of this shift.

He had long argued that the United States should remain involved in China's AI chip market.

In his view, selling GPUs was not only a way to make money, but also a way to keep more developers dependent on American computing technology and the CUDA ecosystem.

But another view has long existed in American politics:

Advanced AI chips have strategic value and exports to China must be tightly restricted.

This is not simply a Democratic-versus-Republican fight. Lawmakers in both parties support stricter chip export limits.

More notable still, the Huang affair did not enter the U.S. government's field of view only recently.

According to reporting by The New York Times, The Washington Post, and others, the Trump administration had intelligence leads as early as 2025.

In May 2026, Secretary of State Marco Rubio briefed Trump on the matter.

Yet the public did not learn of the controversy through the media until October 29.

Then came the November 3 midterms, the Nvidia board investigation, and the November 12 resignation announcement.

Why was intelligence already in hand not made public or acted on sooner?

Why did the whole matter escalate so quickly during this period?

These questions deserve scrutiny.

But there is no reliable evidence that a political party orchestrated the disclosure. The initial major revelation occurred before the election, and temporal proximity alone cannot be treated as causation.

On November 11, Democratic Senators Elizabeth Warren and Chris Coons formally asked the Trump administration to preserve records related to Nvidia's chip exports to China.

Huang's personal controversy is becoming a lens through which to re-examine chip export controls and government regulatory conduct.

For the past few years, the main theme of the AI industry has been building compute as fast as possible.

Next, it may have to answer another question:

Technology can keep advancing, but where exactly is the boundary of expansion?

04Dario, Huang's public rival, happens to be at the door of a trillion-dollar IPO

There is another person worth watching in this affair.

Dario Amodei.

Anthropic's founder and the CEO behind Claude.

Dario and Huang had a public disagreement on an important question:

How should the United States control the spread of advanced AI technology to China?

Huang worried that if the United States withdrew from the Chinese market, it might actually push China to build an independent chip and software ecosystem.

Dario placed more weight on the security risks of the most advanced AI compute and argued for strict limits on exports of high-performance AI chips to China.

In 2026, the two clashed publicly on the issue several times.

Dario once compared the risk of advanced chip exports to nuclear proliferation; Huang rejected the comparison.

Their technology philosophies also differ.

Nvidia actively supports an open model ecosystem, while Claude is built mainly on proprietary models. But Dario does not advocate a blanket ban on open-weight models; the two cannot be reduced to a simple open-versus-closed divide.

Ultimately, they represent different industrial approaches.

Huang emphasized building broad global commercial influence for American computing technology.

Dario emphasized controlling the spread of the most advanced AI capabilities.

These differences were once just a route debate inside Silicon Valley.

But now Huang had suddenly departed, while Dario was taking Anthropic toward an IPO.

According to the prospectus updated on November 10, Anthropic was seeking a trillion-dollar valuation.

And so speculation appeared on X and other social platforms:

Could Dario have been behind Huang's crisis?

It is a dramatic story.

Two important figures in the AI industry, with publicly stated philosophical differences. One suddenly exits; the other reaches an important moment in the capital markets.

But a good story is not the same as the truth.

So far, no credible evidence has been found that Dario or Anthropic planned, encouraged, or leaked the details of Huang's private relationships.

Moreover, Nvidia and Anthropic are not simple competitors.

Nvidia is itself an important business partner of Anthropic and participated in investment arrangements related to its IPO.

If investors begin to worry that Nvidia will grow more slowly without Huang, they may indeed reconsider allocating some capital to AI model and application companies.

But that does not mean money will automatically flow into Anthropic when Nvidia's stock falls.

When the market reassesses returns across the entire AI industry, model companies can be hit as well.

So the question more worth asking than who was plotting behind the scenes is:

After Huang leaves, how will influence in the American AI industry be redistributed?

Washington is discussing stricter regulation.

Some model companies emphasize controlling the spread of advanced AI capabilities.

Cloud giants want cheaper, more diversified chip supply.

And Huang, who spent years driving the global AI industry's rapid expansion, no longer holds Nvidia's steering wheel.

05With Huang gone, who in the global AI supply chain is most exposed?

Nvidia does not produce everything on its own.

Its GPUs are manufactured by TSMC and others, require high-bandwidth memory from SK Hynix, Samsung, and Micron, and are assembled into computing systems by server makers and data-center operators.

Huang's departure will not bring that supply chain to a sudden halt.

But two areas deserve attention.

First, suppliers' confidence in expanding capacity ahead of demand.

A Taiwanese supplier executive interviewed by Reuters on November 12 worried that the new CEO might not lock in manufacturing capacity as aggressively as Huang did.

This is a very practical issue.

Advanced chip manufacturing, HBM memory, and server production all require planning and investment well in advance.

Huang was willing to use his judgment about future demand to push suppliers into bold expansion.

If the new management focuses more on short-term returns, upstream manufacturers may reassess new capacity.

Still, some chip products for the next year or two are already in design and production, so existing supply will not be interrupted immediately by a management change.

Second, the opportunity this may create for competitors.

AMD, Broadcom, and technology companies with in-house chip programs such as Google and Amazon all want to reduce their dependence on Nvidia.

Huang's departure may push large customers to adopt multi-vendor strategies more aggressively.

But Nvidia's hardest advantage to replicate is not GPU performance alone.

It is also the CUDA software ecosystem, system integration capability, and the software and R&D investment customers have accumulated over years.

Competitors will find it difficult to replace all of that quickly.

The real worry may not be how many fewer chips Nvidia sells next year.

It is whether, three to five years from now, when the next generation of computing technology arrives, Nvidia can still be the first to find the right direction.

06The real danger may be the AI industry's capital cycle

There is an angle more important than chips for understanding this crisis.

Money.

Over the past few years, Nvidia has not only been one of the largest equipment suppliers in the AI boom; it has also gradually become an important investor across the AI ecosystem.

Take a simplified business model.

Nvidia invests in an AI compute company.

The compute company uses the capital, along with its own funds, loans, or other financing, to buy Nvidia GPUs and build data centers.

It then rents the computing capacity those GPUs provide to AI model companies.

The model companies raise more money and train more powerful models.

The new models need more compute, which drives the compute companies to buy more GPUs.

This mechanism is not inherently problematic.

Cross-investment among companies in an industrial chain is not unusual in the history of technology.

But it has one feature:

The same company may simultaneously be another company's supplier, customer, investor, and strategic partner.

Nvidia's relationship with AI compute companies is no longer a simple buyer-seller arrangement.

New cloud computing companies such as CoreWeave and Nebius, for example, are committing large amounts of capital to AI infrastructure.

They need to keep buying equipment, and they need to ensure enough customers will rent their compute in the future.

When capital is abundant and demand keeps growing, the model runs smoothly.

But there is an easily overlooked problem.

Revenue in the AI industry must ultimately come from real usage demand.

No matter how high a valuation the capital markets are willing to give model companies, and no matter how many GPUs data centers can buy, someone ultimately has to pay for AI services.

If end-user revenue growth cannot sustainably cover equipment depreciation, electricity, operating costs, and financing costs, the whole system will come under pressure.

At that point, Nvidia's special position could cut both ways.

It is not only an equipment seller; it also helps customers in the ecosystem expand through investment and commercial partnerships.

Suppose that after Huang leaves, the new management decides to tighten ecosystem investment and cut projects with unclear short-term returns. What happens?

Some compute companies may need to find new sources of funding.

If the capital markets also lower their valuations of AI companies, those companies could face higher financing costs.

Higher financing costs slow data-center construction.

Slower construction may reduce GPU purchases.

Lower GPU demand then feeds back into Nvidia's revenue expectations.

The capital cycle that once drove growth could turn into reverse pressure.

That is the risk the entire AI industry should really watch.

Not that TSMC suddenly stops making chips after Huang resigns.

Not that AMD replaces Nvidia overnight.

It is that the capital markets begin to doubt:

Of the rapid AI demand growth of recent years, how much is supported by end users' ability to pay, and how much depends on financing and infrastructure investment continuing to expand?

Two things must be distinguished here.

The GPUs Nvidia sells are real, and demand for AI compute is real.

The existence of cross-investment in the supply chain does not make the financial revenue fake.

But real revenue does not automatically mean all downstream investment will earn an adequate return.

As companies keep pouring enormous sums into data centers, investors will eventually need to know how much cash flow those assets can generate.

And right now, Huang's departure may not be the only thing bringing that question forward.

There are also shifts in U.S. AI regulatory expectations, data-center power costs, the financing environment, and the public-market valuation test that model companies such as Anthropic are about to face.

People used to imagine the AI industry as a machine that only accelerates.

More money in, more GPUs bought, stronger models trained, more users attracted.

But if any link in the chain begins to slow, the whole system must find a new equilibrium.

Huang's departure does not prove the AI bubble has burst, but it may become the moment the capital markets begin to re-examine this funding cycle.

Because the key figure who kept the system running at high speed has left.

07Three possible outcomes for Nvidia from here

The first is the calmest scenario.

Huang leaves, but Nvidia's engineering teams and supply chain continue to operate normally.

The new CEO continues the existing strategy, products ship on schedule, and large customers keep buying.

The market gradually treats the episode as a serious but manageable corporate-governance crisis.

The second scenario deserves more long-term attention.

Nvidia remains a highly profitable company, but the new management focuses more on financial returns and no longer bets on long-term technology directions as aggressively as Huang did.

The company keeps growing; the stock may even keep rising.

But at the same time, AMD, Broadcom, and the in-house chips of large cloud providers gradually take market share.

Nvidia does not lose its advantage suddenly, but it may slowly lose the ability to define the industry's future.

The third scenario is pressure shared across the entire AI industry.

U.S. AI regulation tightens, data-center expansion faces more limits, and model-company revenue growth fails to cover rapidly rising costs.

The capital markets begin to recalculate AI investment returns.

Some highly leveraged compute companies tighten spending, upstream suppliers delay expansion, and growth expectations for chip companies suffer.

If this coincides with Nvidia's management crisis, the effects could travel through the entire supply chain.

Of course, there is not enough evidence today to determine which scenario will unfold.

Nvidia's quarterly earnings, due on November 17, will be the first important window into the answer.

Investors will look not only at revenue and profit, but also at the new management's attitude toward the product roadmap, customer investments, and supply-chain expansion.

08Can Nvidia still be great without Jensen Huang?

In 1993, Jensen Huang founded Nvidia.

At the time, AI looked nothing like it does today.

Over the following three decades, he turned a graphics-chip company into a core supplier of global AI infrastructure.

His greatest contribution was not the revenue he generated.

It was his ability to commit enormous capital and technical resources to the future while most people still could not see the direction.

Today, Nvidia still has a strong engineering organization, a mature software ecosystem, and partnerships with suppliers around the world.

Those advantages will not disappear because Huang leaves.

But how long a technology advantage lasts and whether a company can keep creating new advantages are two different questions.

After Jobs died, Apple remained a great business.

Yet people still miss the Jobs era, when the company dared to redefine the phone.

Nvidia may face a similar debate.

It may keep selling the world's most advanced GPUs and keep generating astonishing revenue and profit.

But when the next computing revolution arrives, can it still be the first company to see the future?

Meanwhile, Washington is rethinking the rules for AI development, model companies are seeking more influence, and large technology companies are looking for alternatives to Nvidia's technology.

And the capital markets are beginning to ask more seriously how much money those enormous AI investments will ultimately earn back.

In the past, Huang was both a driver of technology direction and one of the most important connectors among these industrial forces.

Now that position has suddenly become vacant.

Nvidia may not lose today because it lost Jensen Huang, but it could lose its ability to create the next era.

As for the entire AI industry, it must finally confront a question that rapid growth has long obscured:

When capital no longer pays only for stories about the future, what will actually sustain the next round of growth?