By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
The panel is divided on AMD's potential. While Gemini sees a significant earnings growth opportunity if AMD captures 15-20% of the inference market, Claude and ChatGPT caution that the revenue impact of the Microsoft deal is uncertain and may not justify AMD's current valuation. The geopolitical risk of supply chain constraints, particularly around TSMC's CoWoS capacity, is a major concern for all panelists.
Risk: Supply chain constraints, particularly around TSMC's CoWoS capacity, could cap AMD's upside and collapse the 'multi-client growth' thesis.
Opportunity: If AMD can secure long-term allocations of TSMC's CoWoS capacity and gain market share in the inference market, it could drive significant earnings growth and outpace current consensus estimates.
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 →
Watching for new or strengthening partnerships in the artificial intelligence (AI) realm is a great way to keep tabs on which companies are excelling and which ones are floundering. AMD (NASDAQ: AMD) spent the better part of the first few years of the AI arms race floundering, but it has made some strides in 2026. One partner AMD has been deepening its relationship with is Microsoft (NASDAQ: MSFT). Microsoft has long looked to diversify its relationship with computing unit suppliers, and its latest deal with AMD is a sure sign of that.
But how big of a deal is this for AMD? Let's find out.
Missed Nvidia in 2009? This Rare Signal Is Flashing Again. In 2009, a "Double Down" signal flashed for a little-known chipmaker called Nvidia. For the first time in years, that same "Total Conviction" signal is flashing for a company 1/100th the size of Nvidia. Continue »
AMD recently announced a new partnership with Microsoft to deploy its Helios platform, which is AMD's AI rack system. This includes AMD graphics processing units (GPUs), central processing units (CPUs), networking, and software, so it's an all-encompassing solution that allows users to completely get away from rival Nvidia products.
Microsoft will be using this to power frontier model inference, which brings up a few interesting points. First, frontier AI models are the most advanced, so using AMD's hardware in this application showcases that Microsoft believes that these are some of the best options available. Second, they are using them for inference. That's another big deal, as inference with an AI model occurs after they're trained. Inference is a completely different type of workload than training, and using AMD's likely cheaper products for inference makes sense because having the best product there doesn't matter quite as much as it does for training AI. So, the takeaway here is that Nvidia isn't necessarily losing the technological battle to AMD; it may just not have as cheap a product in inference settings.
However, that's not a bad thing for AMD. After the initial phase of AI is complete, inference will make up the majority of workloads versus training. If AMD can capture a large chunk of this market, then it could become a giant in this space and become more of a true rival to Nvidia, as right now it's much smaller.
The primary problem here is that AMD already has a lot of this success priced into the stock.
AMD has been a hot stock in 2026, more than doubling so far. However, a lot of that has come on expectations of success rather than several quarters of impressive results. Right now, AMD trades for a pretty pricey 68 times forward earnings.
AMD PE Ratio (Forward) data by YCharts.
For comparison, Nvidia trades for 25 times forward earnings. That's a huge premium over its peer, and AMD needs to deliver impressive growth to come back down to a more reasonable valuation.
This deal with Microsoft is a good start, but the market will need to see more from AMD from other clients if it's going to grow into its pricey valuation.
So, is AMD a buy here? I'd say no. The stock has run up too far, too fast without any major, needle-moving deals. The Microsoft announcement was a good start, but it's going to need a lot more than that to justify its price tag. I'd much rather invest in Nvidia, which is growing at a faster rate, is still the top option in the industry, and is trading at a far cheaper valuation.
Over the next few years, I'd expect Nvidia to outperform AMD, as AMD has a much higher starting point that could take multiple years of strong growth to return it to a normal valuation level. This is a red flag for me, so I'm steering clear of AMD's stock for the time being.
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 $421,511! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,381,960!
Now, it’s worth noting Stock Advisor’s total average return is 981% — a market-crushing outperformance compared to 216% 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 August 18, 2026. *
Keithen Drury has positions in Microsoft and Nvidia. The Motley Fool has positions in and recommends Advanced Micro Devices, Microsoft, and Nvidia. The Motley Fool has a disclosure policy.
Four leading AI models discuss this article
"The pivot from AI training to high-volume inference creates a structural tailwind for AMD that justifies a premium valuation despite current high forward P/E multiples."
The market is fundamentally mispricing the transition from training to inference. While the article correctly identifies AMD's 'Helios' platform as a win, it dismisses the margin profile of inference too easily. If Microsoft shifts to AMD for inference, the volume-to-margin trade-off is massive. Trading at 68x forward P/E is undeniably rich, but if AMD captures 15-20% of the inference market, the earnings per share (EPS) growth will likely outpace the current consensus, forcing a re-rating. The article's comparison to Nvidia's 25x forward P/E ignores that Nvidia's growth rate is facing tougher year-over-year comps, whereas AMD is starting from a lower base in the data center segment.
The bull case assumes AMD’s software stack, ROCm, can achieve parity with Nvidia’s CUDA ecosystem; if developers continue to prioritize CUDA, AMD’s hardware efficiency gains will be rendered irrelevant by high switching costs.
"A single partnership announcement for lower-margin inference work does not justify a 68x forward P/E multiple when the company has yet to prove it can convert partnerships into sustained revenue growth at scale."
The article conflates a partnership announcement with material revenue impact—a common trap. Yes, Microsoft deploying AMD's Helios for inference is real, but the article never quantifies: what's the TAM (total addressable market) here? What's AMD's expected revenue contribution? Inference is lower-margin than training, and the article admits Nvidia isn't losing on technology—just price. AMD at 68x forward P/E versus Nvidia at 25x is a real valuation gap, but that gap exists because Nvidia has proven execution and market share. One Microsoft deal doesn't close that gap. The article also omits: does this displace existing Nvidia deployments, or is it additive? If additive, it's less bullish than framed.
If AMD's inference workloads scale to 40%+ of Microsoft's AI compute (plausible given inference dominates post-training), and AMD's gross margins improve to 50%+ on volume, the stock could compress to 35-40x forward P/E and still outperform Nvidia on a 3-year horizon—making today's premium defensible.
"AMD’s MSFT deal isn’t enough to justify the current valuation without broad multi-client adoption and sustained margins; the stock likely needs more durable, multi-quarter revenue visibility to justify a ~68x forward P/E."
MSFT's Helios deal signals cloud-scale validation for AMD, but it's not a game-changing revenue driver by itself. The incremental payoff may be modest and elongated, while Nvidia still dominates training and many inference use cases, keeping AMD in a pricing and execution risk corridor. At 68x forward earnings, AMD hinges on multi-client growth and margin leverage that the article glosses over. Supply constraints, competitive AWS/MSFT budget cycles, and the risk that Helios remains a niche or one-off deployment could cap upside. In short, the catalyst exists, but the risk-reward requires sustained multi-client adoption and solid profitability.
Bullish counterpoint: a high-visibility deal with Microsoft could validate AMD’s roadmap, unlock scale across other hyperscalers, and justify a higher multiple if multi-client traction follows. If that occurs, Nvidia’s edge could erode and AMD’s valuation could re-rate.
"N/A"
[Unavailable]
"AMD's growth thesis is constrained by TSMC's CoWoS packaging capacity, which creates a hard ceiling on scaling regardless of software adoption."
Claude, you’re missing the geopolitical risk inherent in this supply chain. Even if Helios scales, AMD relies heavily on TSMC’s CoWoS (Chip-on-Wafer-on-Substrate) capacity, which is currently the industry’s primary bottleneck. If Microsoft prioritizes AMD, they aren't just fighting Nvidia; they are fighting Nvidia for the same limited packaging capacity. AMD’s valuation at 68x forward P/E assumes frictionless scaling, but if they hit a supply ceiling, that 'multi-client growth' thesis collapses regardless of software parity.
"CoWoS scarcity could be a competitive advantage for AMD if Microsoft secures long-term allocation, not a cap on growth."
Gemini's CoWoS bottleneck is real, but it cuts both ways. If TSMC capacity is truly constrained, Microsoft's Helios deployment *forces* capacity allocation—AMD doesn't need to outcompete Nvidia for packaging; they need contractual priority. That's actually a moat. The real question: does TSMC expand CoWoS capacity for AMD, or does Nvidia's scale lock it down? Article doesn't address this at all.
"CoWoS constraints may become a moat via long-term allocations, while the real risk is demand dynamics and software ecosystem adoption, not packaging limits alone."
Response to Gemini: The CoWoS bottleneck is real but not a binary ceiling. If MSFT and peers secure long-term allocations, capacity becomes a moat and drives pricing power, not collapse. The bigger unknown is demand elasticity for inference and how fast ROCm gains traction; even with constraints, AMD can monetize higher-margin inference through multi-client deals if volumes scale without relying on a miracle.
[Unavailable]
The panel is divided on AMD's potential. While Gemini sees a significant earnings growth opportunity if AMD captures 15-20% of the inference market, Claude and ChatGPT caution that the revenue impact of the Microsoft deal is uncertain and may not justify AMD's current valuation. The geopolitical risk of supply chain constraints, particularly around TSMC's CoWoS capacity, is a major concern for all panelists.
If AMD can secure long-term allocations of TSMC's CoWoS capacity and gain market share in the inference market, it could drive significant earnings growth and outpace current consensus estimates.
Supply chain constraints, particularly around TSMC's CoWoS capacity, could cap AMD's upside and collapse the 'multi-client growth' thesis.