AI News

Jensen Huang Resigns as NVIDIA CEO; AMD Shares Rise 44.7% Cumulatively Over Five Consecutive Days

36Kr · AIUpdated Nov 14 19:00

AI summary

On November 12, Jensen Huang announced his resignation as NVIDIA's President, CEO, and board director. NVIDIA's stock fell as much as 14.4% intraday, with market capitalization briefly dropping by about $600 billion. On the same day, AMD closed up 5.9%, and from November 5 to 12 it rose 44.7% cumulatively. The article compares the two companies' data center businesses: NVIDIA's total second-quarter revenue was about $96.2 billion, with data centers contributing $89 billion, while AMD's data center revenue was $6.7 billion. It also mentions that AMD's Helios racks have secured a $1.2 billion order signed by HPE and Vultr. The author believes that the biggest obstacle to AMD winning orders remains the CUDA ecosystem, and that ROCm can only reduce the workload of code changes.

Why read this

SelectedAI score 76

Using Jensen Huang's departure as the point of departure, the article examines AMD's market share, orders, and CUDA ecosystem gap across both data center and desktop GPU lines.

Topics

Tags

Body · AI translation

On November 12, Jensen Huang publicly announced his resignation as president, chief executive officer, and director. Nvidia's stock fell as much as 14.4% intraday, with market value briefly dropping by about $600 billion, before the closing decline narrowed to 4.3%.

Meanwhile, AMD, Nvidia's biggest rival, closed up 5.9% that day. This was already AMD's fifth consecutive trading day of gains, rising a cumulative 44.7% from the November 5 close to the November 12 close.

This reminds me of the Polymarket betting pool on who would succeed Jensen Huang. In a screenshot posted by the official account on November 12, Huang's daughter Madison Huang ranked first at 12%, while AMD CEO Su and Nvidia CFO Colette Kress were tied for second at 7% each. AMD really is all too practiced at being the "runner-up"!

In the PC era, the CPU was the core hardware, and AMD was always chasing Intel; in the AI era, cloud providers buy GPUs in batches to build data centers, and AMD is still the runner-up—only the leader has changed to Nvidia.

Because Nvidia will not release its third-quarter earnings until November 17 US Eastern Time, let's first look at the second-quarter earnings data.

According to Nvidia's second-quarter earnings this year, Nvidia's total revenue was about $96.2 billion, of which the data center business contributed $89 billion, accounting for more than 90%. In AMD's earnings, by contrast, AMD's data center business was only $6.7 billion, and that includes some server CPU revenue.

AMD packaged its server CPUs, AI accelerator cards, and high-speed networking into a rack system called Helios; the corresponding Nvidia rack-scale product is called Vera Rubin NVL72.

On the left is AMD Helios, on the right is NVIDIA Vera Rubin NVL72.

On September 30, enterprise IT equipment vendor Hewlett Packard Enterprise (HPE) disclosed that it had signed a $1.2 billion Helios rack order with US cloud service provider Vultr. AMD can already participate in billion-dollar-level cloud provider procurement.

The Sabey SDC Columbia data center campus in East Wenatchee, Washington, USA. Vultr once expanded GPU compute capacity in the Seattle region here.

But for Nvidia, an order of this scale can only be considered a "small order."

On November 2 last year, cloud compute company IREN signed a procurement agreement with Dell worth about $5.8 billion for Nvidia GPUs and supporting products and services. In this year's second-quarter earnings, a certain "patron" contributed 16% of Nvidia's total revenue, equivalent to about $15.4 billion, more than twice AMD's second-quarter data center revenue.

To put it exaggeratedly, a single hair plucked from Nvidia would be thicker than AMD's waist.

However, now that Jensen Huang has resigned from his positions at Nvidia, CFO Colette M. Kress is serving as interim CEO. After the leadership change, some customers may reconsider their next round of procurement.

Colette M. Kress, Nvidia's interim chief executive officer and chief financial officer.

Just nine days before Nvidia announced Jensen Huang's departure, AMD had just released its third-quarter earnings. Data center revenue rose to $8.546 billion, up 27.2% quarter over quarter and up 97% year over year, mainly driven by demand for EPYC server CPUs and the launch of the Instinct MI450 series GPUs. On the call that day, Su said this was the first quarter in which AMD shipped MI455 and Helios racks and recognized revenue.

Although the earnings report did not separately disclose Helios revenue, it is clear that rack-scale products have begun contributing revenue to AMD.

Nvidia is about to release its third-quarter data on November 17 US Eastern Time, with the statistical period ending October 25. At that point, Jensen Huang was still in office, so the data should not have been affected. What is more worth watching is the fourth-quarter guidance given by management.

For AMD, this is like buying its own equipment and waiting for the jungle monsters to respawn, when suddenly word comes from the top lane that the opponent has been solo-killed. To win this game, now is a good opportunity, but whether it can "snowball" and push straight through—there are still quite a few difficulties waiting.

02 Desktop graphics cards: truly no match at all

AMD was founded in Silicon Valley in 1969, the same generation as Intel and even earlier than Nvidia. It made CPUs, and in 2006 it bought graphics card company ATI, bringing Radeon into AMD. Since then, the company has been fighting Intel for CPU business while fighting Nvidia for GPU business.

At the beginning, AMD could still go toe to toe with Nvidia, but now, in desktop graphics cards, AMD is completely outmatched by Nvidia.

According to Jon Peddie Research data, in the second quarter of 2026, global desktop discrete GPU shipments were about 12.5 million units, with Nvidia accounting for about 90%, AMD about 8%, and Intel about 2%.

The first obvious gap opened in 2014 to 2015. Nvidia launched the more power-efficient GTX 970 and 980 first, while AMD did not launch the Radeon 300 series and Fury X until June of the following year. By the time AMD's new cards hit the market, its quarterly shipment share had already fallen to 18%.

In the second quarter of 2018, AMD once returned to 36.1%, then fell back to 18.8% by the end of the year. That year, mining demand receded, and both companies ran into channel inventory problems. Nvidia launched the RTX 20 series, focused on real-time ray tracing, in the second half of the year; AMD did not fight back into mainstream price points until 2019 with the RX 5700 series, and its share returned to 31.1% by the end of that year. The later RX 9070 XT could also achieve higher frame rates on a similar budget, yet by the second quarter of 2026, AMD's desktop discrete GPU shipment share was still only about 8%.

Today, Nvidia's RTX 5090 still leads the flagship tier, and AMD can only compete with Nvidia cards at the same level on price-performance. TechSpot tested 52 games in September 2026: the RX 9070 XT, priced at about $730 to $750, had an average frame rate at 1440p 24% higher than the RTX 5070, which started at $800 at the time.

On the left is the Radeon RX 9070 XT, and on the right is the GeForce RTX 5070.

Much of the "crushing" in desktop graphics cards is thanks to Jensen Huang.

In November 2006, Nvidia launched the CUDA-capable GeForce 8800. Huang Renxun later recalled that putting CUDA into gaming cards would eat into a lot of profit, but he insisted on doing it anyway. By May 2008, cumulative shipments of CUDA-capable GeForce 8 and 9 series GPUs had exceeded 70 million units.

What is CUDA?

CUDA is a set of software tools made by Nvidia. Engineers use it to hand tasks such as training models to the GPU, and they can also directly call code that Nvidia has already written and debugged. For example, when dozens of cards train together, there are ready-made tools for how the cards exchange data with one another.

CUDA parallel computing schematic. The same piece of code can be handed to multiple GPU threads to process different data at the same time.

People who buy gaming cards may not use CUDA, but developers have hardware they can buy at any time to write programs and run experiments.

Although Huang Renxun has left, the CUDA that developers are used to is still there. It is still relatively "easy" for AMD to win over gamers, but if it wants to win over people who buy graphics cards for development, it has to get past the CUDA hurdle.

03 Can CUDA's moat really be crossed?

The biggest difficulty for AMD in trying to win Nvidia's customers is CUDA.

This set of things was built up bit by bit by Nvidia. "The Nvidia Way: Jensen Huang and His Tech Empire" wrote that Nvidia not only put CUDA into gaming cards that developers could afford, but also sent people to help developers move molecular simulation programs onto GPUs.

Nvidia disclosed that by fiscal year 2026, more than 7.5 million developers were using CUDA and its other software tools, with hundreds of accompanying dedicated software libraries.

Of course, AMD is not without countermeasures.

ROCm is AMD's software tool benchmarked against CUDA; it lets developers change less code when switching cards. For example, if you train a photo classification model with PyTorch on your own computer, as long as all the features used are supported, installing the ROCm version of PyTorch will let the original training script continue to run on Radeon.

In AMD's official tutorial, the Radeon RX 9000 series local AI demo shows ComfyUI running SDXL image generation.

But what is really hard to catch up with in CUDA is the ready-made tools and code. The more features a project depends on, the more items have to be checked one by one when switching to AMD.

And for data centers, switching from Nvidia to AMD is even more complicated: a single card can run, but it still has to be verified whether dozens of cards training together work properly. This is not just a matter of money; time is actually more important.

Midjourney ran into a similar problem. In May this year, Midjourney founder David Holz said in an interview that Midjourney chose Google TPU early on; if it had used Nvidia's full set of tools from day one, the team's research progress probably could have been a year earlier. If he could do it over, he would use Nvidia GPUs from day one. Different technical approaches really do affect the pace of business.

A file photo of Midjourney founder David Holz.

When talking about catching up with Nvidia graphics cards, Liang Wenfeng said, "Technology has no secrets, but resetting takes time and cost."

AMD's approach is to minimize the amount of code customers need to change. For example, if a customer writes a model in PyTorch, as long as the features used are already supported on AMD, most of the model code can stay the same, and PyTorch hands the computation to AMD's GPU through ROCm. When it comes to CUDA code written by customers themselves, AMD also has tools that can help convert part of it.

Anthropic said in its IPO roadshow that it used Claude Mythos Preview to help modify computational code and got a model's input processing running on AMD MI355X in about three days, with speed reaching a usable level.

Meta and AMD adapted 10 recommendation models to both companies' GPUs; training with the same number of cards, each side won some and lost some, and overall training speed was about the same.

Cloud providers have already started using both companies together, no longer putting all their eggs in one basket.

On October 26 of this year, Nvidia announced an expansion of its model collaboration with Meta, and the next day, AMD disclosed that Meta was already using Helios. For procurement departments, this means the same round of capacity expansion can be split by task. Training that relies on CUDA continues to use Nvidia, while recommendation model training already validated on ROCm can separately procure AMD compute.

Of course, losing Jensen Huang as the face of the company will inevitably affect Nvidia's business—mainly because Jensen really can win people over with his name alone.

In 2025, Nvidia signed an agreement with CoreWeave initially valued at $6.3 billion, under which it would purchase compute capacity CoreWeave had not rented out, with the term running through 2032. Jensen Huang's face represents Nvidia.

After he leaves, whose face would dare make commitments on Nvidia's behalf?

04 AMD nearly died, and it was even less willing to settle for second place

In 2006, AMD bought ATI, and from then on it fought Intel on CPUs while fighting Nvidia on graphics cards.

But over the next decade, it lost ground on both fronts. By 2015, AMD's global x86 shipment share had fallen from 25.3% at the end of 2006 to 12.1%, and its desktop discrete GPU share at one point was down to just 18%.

By the end of 2015, AMD had only $785 million in cash and marketable securities, against $2.26 billion in debt. Revenue in the division selling PC processors and graphics cards fell 42% that year.

At the time, AMD pitted the FX-8370 against Intel's Skylake Core i7-6700K on desktop. In a contemporaneous test covering 14 applications, the FX-8370's overall score was about 39% lower.

With CPU share down to just over 10%, AMD first had to survive until its next-generation products arrived.

Game console orders and cash from selling equity bought AMD time. Lisa Su later recalled that AMD first concentrated resources on redoing its CPU cores, then expanded investment in GPUs. A new set of CPU cores could help it win back the desktop business and also return to the server market.

That new design was Zen. At the same frequency, one core could do more work.

The Zen series brought AMD back from the dead. AMD used it to build desktop Ryzen, winning back the desktop business, and used it to build server EPYC, re-entering the server market. By the second quarter of 2026, looking only at EPYC and Xeon in servers, AMD's shipment share had caught up to 46.4%, just 7.2 percentage points behind Intel's 53.6%.

In 2017, Lisa Su showed off the first-generation Ryzen processor.

The GPU line was not abandoned.

At the time, Sony's PS4 and Microsoft's Xbox One both used AMD custom chips, with a CPU and graphics processor in a single chip. The technology accumulated from making graphics cards could be sold both to PC gamers and to console makers.

The first-generation PlayStation 4 and DualShock 4 controller unveiled at E3 2013.

In December 2016, the day before AMD showed off its Zen desktop processors, it announced the Radeon Instinct plan for deep learning.

While saving its CPUs, AMD was already making GPUs for data centers to train AI models.

That courage is finally starting to pay off. AMD launched the Instinct MI100 for data center computing in 2020, and in 2023 took the MI300X to compete for large-model compute. Today, the rescued EPYC and the Instinct line developed along the way have become the core of the Helios rack.

The cover of AMD's official "Building Helios" engineering video, showing the development site.

Oracle Cloud is a good example, because it both buys racks and rents compute to customers.

Oracle plans to use 50,000 AMD Instinct MI450 series GPUs, with deployment starting in the third quarter of 2026, to build an AI cluster open to cloud customers, and to continue expanding deployment in 2027 and beyond. At 72 cards per Helios rack, that is nearly 700 fully configured racks at the start.

Take Vultr as a reference. Its planned previous-generation AMD AI cluster in Ohio is set to use about 24,000 MI355X cards, with the entire cluster project planned to invest more than $1 billion. Oracle's first batch of deployment alone plans to use more than twice as many cards as that project.

With Jensen Huang gone, the sky has brightened for AMD.

Desktop graphics cards can finally breathe a little, and although developers are still used to CUDA, winning over some gamers is still worthwhile.

Data centers are where AMD can seize orders. A full Helios rack is packed with Lisa Su's hopes, ROCm can help bypass CUDA's moat, and customers have even signed contracts and are just waiting for shipments.

But I guess AMD won't laugh out loud; they'll only be unable to hold it in when Lip-Bu Tan also leaves Intel.

This article is from the WeChat public account "Hardcore Kanban" (ID: yinghekb), author: the one embracing AI, published by 36Kr with authorization.

Source

36Kr · AI