AI News

Jensen Huang Resigns as NVIDIA CEO, Adding Uncertainty to Long-Term Orders in the A-Share "NVIDIA Chain"

HuxiuUpdated Nov 14 15:31

AI summary

On November 12, NVIDIA announced that Jensen Huang has resigned as President, CEO, and Director, effective immediately; CFO Colette Kress has been appointed interim CEO, and Director Harvey Jones has been appointed to the newly created position of Chairman. The stock fell more than 14% intraday that day and closed down 4.34%. Products and orders for the next two years have already been scheduled; the Rubin Ultra and Feynman roadmaps have been set, most 2027 HBM contracts have been finalized, and customers are already discussing supply for 2029 and beyond. Work previously driven by Jensen Huang, including specifications, capacity reservations, and customer financing, will be handed over to his successor. In the short term, optical module, PCB, and liquid cooling companies in the A-share "NVIDIA chain" will follow cloud capital expenditure and platform cadence; the PCB value per rack rises from about $35,000 for GB300 to more than $110,000 for Vera Rubin.

Body · AI translation

[[**Jensen Huang Resigns as NVIDIA CEO, Leadership Change Completed Within Four Days; Orders Locked In for Two Years, but Specifications, Capacity and Customer Support After 2029 to Be Decided by Successor, Adding Uncertainty to Long-Term Growth of A-Share "NVIDIA Chain."**

**Key Points**

**1. Leadership Change Lands Ahead of Earnings, Market Falls Then Narrows Losses**

On November 12, it was announced that Jensen Huang would resign immediately, with CFO Kress serving as interim CEO and Jones as chairman; that day the stock fell more than 14% intraday and closed down 4.34%.

**2. Short-Term Orders and 2027 Supply Already Locked In**

Products and orders for the next two years have been scheduled, the Rubin Ultra and Feynman roadmaps are set, most HBM 2027 contracts have been finalized, and customers are already discussing supply after 2029.

**3. Long-Term Variables Center on Specifications, Capacity and Customer Financing**

These were originally driven personally by Jensen Huang; as of the second quarter of fiscal 2027, external guarantees totaled about $530 billion, and the $500 billion financing platform was also driven by him.

**4. A-Share Chain Follows Platform Transitions in the Short Term, Long-Term Depends on Upgrade Magnitude**

Optical modules, PCBs, liquid cooling and others still follow cloud capital expenditure and platform cadence in the short term; PCB value per rack rises from about $35,000 for GB300 to more than $110,000 for Vera Rubin.

**5. NVIDIA's China Share Declines, Domestic Substitution Constrained by Capacity and HBM**

In the first half of 2026, NVIDIA's share of China's data center AI chip market fell to 34%, with Hopper revenue to China accounting for less than 1% of data center business; domestic substitution remains constrained by process technology and HBM supply. 2026-11-14 12:03 Jensen Huang Is Gone, "Da Chain" Bosses Start Reviewing Orders Economic Observer Quick Read

This article is from the WeChat public account: Economic Observer, author: Zheng Chenye

In the past, what specifications each generation of NVIDIA products would use, how much capacity to book upstream, and how much financial support to provide downstream customers were mainly decided by Jensen Huang himself. Now that he has left, although NVIDIA's products and orders for the next two years have been scheduled, longer-term planning must be handed to his successor, and A-share listed companies in this industrial chain have therefore gained added uncertainty.

At 22:00 Beijing time on November 12, half an hour before the U.S. stock market opened, NVIDIA (NVDA.O) announced on its official website that founder Jensen Huang had resigned from the positions of president, chief executive officer and director, effective immediately. After the news was released, NVIDIA's stock fell more than 14% at one point after the market opened that day, and the decline narrowed to 4.34% at the close.

According to the announcement, NVIDIA Executive Vice President and Chief Financial Officer Colette Kress will serve as interim chief executive officer while continuing as chief financial officer; Harvey Jones, who has served on NVIDIA's board since 1993, will serve as chairman, a position NVIDIA had never established before.

The board also launched a search for a permanent chief executive officer, considering both internal and external candidates. The announcement did not explain the reason for Jensen Huang's resignation, and no statement from him appeared anywhere in the text.

On November 8, Jensen Huang wrote to employees regarding a personal emotional controversy, and NVIDIA's board announced that day that an investigation led by independent directors would be launched, while also expressing confidence in Jensen Huang and the management team he led. But just four days later, the company saw its first leadership change since its founding 33 years ago.

Public information shows that Jensen Huang is 63 years old this year and has been the only chief executive officer of the company since he founded NVIDIA in 1993. During his tenure, NVIDIA grew from a graphics chip company into a core global supplier of AI computing power, and in October 2025 became the world's first listed company to surpass a market value of $5 trillion.

More than two months earlier, on September 2, Jensen Huang said in a conversation with U.S. Commerce Secretary Howard Lutnick: "Of course I could stop working, but I don't want to miss it. I would never miss the next 10 or 15 years."

In the global AI computing power supply chain, from TSMC's advanced processes and South Korean manufacturers' high-bandwidth memory chips (HBM) to the optical modules, printed circuit boards (PCBs) and liquid cooling equipment produced by A-share listed companies, most upstream capacity expansion has followed NVIDIA's one-generation-per-year product cadence.

In the past, what specifications each generation of NVIDIA products would use, how much capacity to reserve from upstream suppliers, and how much financial support to provide to downstream customers were mainly decided by Jensen Huang himself. Now that he has left, although NVIDIA's products and orders for the next two years have been scheduled, longer-term planning must be handed to his successor, and A-share listed companies in this industrial chain have therefore gained added uncertainty.

Leadership change in four days

Public information shows that, unlike many U.S. technology companies where founders hold super-voting rights, NVIDIA implements one share, one vote, and Jensen Huang's shareholding is approximately 3.58%. Before Jensen Huang's resignation, among the 11 board members, the 10 other than him were all independent directors. Regarding whether Jensen Huang should stay or go, the board's attitude appears to have changed within four days.

On November 8, NVIDIA's board said in a statement that the investigation would be handled by a committee composed of independent directors, assisted by outside independent counsel, and that Jensen Huang supported the investigation and promised to cooperate. The statement concluded by saying that the board had confidence in Jensen Huang and the team he led.

Two days later, on November 10, NVIDIA's independent directors issued a separate statement. The statement contained only one sentence: the independent directors had engaged an outside law firm to conduct a thorough and independent investigation. It no longer mentioned confidence in Jensen Huang.

According to documents NVIDIA filed with U.S. securities regulators, on November 11 the board accepted Jensen Huang's resignation and that same day appointed 73-year-old Jones as chairman. Before the U.S. stock market opened on November 12, the outcome was announced to the public.

A veteran who has long followed the semiconductor industry told Economic Observer reporter that NVIDIA had originally planned to release its financial results and provide guidance for the new quarter on November 17, and that it still had a large number of long-term agreements with customers and suppliers to advance. If the chief executive's future remained unresolved, all of these matters would be affected, and the board needed to deliver a clear outcome before the earnings release.

Previously, NVIDIA did not have a chairman, and the board was chaired by the lead independent director. According to the announcement, Jones, who became chairman, is a former chief executive of Synopsys. He said in the announcement that the board has full confidence in Cress and the management team, and that NVIDIA's day-to-day operations, commitments to customers and product roadmap will not be interrupted.

Public information shows that Cress, who became interim chief executive, is 59 years old and joined NVIDIA in September 2013, serving as chief financial officer ever since. When she joined, NVIDIA's market value was still less than $10 billion. Over the past 13 years, NVIDIA's quarterly results and guidance have basically all been explained to investors by her on earnings calls.

In a letter to employees on November 12, Cress said that the past period had been extremely difficult for employees, and that management had scheduled an all-hands meeting for November 18. Before that, on November 17, NVIDIA will release its financial results for the third quarter of fiscal 2027 as originally planned, with previously provided revenue guidance of about $108 billion. At that time, Cress will face investors for the first time as interim chief executive.

Most of NVIDIA's key decisions over the past 33 years came from Jensen Huang. In 2006, NVIDIA launched the software platform CUDA, enabling graphics processing units (GPUs) to handle general-purpose computing tasks. This investment did not generate corresponding revenue for many years afterward, and only showed its value when the deep learning era emerged in 2012. Today it is one of NVIDIA's most important competitive moats.

NVIDIA's financial reports show that November 2022, when ChatGPT was released, fell in NVIDIA's fiscal 2023, in which NVIDIA's full-year revenue was about $27 billion; by the second quarter of fiscal 2027, ended July 2026, NVIDIA's quarterly revenue had reached $96.2 billion, of which $89 billion came from the data center business. During this period, NVIDIA's product cadence and sales methods were all determined by Jensen Huang himself.

In terms of product cadence, in June 2024, Jensen Huang announced during Computex Taipei that the AI chip iteration cadence would be accelerated from one generation every two years to one generation every year.

After that, the Blackwell platform was launched in 2024, and the Vera Rubin platform began mass production and shipment in August 2026. NVIDIA's previously published roadmap shows that the Rubin Ultra and Feynman platforms are planned to be launched in 2027 and 2028, respectively.

In terms of sales methods, starting with the Blackwell generation, NVIDIA's main products expanded from a single chip to a complete rack, with one rack integrating 72 GPUs as well as supporting central processing units (CPUs), switches, liquid cooling and power supply systems. In a research report published in May 2026, Morgan Stanley estimated that the bill of materials cost for one Vera Rubin rack is about $7.8 million, about 95% higher than the previous-generation rack.

Colette Kress also provided a set of figures on the earnings call in August 2026: for every 1 gigawatt (GW) of data center built, NVIDIA's corresponding potential revenue rises from about $18 billion for the Hopper generation launched in 2022 to about $25 billion for the Blackwell generation and about $40 billion for the Vera Rubin generation.

What Kress and the future official chief executive are taking over is precisely this set of product cadence and sales methods established by Jensen Huang.

NVIDIA Without Jensen Huang

NVIDIA's products and orders for the next two years had already been determined before Jensen Huang's departure, so a change of leadership will have limited impact on this part of the business.

On the earnings call in August 2026, Kress said NVIDIA's revenue for fiscal 2028 (ending January 2028) is expected to grow about 70% year on year, and at least until the end of that fiscal year, products will remain in short supply. According to her, capital expenditure by the world's top five cloud providers is approaching $800 billion in 2026 and is expected to reach $1.3 trillion in 2027.

A technology analyst at a large securities firm in South China told Economic Observer that the platforms and architectures of the Rubin Ultra and Feynman products have both been determined, and most of NVIDIA's supply for 2027 and 2028 has already been booked by customers. What customers are currently discussing with NVIDIA is already supply for 2029 and beyond. The impact of the leadership change will therefore only become apparent later, and will be concentrated in several areas of work that were previously the responsibility of Jensen Huang himself.

On the product side, TF International Securities analyst Ming-Chi Kuo recently posted on social media that Jensen Huang may not be the person inside NVIDIA who best understands any single new technology such as high-voltage direct current power supply or co-packaged optics (CPO); his value lies in his ability to turn technologies that the engineering team believes are feasible into products that the whole company is willing to commit resources to, take risks on, and must deliver.

Kuo believes that this has also given the supply chain high order visibility, making suppliers willing to cooperate with NVIDIA's upgrades and capacity expansion. In his view, the impact of a change of CEO on NVIDIA's product development will not become apparent until 2028 at the earliest, and can be observed from NVIDIA's subsequent capacity bookings and the scale of specification upgrades.

Beyond products, upstream capacity has also long been secured personally by Jensen Huang.

In a report recently released by semiconductor research firm SemiAnalysis, Jensen Huang's earlier trip to South Korea was precisely to ensure the supply of memory chips; NVIDIA's supply and capacity commitments to suppliers increased from $119 billion to $279 billion within a quarter, most of which was for memory chips. Most of these agreements were negotiated personally by Jensen Huang.

"HBM supply contracts are usually finalized about a year in advance; most of the volume and prices for 2027 have already been set and will not change in the short term because of a change of CEO," Lin Meibing, chief analyst at chip industry outlet ICTIME, told the Economic Observer.

According to him, NVIDIA has long been a priority customer for South Korean memory manufacturers, which is related to the communication Jensen Huang has maintained with senior executives of those manufacturers over the years. As is customary, negotiations on 2028 volumes will begin in 2027, and only then will it be clear whether his successor can sustain such relationships.

Support for downstream customers is another task for which Jensen Huang was responsible. According to the SemiAnalysis report, as of the second quarter of fiscal 2027, NVIDIA's total various guarantees provided externally amounted to about $530 billion, up from $184 billion in the previous fiscal quarter; many computing power leasing companies have been able to obtain financing at lower cost precisely by relying on NVIDIA's guarantees.

Public information shows that on August 10, 2026, NVIDIA also announced that it would join with multiple financial institutions to build a financing platform of about $500 billion to provide financing for customers purchasing NVIDIA chips and to guarantee the residual value of some chips, with the guarantee ratio at up to 25% of a single transaction.

A veteran figure in the semiconductor industry said that this financing platform was driven by Jensen Huang personally, and that even if the new management continues support for downstream customers, there may be a pause in between, and competitors such as Broadcom and AMD may gain more room in the next round of order competition. "NVIDIA's organizational structure is built around Jensen Huang alone," the person said. Most heads of NVIDIA's various businesses report only to Jensen Huang, and coordination among businesses is also completed by him; whoever takes over from within will need time to become familiar with businesses they were not originally responsible for.

SemiAnalysis also mentioned in the aforementioned report that Jensen Huang has publicly said he has about 60 direct subordinates. The two names most frequently mentioned inside NVIDIA are Vice President Ian Buck, who led the development of CUDA, and Senior Vice President Jonah Alben, who is responsible for GPU design. The two are respectively strong in software and chip design, but may not necessarily possess Jensen Huang's control over the supply chain.

The institution believes that a founder with the full support of the board can bear the risks of long-term investment, while a successor, no matter how long they have served at NVIDIA, must be accountable to the board and shareholders in a different way, and the likelihood of making similar decisions will be significantly lower.

In other words, customers are already discussing supply for 2029 and beyond with NVIDIA, and the specifications and production capacity of these products will need to be decided by the successor.

Variables in China's industrial chain

While Jensen Huang was in office, his dinners were always a topic eagerly discussed by the market and also a window for outsiders to observe NVIDIA's supply chain.

Public information shows that in October 2025, Jensen Huang dined in Seoul with Samsung Electronics Chairman Lee Jae-yong and others; on the evening of January 31, 2026, he hosted a dinner at a restaurant in Taipei for nearly 40 heads of supply chain companies. TSMC Chairman C.C. Wei, Quanta Chairman Barry Lam, Hon Hai Chairman Young Liu and others were among the attendees, including heads of PCB companies from mainland China. This dinner was described by outsiders as a "trillion-dollar banquet."

Every time a group photo circulated, the market would identify one by one the companies to which the attendees belonged and interpret it as a list of NVIDIA's supply chain. Jensen Huang himself also joked on the earnings call in August 2026 that one of the most interesting things in the past year was that outsiders were always wondering where he went to dinner and with whom.

The aforementioned technology analyst at the brokerage said that a group photo from a dinner being interpreted one by one by the market shows from the side Jensen Huang's control and influence over the supply chain: which companies can enter NVIDIA's supply system and how much share they obtain depends to a large extent on him personally.

In the A-share market, the companies known as the "NVIDIA chain" are mainly concentrated in segments such as optical modules, PCBs, liquid cooling, and server assembly. The revenue these companies derive from NVIDIA's industry chain depends on cabinet shipment volumes and the value of the components used per cabinet; the former follows cloud providers' capital expenditure, while the latter is determined by the specifications of each generation of NVIDIA's platform.

During interviews, Economic Observer reporters learned that domestic supply chain companies generally believe that the change of CEO will not affect cabinet shipments or current orders in the short term; the variable lies in whether NVIDIA's subsequent platform upgrades can maintain their original pace and magnitude.

The marketing head of an optical module company in East China told Economic Observer that orders signed between optical module companies and leading overseas customers fall into two categories: one is long-term orders that cannot be modified or canceled, and the other is annual framework orders that can be fine-tuned monthly. The former accounted for 30% to 40% in 2025 and is expected to exceed 50% in 2027.

According to this person, in 2027, industry-wide demand for 1.6T optical modules will exceed 100 million units, but constrained by the supply of core chips such as digital signal processors (DSPs), effective supply will be only about 70 million units; with supply falling short of demand, customers have no incentive to cut orders.

The marketing head of a large server assembly company in South China also told Economic Observer that system integrators' production schedules follow cloud providers' capital expenditure plans and the pace of NVIDIA's platform transitions; replacing the chief executive itself will not change these two factors. The industry is more concerned about whether subsequent generations of platforms can be launched on the original schedule and with the original specifications.

This person believes that the specification improvements in NVIDIA's last two generations of platforms have already significantly raised the value per cabinet. According to industry estimates, from the GB300 cabinet of the previous-generation Blackwell platform to the Vera Rubin cabinet, the value of PCBs used per cabinet rises from about $35,000 to more than $110,000; one reason is that the new platform uses PCB midplanes to replace some of the cables inside the cabinet.

Vera Rubin is also NVIDIA's first fully liquid-cooled platform, extending heat dissipation from the GPU to all heat-generating components including memory, network cards, and power supplies, thereby raising the liquid cooling value per cabinet to about $120,000.

The next round of upgrades is already on the roadmap. According to the source, based on information currently available in the supply chain, the new-generation cabinets equipped with Rubin Ultra are planned to enter mass production in the second half of 2027, with power consumption per chip rising from 2,300 watts for Vera Rubin to 3,600 watts, the copper cables inside the cabinet replaced by a 78-layer PCB backplane, the supply voltage raised from 54 volts to 800 volts, and power consumption per cabinet rising to about 600 kilowatts. The high-end capacity investment plans already announced by leading domestic PCB companies total more than 40 billion yuan.

In his view, domestic suppliers' capacity expansions are all built on the expectation that NVIDIA will continue to upgrade specifications. If the successor becomes more conservative on specification upgrades and capacity bookings, growth for upstream companies after 2028 will slow, which is more worth watching than short-term order fluctuations.

Beyond specifications, supplier admission and key materials are also controlled by NVIDIA.

The market head of the aforementioned server assembly company said that when purchasing components such as liquid cooling, system integrators can only choose from the supplier list certified by NVIDIA, and this list is constantly adjusted as platforms iterate. At present, manufacturers from Taiwan, China, and from Europe and the United States still dominate core components such as cold plates and manifolds, while mainland Chinese manufacturers have only begun entering this list in areas such as coolant distribution units and quick connectors over the past year.

The market head of the aforementioned optical module company also pointed out that core materials such as laser chips used in optical modules supplied to NVIDIA need to pass NVIDIA's certification, and when materials are in short supply, NVIDIA often steps in to coordinate with upstream manufacturers.

According to the source, as optical modules evolve from pluggable solutions to new solutions such as near-packaged optics (NPO) and CPO, when and on what scale they are introduced also depends on the platform plans of leading customers such as NVIDIA. According to current industry expectations, NPO solutions will begin ramping up in the second half of 2027.

Beyond supply relationships, changes at NVIDIA will also affect more Chinese companies through memory chips.

Lin Meibing said that NVIDIA is the largest buyer of HBM (high-bandwidth memory), and memory manufacturers are shifting large amounts of capacity to HBM, which is one reason for the tight supply and rising prices of general-purpose memory chips.

In his view, if NVIDIA's pace of HBM procurement changes in the future, the supply and prices of general-purpose memory chips will change accordingly, and this will be transmitted to domestic module manufacturers and terminal manufacturers such as mobile phone makers.

Public information shows that Jensen Huang had previously publicly advocated on multiple occasions that NVIDIA continue selling products to the Chinese market. After his departure, several brokerages believe that the United States is expected to enforce export controls on advanced chips more strictly, and domestic AI chips are ushering in a window for shipments.

However, NVIDIA's share of the Chinese market had already declined significantly. According to Colette Kress at the earnings call in August 2026, in the second quarter of fiscal 2027, NVIDIA's revenue from sales of Hopper-series chips to Chinese customers accounted for less than 1% of data center business revenue, and subsequent guidance also did not include data center computing revenue from China.

According to IDC statistics, in the first half of 2026, in China's data center AI chip market, NVIDIA's share had fallen to 34%, Huawei Ascend's share had risen to 28%, and the shares of Alibaba Pingtouge, Cambricon, Baidu Kunlunxin, and Hygon Information were each between 3% and 7%.

Lin Meibing believes that domestic computing power chips are currently mainly used for inference, and shipments are still constrained by advanced process capacity and HBM supply. The market space vacated by NVIDIA has long existed, and Jensen Huang's departure has not changed the current supply-demand state. How much of that share domestic computing power chip companies can take depends on their own capacity and supply chains.

Capacity expansion across the upstream and downstream of the industrial chain is, to a certain extent, built on the judgment of Jensen Huang alone. After his departure, whether NVIDIA can continue its previous pace of advance remains to be tested by time. Channel: Finance and Economics. This content comes from the internet, original link. The views represent only the author and do not represent the position of Huxiu. If there is a copyright issue, please contact hezuo@huxiu.com, and we will verify and handle it promptly. Everyone who is changing and wants to change the world is on the Huxiu APP[]

Source

Huxiu