By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
The panel is largely bearish on AMD's current valuation, citing high P/E ratios, Nvidia's software ecosystem dominance, and potential risks from custom silicon and supply constraints. They agree that AMD has competitive technology but question whether its stock is fairly priced.
Risk: Custom silicon risk, where hyperscalers prioritize vertical integration, reducing AMD's merchant silicon market.
Opportunity: AMD's hardware efficiency and potential for rapid AI revenue growth if it hits its $4B+ AI revenue target for 2024.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Advanced Micro Devices (NASDAQ: AMD) has been a big winner in artificial intelligence (AI) boom. The chip designer's stock has gained 107% since OpenAI launched the game-changing ChatGPT platform. While that price climb can't compare to AI market darlings Nvidia (NASDAQ: NVDA) and Super Micro Computers (NASDAQ: SMCI), it's a market-stomping performance.
Less than 30 stocks among the 503 members of the S&P 500 (SNPINDEX: ^GSPC) index have done better in this period, and the index itself "only" gained 41%. The index gain works out to a compound annual growth rate (CAGR) of 21.8%. If I could match that soaring speed over a full decade, I'd turn a $1,000 investment into a $7,200 return. And AMD is smoking that wealth-building S&P 500 return right now.
But that's already old news. Past results are no guarantee of future success. The burning question existing and potential AMD investors should ask today is whether the stock has room to run any higher from this lofty perch.
AMD is no spring chicken in the high-performance computing market. The AMD Instinct MI250X accelerator chip Is found in 10 systems on the latest list of the world's 500 largest and most powerful supercomputers.
Nvidia dominates that system list, led by at least 48 systems using the V100 accelerator and more than 80 relying on some version of the A100 chip. But when you look at the performance numbers, the leading AMD chip with a 2% share of system counts delivers a staggering 21.4% of the entire list's number-crunching performance.
The Instinct MI250X didn't do it through unbeatable performance per chip. This achievement was built on massive installation numbers. The AMD Instinct systems have 12.7 million processor cores at work in those ten massive supercomputers, led by 8.7 million Instinct cores in the greatest performer of them all -- the Frontier system at the Oak Ridge National Laboratory. The largest Nvidia-based systems stop around 2 million processor cores.
I apologize if that got a bit nerdy with processor details and big numbers. In short, a lot of cheaper chips can do the same job as a few expensive, top-of-the-line processors. I'm trying to make a simple point, though:
AMD's AI accelerators are very competitive, especially if you include unit pricing. Throwing lots of lower-priced hardware at AI problems can make up for any edge Nvidia's chips might have in terms of plain performance.
In other words, AMD may deserve a bigger slice of the AI market than it gets credit for today. CFO Jean Hu and CEO Lisa Su like to focus on the total cost of ownership (TCO) when comparing AMD's products to Nvidia's, and the company is already winning system deals from that angle. This concept includes chip prices, power consumption, support for more memory per chip, and other total system aspects that can outweigh Nvidia's lead in pure performance per chip.
So AMD should be on your list of AI hardware providers to watch. Should it be your next buy in the AI space?
As it turns out, AMD's stock might already have the market respect it deserves. As noted earlier, AMD's stock chart makes the broader market look stale and stalled. Moreover, these shares are more richly valued than Nvidia's in many ways:
| Valuation Metric | AMD | Nvidia | |---|---|---| | Price to earnings (P/E) | 183 | 72 | | Price to free cash flow (P/FCF) | 183 | 80 | | Enterprise value to earnings before interest and taxes (EV/EBIT) | 275 | 63 |
AMD's shares are more affordable in terms of price to sales or the price to earnings to growth ratio, so it's not a slam dunk in either direction. Still, it's hard to call AMD's stock "affordable" in any way, shape, or form.
Longtime AMD investors should be used to uncomfortably high valuation ratios. This stock has a long history of soaring to lofty valuation ratios.
Due to AMD's tendency toward low profit margins, the math gets a little bit absurd sometimes. P/FCF and P/E ratios were measured in the thousands in 2004,as the K2 line of PC processors put up a spirited fight against Intel's (NASDAQ: INTC) Celeron and Pentium 4 lineups. AMD stock was also more pricey than Nvidia's from 2018 to 2021, when both companies had unexpected manufacturing advantages over a struggling Intel.
So the real question is, will AMD be able to carve out a lucrative niche for itself in this era of heavy AI interest?
I'm thinking that it can. Nothing seems impossible under the leadership of Lisa Su, who also seems to have a good handle on how to make the most of this game-changing moment. AMD isn't trying to be Nvidia 2.0, but choosing a different path with significantly different plans and designs.
So the stock is expensive, but then it kind of always is. And I think there's plenty of room for a different approach to AI computing from a company (or several) not named Nvidia. Check your risk tolerance before going any further, but AMD could be an interesting buy for growth investors who respect Lisa Su's leadership.
Before you buy stock in Advanced Micro Devices, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Advanced Micro Devices wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.
Consider when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $792,725!*
Stock Advisor provides investors with an easy-to-follow blueprint for success, including guidance on building a portfolio, regular updates from analysts, and two new stock picks each month. The Stock Advisor service has more than quadrupled the return of S&P 500 since 2002*.
*Stock Advisor returns as of August 22, 2024
Anders Bylund has positions in Intel and Nvidia. The Motley Fool has positions in and recommends Advanced Micro Devices and Nvidia. The Motley Fool recommends Intel and recommends the following options: short August 2024 $35 calls on Intel. The Motley Fool has a disclosure policy.
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.
Four leading AI models discuss this article
"AMD’s valuation is currently disconnected from near-term earnings, making its stock a pure bet on the company's ability to successfully scale its software ecosystem and capture the inference market."
The article correctly highlights AMD's TCO (Total Cost of Ownership) advantage, but it glosses over the 'software moat' that Nvidia’s CUDA ecosystem provides. While AMD’s MI300X series is impressive hardware, enterprise adoption is gated by developer tooling. At a 183x P/E, the market is pricing AMD for perfection—assuming they can capture a significant percentage of the inference market from Nvidia. However, if enterprise AI spending shifts from training to inference, AMD’s hardware efficiency could actually become a major tailwind. The valuation is aggressive, but if they hit their $4B+ AI revenue target for 2024, the forward P/E compresses rapidly, making the current premium look like a growth-priced entry point rather than a bubble.
The bear case is that AMD remains a perennial 'second-best' hardware vendor whose margins will be perpetually squeezed by Nvidia's software dominance and the emergence of custom silicon from hyperscalers like Amazon and Google.
"AMD's technology competitiveness does not justify a 2.5–4.4x valuation premium to Nvidia when AMD has yet to prove it can capture meaningful server CPU or accelerator share at scale."
The article conflates two separate questions: whether AMD's *technology* is competitive (likely true) versus whether AMD's *stock* is fairly priced (highly questionable). AMD trades at 183x P/E and 275x EV/EBIT versus Nvidia's 72x and 63x—a 2.5–4.4x premium—while the article offers no earnings growth differential to justify it. The supercomputer data is cherry-picked: 10 systems with AMD chips versus 128+ with Nvidia doesn't suggest market-share capture. TCO arguments are real but not new; they haven't dented Nvidia's moat for three years. The author's historical precedent (2004, 2018–2021) actually shows AMD's valuation excesses *precede* disappointment, not outperformance.
AMD's MI300X (not discussed) ships in 2024 with materially better performance-per-watt than MI250X, and if hyperscalers genuinely adopt it at scale, the valuation gap could compress upward for Nvidia, not downward for AMD—meaning AMD could still underperform despite winning design wins.
"AMD's 183x P/E premium is unsupported given its negligible AI accelerator market share versus Nvidia's ecosystem dominance."
The article correctly flags AMD's stretched multiples (183x P/E, 275x EV/EBIT) versus Nvidia yet still endorses the stock on TCO and Lisa Su's execution. What it underplays is AMD's minimal data-center AI share—MI250X powers just 2% of top-500 systems while Nvidia holds the ecosystem via CUDA. Without rapid MI300 adoption or share gains in 2025, the 107% run since ChatGPT looks like a one-time re-rating rather than sustainable growth. Supercomputer wins also rely on volume of cheaper cores, not recurring high-margin software revenue Nvidia captures.
AMD could still close the gap if hyperscalers prioritize cost and power efficiency over CUDA compatibility, allowing Su to convert TCO wins into durable 20%+ share.
"Valuation already prices in a prolonged AI-driven reinvestment cycle; without durable data-center share gains and margin expansion, AMD risks multiple contraction even if AI demand stays healthy."
AMD has a credible niche in AI acceleration on a total-cost-of-ownership basis, yet the article understates several headwinds. Nvidia still dominates software ecosystems, developer momentum, and big-system demand, making AMD's share gains look incremental rather than revolutionary. The stock's valuation is extreme (P/E and P/FCF well over 100x versus Nvidia’s ~70x in the cited context), implying a long, undiluted bet on AI capex staying red-hot. Even if AMD's accelerators offer cost advantages, a cyclical AI rush could fade and compress multiples. The risk also hinges on supply by TSMC, customer concentration in hyperscalers, and competition from entrenched players—execute well, but execution risk remains high.
The flip side is that the AI cycle may not be Nvidia-centric for long; if AMD's TCO edge and ROCm platform win meaningful hyperscaler traction, the stock could re-rate faster than skeptics expect. In that scenario, the optimistic multiple becomes justified rather than a mispricing.
"The rise of custom silicon from hyperscalers creates a structural ceiling for AMD's merchant accelerator market share regardless of their TCO advantage."
Claude and Grok focus on the P/E gap, but you are all ignoring the 'Custom Silicon' risk. Hyperscalers like Amazon (Trainium) and Google (TPU) are building internal alternatives, not just choosing between AMD and Nvidia. If these clouds prioritize vertical integration, AMD’s merchant silicon market is smaller than the bull case assumes. The valuation isn't just a bet on beating Nvidia; it’s a bet that the 'AI pie' grows fast enough to support both merchant and custom providers.
"Custom silicon threatens AMD's margin structure more than its total addressable market, but the article never quantifies the capex split assumption."
Gemini's custom silicon risk is real, but underweighted here. Amazon and Google build TPUs/Trainium for *their own* workloads, not merchant sales. AMD's TAM isn't shrinking—it's the addressable *margin* that's at risk. The real question: do hyperscalers reserve 30% capex for custom silicon and 70% for merchant (Nvidia/AMD), or flip that ratio? Article assumes the former; nobody's modeled the latter credibly.
"Nvidia's software moat protects its merchant margins more effectively than AMD's TCO story can offset custom-silicon encroachment."
Claude correctly flags margin pressure from custom silicon but misses how Nvidia's CUDA lock-in lets it retain high-margin merchant share even as hyperscalers divert capex. AMD lacks that defense, so any shift toward 30-50% custom workloads compresses AMD's addressable revenue faster than Nvidia's. The TCO edge cited earlier then matters less if the merchant pie itself shrinks unevenly.
"The real risk to AMD's bull case isn't just custom silicon, but ramp cadence and supply constraints that could cap AI revenue and prevent the merchant TAM from supporting the current valuation."
Gemini's 'custom silicon' risk is valid, but the bigger overlooked risk is supply and ramp cadence. MI300X volumes hinge on TSMC capacity and hyperscaler mix; a slower-than-expected ramp or allocation of wafer fab to other customers could cap AMD's AI revenue upside and let Nvidia's moat matter even more. If capex shifts toward internal accelerators, the merchant TAM may never justify the current multiple.
The panel is largely bearish on AMD's current valuation, citing high P/E ratios, Nvidia's software ecosystem dominance, and potential risks from custom silicon and supply constraints. They agree that AMD has competitive technology but question whether its stock is fairly priced.
AMD's hardware efficiency and potential for rapid AI revenue growth if it hits its $4B+ AI revenue target for 2024.
Custom silicon risk, where hyperscalers prioritize vertical integration, reducing AMD's merchant silicon market.