Jensen Huang's Troubles Are Just Beginning: The Triple Pressure of Nvidia's Governance, Policy, and Capital Expenditure
A commentary article republished by Huxiu argues that beyond Jensen Huang's personal controversies, Nvidia also faces triple pressure: a heavy reliance on a single leader, unstable policy support from Washington, and the capital-intensive nature of AI infrastructure. The article notes that Jensen Huang has more than 60 direct reports and that the company has long lacked a COO role; that 40 of the 41 local ballot measures tallied by Reuters supported restricting data center construction; that the five major cloud providers' capital expenditure in 2026 is projected to approach $800 billion; and that Nvidia plans to partner with financial institutions to mobilize more than $500 billion in third-party capital.
Why read this
AI score 60
The article breaks down the risks currently facing Nvidia across three lines—governance structure, policy environment, and capital expenditure—offering a lens for understanding the shift in its valuation logic.
Topics
Body · AI translation
**Key points**
**1. Nvidia's governance relies heavily on Jensen Huang**
He has more than 60 direct reports and holds few one-on-one meetings; the flat management structure depends on his personal judgment. Although the board has launched an independent investigation, the company lacks a COO and a clear mechanism for succession and delegation of authority.
**2. Washington's support is not solid**
Beyond export controls, the regulatory cycle after the midterm elections and local opposition to data center construction will simultaneously constrain its global sales and domestic infrastructure. Of the 41 local referendums tallied by Reuters, 40 supported restrictions.
**3. AI infrastructure has become a capital-intensive business**
The five major cloud providers are expected to spend nearly $800 billion in capital expenditure in 2026. Nvidia plans to join with financial institutions to mobilize more than $500 billion in third-party capital, but risks related to demand, returns and customer concentration remain unresolved.
**4. The valuation logic is shifting toward infrastructure credit**
The market will simultaneously scrutinize compute utilization, whether AI revenue can cover costs, governance compliance and long-term stability. Nvidia needs to build institutions that do not depend on Jensen Huang to answer every question. 2026-11-09 18:13 Jensen Huang's troubles have only just begun Beyond the Page Quick Look
This article is from the WeChat public account: Beyond the Page, author: Huahua
At an industry cocktail party in Taipei in June 2024, Jensen Huang, at the peak of his reputation, was surrounded by a group of tech leaders raising their glasses to toast him. Right beside him was a woman in pink.
Two years later, this photo appeared in a New York Times report.
The ensuing controversy escalated further after Jensen Huang issued a public apology letter on November 8. In the letter, he admitted to an extramarital affair and apologized to his wife, family, as well as Nvidia employees, the board, partners and shareholders.
As things have developed, this has long since gone beyond a simple personal moral crisis.
When a tech giant with a market value of trillions of dollars has its business model, global customer trust and even globalization issues highly tied to the personal reputation of the same person, a personal controversy may spread outward along the company's business relationships and ultimately touch the foundation of trust in the global AI industry chain.
Jensen Huang's troubles have only just begun.
I. Nvidia cannot do without Jensen Huang
The first core contradiction exposed by this crisis is that behind Nvidia's flat organizational management lies a high dependence on a single leader.
What exactly has Jensen Huang built for Nvidia?
Over the past few years, he has completed a key business transformation, taking Nvidia from a GPU chip supplier to the core position of global AI data centers.
In an interview with Lex Fridman in March this year, Jensen Huang revealed that he has more than 60 direct reports. He rarely schedules one-on-one meetings and prefers to resolve issues through collective discussion.
This de-layered management system allows decisions among chip architecture, software ecosystem and supply chain to collide directly, reducing the losses caused by hierarchical transmission.
But it also leaves a problem: the company's decision-making efficiency is highly correlated with Jensen Huang's personal judgment.
When a large number of important decisions revolve around one person, that person's private risks may enter the company's governance system. It is difficult for Nvidia to remain completely aloof from how a personal controversy affects customer relationships, internal information management and business decisions.
The question is, is there a way to reduce this dependence?
Silicon Valley is not without precedents.
This year, Apple's transition of the CEO position from Cook to Ternus provided a reference. Ternus took over as CEO and is responsible for the company's daily operations; Cook moved to executive chairman of the board and continues to participate in some matters, including global policy communication. This handover was long planned and approved by the board.
The value of this arrangement lies in clarifying management responsibilities and the succession path in advance. Even if the top management changes, the company's daily operations, product development and external communication have corresponding responsible persons and institutional arrangements to carry them on.
By contrast, Nvidia has long lacked a COO role and has not shown the outside world a clear mechanism for delegation of authority and succession. From supply chain procurement and customer allocation to software product roadmaps, Jensen Huang is deeply involved in all of it.
Nvidia does not lack experienced executives. What it truly lacks is a core manager other than Jensen Huang who can gain broad recognition from the board, the technical team and the capital market.
At present, Nvidia's board has launched an independent investigation. But the investigation itself can hardly eliminate the company's governance risks in the short term. It can answer whether specific events have affected the company, but it cannot replace a long-term effective management mechanism.
Nvidia also needs to prove to the market that even if Jensen Huang is temporarily unable to perform his duties, the company's customer management and core business can still operate normally.
II. Washington's fragile bargain
Nvidia's risks are not only internal. For a GPU company with business across the globe, Washington's attitude directly affects whether it can continue to expand in the global market.
Jensen Huang has always sought a balance between U.S. export controls and global commercial interests. In order to obtain export licenses for chips such as the H200 and B20, he tried to explain to the U.S. government that allowing American companies to participate in global compute market competition helps maintain the United States' technological and commercial advantage in the AI industry.
But this lobbying narrative is facing a more complex political environment.
The U.S. midterm elections concluded on November 3, with the Democratic Party winning control of both chambers of Congress. As the new Congress takes office in January next year, AI industry policy and regulation of tech giants will enter a new political cycle.
At that time, how will Washington view Nvidia?
Export controls are of course a key focus, and internal information security and compliance management within companies are equally unavoidable. For companies involved in advanced chip exports, customer backgrounds, access to technical information, and internal security management are already under strict scrutiny.
For Nvidia, there is an even more practical source of policy pressure: the U.S. government needs to strike a balance between restricting the flow of sensitive technology and maintaining the global competitiveness of domestic companies.
Export licenses, chip performance restrictions, and customer access rules can all affect what companies can sell, to whom they can sell it, and how much market space they can obtain.
An even more severe obstacle comes from grassroots voters.
Reuters' tally of 41 local referendums in 10 states shows that 40 of them voted in favor of restricting data center construction. Electricity prices, water resources, and environmental impact have turned compute infrastructure from an investment project for tech companies into an increasingly controversial issue in local public policy.
This presents Washington with another difficult problem. Jensen Huang stresses to the government the competitiveness of the U.S. AI industry, while local voters care about the costs that data center construction brings to their communities. On one side is technological competition at the national level; on the other is resource allocation at the local level. The government needs to respond to both demands at the same time.
Data centers require land, electricity, and water resources, and projects also need local government approval and community support to get off the ground. Even if the market has sufficient demand for compute, investment plans may still be delayed if energy supply, land planning, and local permits cannot keep up.
Jensen Huang's personal influence can help Nvidia open a window for policy communication, but it cannot solve these structural problems for the company.
III. An increasingly heavy business
In the early days of the AI race, Nvidia's business model was relatively straightforward: sell GPUs to cloud service providers, let customers build data centers, and then use compute to provide cloud services or train models.
As demand for compute grew rapidly, chip sales alone could no longer fully explain the investment scale of the entire industry. Data center construction also requires land, electricity, cooling systems, networking equipment, and long-term funding, and a shortage of any one resource can slow down compute deployment.
Nvidia CFO Colette Kress disclosed at the company's earnings call in August this year that capital expenditure by the world's top five cloud service providers is expected to approach $800 billion in 2026 and may reach $1.3 trillion in 2027.
On August 10 this year, Nvidia announced cooperation with financial institutions including Apollo, BlackRock, Blackstone, and Goldman, planning to mobilize more than $500 billion in third-party capital for AI infrastructure construction.
This cooperation connects Nvidia with financial institutions, data center developers, and parties demanding compute. What Jensen Huang hopes to promote is making compute infrastructure a productive asset capable of attracting long-term capital.
In specific projects, financial institutions can participate in data center investment through project companies and other vehicles, and then lease the completed facilities to cloud service providers or AI companies. Nvidia participates by virtue of its chip supply, technical capabilities, and industry ecosystem, helping projects connect compute demand with infrastructure supply.
But problems follow: when the chip business extends into infrastructure investment, who bears the long-term risks of the projects?
First is demand risk. The construction cycle of data centers is measured in years, while GPUs iterate much faster. Whether the computer rooms built with huge investment today can continue to obtain enough compute orders in the future depends on the actual demand of cloud service providers and AI companies.
Second is asset returns. Investment in data center power, cooling, networking, and land has strong long-term characteristics, and equipment also needs continuous upgrading. If compute demand growth falls short of expectations, or if the efficiency and rental prices of old equipment decline after a new generation of chips is launched, project cash flow may be affected.
There is also the issue of customer concentration. Large data center projects often rely on a small number of cloud service providers or AI companies to provide long-term leasing demand. Once major customers cut capital expenditure or adjust procurement plans, project revenue and financing arrangements may both be hit.
IV. The credit of a giant
What Nvidia faces today is in fact a question that all growth-stage tech companies may encounter when they move toward an infrastructure role: when the scale of growth is large enough, by what standards will the market reassess a company's value?
In business history, Standard Oil, IBM, and Microsoft all grew from industry leaders into giants with infrastructure characteristics. The larger a company becomes and the more deeply its business penetrates the operation of society, the more complex the external demands on it become.
Technological innovation and growth speed remain important, but customer fulfillment, governance standards, regulatory compliance, and long-term operational stability will also enter the valuation framework of the capital market.
Nvidia's growth story has largely been built on expectations of continued explosive demand for AI compute. As long as cloud service providers keep expanding purchases and GPUs remain in short supply, the market is willing to pay a premium for its technological leadership and growth space.
But as hundreds of billions of dollars in capital begin to flow into data center construction, the market has more things to calculate. After GPUs are sold, can they be fully utilized? Can the pace of data center construction keep up with the commercialization of models and applications?
If revenue created by AI applications can cover compute procurement, electricity, and infrastructure costs, then massive capital expenditure has a reason for continued investment. Once the pace of commercialization cannot keep up with infrastructure expansion, companies across the industrial chain will all face pressure from declining investment returns.
The value of an infrastructure company depends both on whether its technological iteration can stay ahead and on whether it can maintain long-term, stable business relationships in different markets.
For Jensen Huang, this may be the biggest cognitive threshold since he founded the company thirty years ago.
He not only has to continue leading Nvidia to win the next chip process and architecture cycle, but also has to push the company to establish governance mechanisms, business rules, and an industrial cooperation system that can support long-term development.
Nvidia has reached a new stage. Growth cannot be driven only by compute demand and product iteration; it must also withstand the test of capital returns, the policy environment, and market cycles.
Jensen Huang can continue to be Nvidia's most important leader, but Nvidia must become a company that does not need him to answer every question.
A note from [Beyond the Page]:
In the business world, great founders are not rare.
What is truly scarce is that after the founder leaves, the company can still carry on his judgment and even surpass his abilities.
A person can create a company's ceiling. But only institutions determine a company's lifespan. Channel: International Hotspots. This content is published with the author's authorization. The views represent only the author and do not represent Huxiu's position. If you have objections or complaints about this manuscript, please contact tougao@huxiu.com. Everyone who is changing and wants to change the world is on the Huxiu APP [ ]