In the red-hot market for artificial intelligence chips, Advanced Micro Devices finds itself in a familiar position: playing No. 2 to the market leader, Nvidia.
AMD took on that role in personal computer and traditional server chips, battling it out for years with industry giant Intel in one of tech's most heated rivalries. And AMD stock has fared just fine. Now, the explosive growth of AI has set the stage for AMD vs. Nvidia, yet another tech showdown, this time pitting the dominant force in AI chips against a challenger with a track record of transforming a runner-up role into a position of strength.
The rivalry takes center stage at Nvidia's big AI event March 18 in San Jose, Calif. The company is expected to unveil its next-generation product, the B100, its first Blackwell series GPU, Monday at Nvidia's GTC 2024 conference for AI developers. Nvidia is holding its AI event as AMD ramps up production of its MI300 AI chip.
"Historically AMD has been a No. 2 in pretty much every field that they've competed in. And a compelling one," Piper Sandler analyst Harsh Kumar told Investor's Business Daily.
AMD Vs. Nvidia In AI Chips
AMD's prowess as a tough, resilient competitor is underscored by its longtime rivalry with Intel. AMD was on the verge of collapse, its stock price worth less than $2, when Chief Executive Lisa Su took over a decade ago.
But AMD battled back to regain market share in PCs and servers, even though it has never come close to overtaking Intel. Intel had 72.4% of total PC chip shipments in the fourth quarter of 2023, followed by AMD with 18.3% and Apple with 7.8%, according to IDC. Intel also is dominant in server chips with 68.7% unit market share in Q3 2023, outpacing AMD's 20.5% share.
But AMD has drawn strength from an important fact: Manufacturers dislike being locked into one vendor. Having more than one choice is always preferred.
AMD stock fell 4% to 187.06 on Thursday as investors took profits following the steep run-up in AI stocks. Nvidia stock retreated 3.2% to 879.44 on Thursday.
Nvidia currently dominates the market for graphics processing units, or GPUs, used for running computationally intensive AI workloads. But AMD has proven to be an able fast-follower.
AMD's Instinct MI300 series accelerators provide a viable alternative to Nvidia's current H100 GPU, analysts say.
'Hyperscalers Want Choice'
AI chip customers, such as cloud service providers and enterprises, appreciate having a strong No. 2 in the market to keep prices down and spur innovation.
"The hyperscalers want choice and any market that is dominated by a single supplier makes them very uncomfortable, and it should," Patrick Moorhead, chief analyst with Moor Insights & Strategy, told IBD.
Since AMD stepped up its presence in the AI accelerator market late last year, Nvidia has moved to coming out with a new flagship product each year to match AMD's annual cadence. Nvidia previously was on a two-year cycle for its product launches in the category.
Nvidia is widely expected to detail the capabilities of its next-generation product, the B100, on Monday at GTC. The B100 will be available later this year. And in 2025, Nvidia plans to leapfrog that product with its X100 GPU.
AI Chips Backlog
For now, Nvidia is selling every H100 GPU it can make and has a sizable backlog. But its inability to meet demand for AI chips created an opening for AMD and others to gain market share. Other players in the market include Intel with its Gaudi AI processors and custom application-specific integrated circuits, or ASICs, from Broadcom and Marvell Technology. Plus, a host of AI chip startups are pursuing the market, including Cerebras and SambaNova Systems.
By all accounts, the market for AI chips in data centers is massive and can support numerous vendors.
Ahead of the B100 launch, Nvidia plans to release an upgraded version of its current Hopper series product H100. That improved product, the H200, is due for release in the second quarter with performance boosts across the board vs. the H100.
Meanwhile, AMD has a new data center GPU, the MI400, in its near-term product pipeline.
AMD Vs. Nvidia AI Chips: A Viable Alternative
Wall Street analysts see no slowdown in spending for data center infrastructure to support generative AI applications. Startup OpenAI kicked off the generative AI trend with its buzzy chatbot, ChatGPT, in late 2022.
Even with the burgeoning competition in the space, Nvidia has more than 90% of the GPU market, analysts say. That's largely because Nvidia provides a full-stack solution for AI computing, including its CUDA software and foundational models and libraries to jump-start development of AI applications.
AMD and Intel, by contrast, are focused on providing raw computational horsepower with their GPUs.
"I do see AMD as a very viable alternative to Nvidia," Piper Sandler's Kumar told IBD.
"The way that AMD is trying to compete is to offer value and performance," he said. "They're not going to provide you the full stack of software and predefined libraries. That's not something that AMD is capable of doing at this point or is focused on. But they want to provide you the best performance."
The AMD AI Chip Playbook
AMD will stick with its trusty playbook of getting a foothold in a market and gradually increasing its presence, Kumar said.
He said AMD has typically controlled market share of "between 20% and 30%." And if it could get to 20% of the AI chips market, that would be a huge business.
"The opportunity to double and triple the business fully exists even with a marginal share because the market is going to be so big," Kumar said.
AMD Is 'A Strong Second Contender' Vs. Nvidia
And AMD is upbeat about where the market is headed.
On a conference call with analysts on Jan. 30, Su, the chip firm's CEO, said the accelerator market could grow to about $400 billion by 2027. That figure took Wall Street by surprise because it was well above their forecasts.
"The beauty of the AI market is it's growing so quickly that I think we have both the market dynamic as well as our ability to gain share in that framework," Su said.
Speaking at the Morgan Stanley Technology, Media & Telecom Conference on March 5, AMD Chief Financial Officer Jean Hu defended the forecast.
"We can debate if it's $400 billion or $300 billion, but the trajectory of this technology trend is truly extraordinary," Hu said. She noted that the market was $40 billion last year and is likely to hit about $100 billion this year.
"It is a large market," she said. "We are a strong second contender in this market."
AMD Vs. Nvidia: How Sales Stack Up
AMD didn't really have a product to rival Nvidia in the AI market until Dec. 6 when it launched its MI300 data center GPU accelerator family. Now it is focused on gaining market share in the space.
Mizuho Securities analyst Vijay Rakesh estimates that Nvidia had 94% unit market share in AI server sales in 2023. But that is likely to drop to about 75% market share in 2025 and 2026 as rivals step up with competing products, he said.
AMD's data center sales were $6.5 billion in 2023, or 29% of its total sales. Its data center sales, which include traditional server central processing units, or CPUs, rose 7%. The rest of AMD's revenue came from PC, gaming and embedded processor sales.
Meanwhile, Nvidia's data center sales were $47.5 billion, or 78% of total sales, in its fiscal 2024 ended Jan. 28. Its data center sales rocketed 217% from the prior year.
Nvidia CEO Jensen Huang is understandably bullish on the AI chip market, which is finally bearing fruit for the graphics-chip maker after years of investment.
"Fundamentally the conditions are excellent for continued growth, calendar 2024 to calendar 2025 and beyond," Huang said on a conference call with analysts Feb. 21. "And let me tell you why. We're at the beginning of two industrywide transitions."
Those transitions are the shift from general to accelerated computing and the boom in generative artificial intelligence, he said.
Nvidia's Lead Vs. AMD Is Probably Safe
Nvidia's growth in the segment will be helped by an improving supply chain. But Huang expects demand to continue to outstrip supply through 2024.
Meanwhile, AMD will try to gain market share by offering a compelling value on a price-performance basis for its AI accelerators, Kumar said.
Data center GPUs can cost $20,000 to $40,000 apiece.
While AMD likely has a pricing edge over Nvidia, it is unlikely to ever lead the market, Moorhead believes.
"Nvidia would have to screw up big time somehow for AMD to ever be a No. 1," Moorhead said. "At the same time, AMD would have to hit on all cylinders."
But that's probably fine with AMD, Kumar said.
"They're letting Nvidia forge the market and they will come in as a second and try to get 20, 30 or more as a percentage of share," he said.
AMD Stock: No. 2 As A Wall Street Winner
Where does that leave AMD stock as the AMD vs. Nvidia battle plays out in AI chips?
To be sure, AMD's run as a runner-up has paid off for the chip giant's shares.
A decade after its struggles on Wall Street, AMD stock has been wowing investors. The shares have soared sevenfold in the last five years, while Intel has dropped nearly 14%. AMD's market cap now stands at roughly $315 billion, towering over Intel's $185 billion.
AMD stock's rise is not as meteoric as Nvidia stock, which has soared roughly twentyfold in the last five years. Nvidia's AI dominance has pushed the chip giant's market cap to more than $2 trillion.
But like Nvidia, AMD is a Wall Street star. Both companies have an IBD Composite Rating of 99, meaning they've outperform 99% of all stocks in terms of overall fundamental and technical strength. Intel's rating is 76. (Is AMD stock a buy now? See our analysis.)
Follow Patrick Seitz on X, formerly Twitter, at @IBD_PSeitz for more stories on consumer technology, software and semiconductor stocks.
