By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
The panelists generally agreed that while AMD and Micron have promising opportunities in inference/agentic AI and memory cycles, respectively, the dominant risks include hyperscaler shifts to custom silicon (ASICs), potential capex deceleration, and cyclical price normalization. The 'supercycle' narrative and high price targets may not be fully sustainable.
Risk: Hyperscaler shift to custom silicon (ASICs) and potential capex deceleration
Opportunity: AMD's inference/agentic AI positioning and Micron's memory supercycle
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 →
AMD is just at the start of a huge opportunity with inference and agentic AI.
Micron is riding a powerful memory cycle that shows no signs of letting up anytime soon.
Two of the hottest stocks in the market right now are Advanced Micro Devices (NASDAQ: AMD) and Micron (NASDAQ: MU). And according to some Wall Street analysts, these stocks still have more room to run.
Baird analyst Tristan Gerra has a street-high $625 price target on AMD, up from $300 ahead of the company's early May earnings report, representing about 25% from here (as of May 26). Meanwhile, UBS analyst Timothy Arcuri just tripled his price target on Micron from $535 to $1,625, representing around 80% upside.
Will AI create the world's first trillionaire? Our team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue »
Let's take a look at why these two artificial intelligence (AI) stocks are so hot right now and whether or not it's too late to get in.
AMD is riding two of the most powerful trends in AI right now in inference and agentic AI. These markets are just starting to boom, and AMD is well-positioned to see explosive growth in the coming years as a result.
While rival Nvidia has dominated the AI model training market, helped by most foundational AI code being written on its CUDA software platform, inference is less technologically demanding and tends to be more memory-bound than compute-bound. With hyperscalers looking to diversify away from Nvidia and looking for all the compute power they can get their hands on, AMD is set to be a big beneficiary. Its chiplet design can pack in more high-bandwidth memory (HBM), making its graphics processing units (GPUs) well-suited for inference. Meanwhile, it has already signed large GPU deals worth over $100 billion each, which should be a huge growth driver in the years ahead.
At the same time, the company has a huge opportunity in the data center central processing unit (CPU) market due to the rise of agentic AI. CPUs act like the brains of a computer, and to manage AI agents, AI data centers are going to need a whole lot more high-power CPUs. While the GPU-to-CPU ratio for training was 8:1, it moves to 1:1 for agentic AI. AMD is one of the leaders in this space in a market that it expects to rise to over $120 billion in the next few years.
Given its huge growth opportunities ahead, AMD looks poised to continue having solid upside from here.
The rise of Micron has been nothing short of spectacular. The stock is up more than 800% over the past year, largely due to earnings growth, and it still trades at a forward price-to-earnings (P/E) ratio of just 8.6 times fiscal 2027 analyst estimates.
Micron is one of the big three dynamic random-access memory (DRAM) makers, along with Korean competitors SK Hynix and Samsung. Right now, the entire DRAM market is very undersupplied due to surging demand for HBM. GPUs and other AI chips need to be packaged with HBM to optimize their performance, and as noted above with AMD, inference tends to be much more memory-bound than compute-bound, which is just adding another tailwind.
With the big memory makers focused on HBM and this specialized form of DRAM requiring upwards of three times the wafer capacity as ordinary DRAM, overall prices have skyrocketed, which has led to huge sales growth and enormous margin expansion for Micron and its competitors. And while these companies are working to increase capacity, they are also constrained by the finite number of EUV and DUV machines that ASML can produce each year.
While the memory market has historically been very cyclical with large boom-and-bust cycles, the big three players have started to lock in three to five-year commitments for the first time, trading some near-term price for more long-term visibility. Together with high HBM growth, ASML machine bottlenecks, and a low valuation, Micron could have a lot more room to run.
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 Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you’d have $471,072! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,303,352!
Now, it’s worth noting Stock Advisor’s total average return is 983% — a market-crushing outperformance compared to 210% for the S&P 500. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.
**Stock Advisor returns as of May 28, 2026. *
Geoffrey Seiler has positions in Advanced Micro Devices. The Motley Fool has positions in and recommends ASML, Advanced Micro Devices, Micron Technology, and Nvidia. 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
"AI infrastructure spending fatigue and Nvidia's entrenched position pose larger downside risks to both stocks than the bullish price targets acknowledge."
The article highlights AMD's inference/agentic AI positioning and Micron's HBM-driven memory supercycle as reasons for 25-80% upside, but downplays execution and demand risks. AMD's chiplet HBM advantage and $100B+ deals face Nvidia's CUDA moat and potential hyperscaler diversification limits, while Micron's 8.6x FY2027 P/E assumes sustained pricing power despite ASML bottlenecks. Historical DRAM cycles suggest the 3-5 year commitments may not fully insulate against 2026-27 capacity ramps or AI ROI scrutiny that could curb capex. Missing context includes AMD's data center CPU share gains versus Intel and whether agentic AI truly shifts GPU:CPU ratios to 1:1 at scale.
Signed multi-year GPU deals and locked-in memory contracts could still deliver the projected growth if inference workloads accelerate faster than training, overriding cyclical fears.
"The article treats booking announcements as revenue catalysts and cyclical DRAM pricing as solved by contracts, when both are timing risks that could compress valuations faster than the 25–80% upside materializes."
The article conflates two separate theses without stress-testing either. AMD's inference opportunity is real—but the 25% upside assumes Nvidia doesn't dominate inference too (it likely will, given software lock-in). More critically: AMD's $100B+ deal announcements are *bookings*, not revenue. Recognition happens over years. Micron's 8.6x forward P/E looks cheap until you remember DRAM is cyclical; the article acknowledges this but then dismisses it via multi-year contracts. Those contracts lock in *prices* during a supply crunch—exactly when you want to sell, not commit. ASML bottlenecks are real but not new; they're already priced into ASML's valuation, not a surprise catalyst.
If inference truly is memory-bound and HBM demand sustains at current levels, Micron's long-term contracts actually protect margin during the inevitable downturn, making the risk/reward genuinely asymmetric at 8.6x forward P/E.
"The market is underestimating the risk of custom silicon cannibalization for AMD and the inevitable return of cyclical volatility for Micron's memory business."
The article's reliance on extreme price targets—specifically the $1,625 figure for Micron—is dangerously detached from historical cyclicality. While AMD has a legitimate path in inference and agentic AI, the '1:1 GPU-to-CPU' ratio claim is a massive oversimplification that ignores the rapid rise of custom silicon (ASICs) from hyperscalers like Google and AWS. Micron is currently benefiting from HBM shortages, but memory is a commodity business; once capacity catches up, the 'supercycle' narrative often collapses under the weight of inventory gluts. Investors are ignoring the massive capital expenditure (CapEx) required to maintain these growth rates, which will likely compress free cash flow margins significantly through 2026.
If the shift toward agentic AI creates a permanent, non-cyclical floor for memory demand, the traditional boom-bust cycle of DRAM may be permanently broken, justifying a structural re-rating of Micron's P/E multiple.
"The upside for AMD and MU relies on a durable AI capex cycle and continued memory tightness, but any slowdown in hyperscaler AI deployments or a memory-price reversion could break the thesis."
Read through the article, the bull thesis says AMD benefits from inference/agentic AI and MU memory cycle, with high price targets. But the strongest risk is that AI capex is highly concentrated among a handful of hyperscalers; chips orders are lumpy and tied to capacity expansions that may take longer to monetize. Inference is memory-bound but demand could shift as software optimizes; Nvidia remains dominant in training, reducing incremental share for AMD. Micron's margin gains rely on HBM demand and ASML supply constraints; any easing would compress profitability. The article glosses over potential price normalization and capital expenditure cycles that could cap multiples.
Counter: AI capex could plateau if hyperscalers reach saturation or shift spending toward Nvidia-centric ecosystems; memory prices also cycle, so MU's margin upside may prove temporary rather than secular.
"AMD and Micron face simultaneous volume risk from ASIC adoption that separate cyclical analyses miss."
Gemini's ASIC warning creates a correlated risk for both names that no one connected. AMD's inference edge and Micron's HBM volumes both depend on continued GPU-centric builds; any hyperscaler shift to custom silicon hits memory demand and AMD share at once. The 8.6x MU multiple and AMD bookings look far less insulated once that single trigger is modeled, rather than treating cyclical and competitive risks as independent.
"ASIC risk is real but secondary to total capex growth; the cycle breaks if spending slows, not if share shifts."
Grok's ASIC correlation is sharp, but it assumes hyperscalers abandon GPU-centric stacks wholesale. Reality: Google TPUs and AWS Trainium coexist with Nvidia; they're not binary. More pressing: even if custom silicon gains 15-20% share, HBM demand still grows if total AI capex expands faster than ASIC adoption. The real trigger is whether hyperscaler capex *decelerates*—not diversification. That's the cyclical floor nobody's modeled.
"Hyperscaler custom silicon adoption will commoditize AI hardware, leading to margin compression that invalidates current premium valuation multiples for AMD and Micron."
Claude, you’re missing the power dynamic: custom silicon isn't just about diversification; it’s about margin capture. Hyperscalers are building ASICs to strip out the 'Nvidia tax,' which directly compresses the total addressable market for merchant silicon. If AMD and Micron are forced to compete against in-house stacks that prioritize cost efficiency over performance, the 'supercycle' narrative crumbles. We are looking at a structural shift toward commoditized AI hardware that will inevitably compress multiples across the entire sector.
"ASIC adoption won't erase AMD's and Micron's AI opportunities; memory demand will persist due to bandwidth needs in hybrid architectures, keeping upside intact even as hyperscalers lean toward custom silicon."
Gemini's ASIC warning is real but not a binary death knell for AMD and Micron. Hyperscalers won't abandon GPU ecosystems; they run hybrids that still need memory bandwidth, so HBM demand could stay resilient even as ASICs gain. The risk is more about timing, capex cycles, and price normalization than a structural collapse in AI demand. In short, the AMD/Micron thesis survives a softer ASIC surprise, but with narrower margins.
The panelists generally agreed that while AMD and Micron have promising opportunities in inference/agentic AI and memory cycles, respectively, the dominant risks include hyperscaler shifts to custom silicon (ASICs), potential capex deceleration, and cyclical price normalization. The 'supercycle' narrative and high price targets may not be fully sustainable.
AMD's inference/agentic AI positioning and Micron's memory supercycle
Hyperscaler shift to custom silicon (ASICs) and potential capex deceleration