Breakthrough Chips, Hyperscale Deals, Valuation Risks

This article first appeared on GuruFocus.

AMD (NASDAQ:AMD) is very much the frontrunner in the next generation of computers, AI, data centers and edge technology. Its lineup, whether it is premium EPYC CPUs and Instinct AI chips or Ryzen game computers and intelligent Xilinx modules, is an all-around choice for new workloads. They are operating well, closing business with the largest hyperscalers, and continue to innovate, so it is no wonder people perceive them as a leader. The only disadvantage is that this optimism was already in large quantities in the stock, leaving hardly any room for error.

AMD Stock has rallied more than 114% in 2025, driven by the AI boom. To long-term players, AMD remains a quality growth opportunity with massive potential, albeit, though, you will have to hold tight and should be prepared for volatility

The AMD Story: Breakthrough Chips, Hyperscale Deals, Valuation Risks

AMD is in the middle of a hard competitive environment. The former competitor Intel (NASDAQ:INTC) remains ahead in terms of revenues and size, but AMD has shortened the distance significantly. In PCs, Intel has approximately 75% of the CPU market, which is being fed upon by AMD in the high-end niche and is already edging away, as share changes demonstrate in recent years.

On the servers, Intel has fallen to approximately 50% (compared to 90% a few years ago), leaving the others to AMD and a slight margin of Arm-based Graviton. The future of Intel will depend on the new products (Xeon Sapphire Rapids/Granite Rapids, new 18A process, etc.) and a stable architecture roadmap.

Their margins are fading (single-digit in data centers), and the historic lag leaves AMD with an advantage, but a turnaround (Diamond Rapids, 18A process) could be an out-of-pocket expense for AMD. Other CPU vendors are also emerging: consider Arm-based (new cloud CPUs by Qualcomm, AWS Graviton) in servers, and M-series chips are competing with Intel in Macs and possibly in laptops. Everything that may decelerate AMD’s PC momentum.

Nvidia is far ahead in the case of GPUs or AI accelerators. They own a majority stake in discrete graphics and data-center AI cards. The majority calculate the add-in shipments of Nvidia at a greater than 90% (the remaining 10% comprises AMD and Intel). These improvements are in the Radeon and Instinct GPUs by AMD, which MI300 will compete with Nvidia, unlike the H100, but the software and market ecosystem (CUDA) is well established with Nvidia. AMD has an open ROCm stack under development, and the collision (such as OpenAI, Oracle, and others) is helping AMD get momentum; however, a significant obstacle is still Nvidia. Other players of AI chipsets (Graphcore, Habana, and others) are significantly smaller.

The AMD vs. Intel dynamic is especially important. AMD is fully fabless; it outsources manufacturing to TSMC/Samsung, whereas Intel still owns and invests heavily in its own fabs. This makes their capital structures very different. Industry analysts note that when AMD spun off its fabs in 2009, it became a formidable design-centric competitor with the money drain from manufacturing gone, allowing focus on chip. Intel, by contrast, has been pouring tens of billions into cutting-edge factories (e.g. Intel 18A, Foundry build-outs). This has led to cost overruns and delays.

AMD and Intel now run on different playbooks. AMD is fabless and capital-light, outsourcing manufacturing to TSMC and Samsung; Intel is rebuilding a foundry business with heavy capex. That divergence matters for returns: a capital-light model can deliver higher free cash flow if growth holds, but it is also exposed to foundry supply constraints

In short, AMD’s capital-light model yields higher returns on invested capital and free cash flow (helping fund R&D) than Intel’s asset-heavy approach. As one commentator put it, Intel saw the positive results with AMD, which soared after distancing itself from foundry headaches. In other words, AMD’s profitability benefits from avoiding massive fab expenses.

Long-term, AMD’s lower capex means its intrinsic value per dollar of revenue can be higher than Intel’s, but only if growth continues.

Talking about AMD’s financial performance is great so far. The latest Q3 results of AMD were off the scale. The company earned a record revenue of $9.246 billion, increasing by nearly 36% compared to the items in the Q3 of 2024. GAAP gross margin stood at 52%, indicating that the mix remains on target. GAAP operating income stood at $1.270 bn and GAAP net income at $1.243 billion, and produced EPS of $0.75 in the diluted version, a 61% increase in net and a 60% increase in EPS over the previous year.

The AMD Story: Breakthrough Chips, Hyperscale Deals, Valuation Risks
The AMD Story: Breakthrough Chips, Hyperscale Deals, Valuation Risks

On the non-GAAP, it was also comparable, with non-GAAP operating income of $2.238bn and non-GAAP net income of $1.965bn, and the EPS at $1.20, doing so, or nearly three times less than the previous year. CFO Jean Hu referred to it as record-free cash flow in order to match the revenue boom.

The headline numbers were maintained by segment results. Data center revenue totals $4.34 billion in Q3 ’25, 22% YoY, and was supported by high demand for 5th-gen EPYC computer processors and the AMD Instinct MI350 AI chips. Client & Gaming segments billed a record of $4 billion, up 73% in that blend. Client (PC) revenue recorded a record of $2.80 billion (up 46%), and Gaming (Radeon plus semi-custom) broke a record of $1.305 billion (up 181%). Embedded and Adaptive revenue fell to $857 million, or approximately 8% year-over-year, with ease in certain industry and networking mixes.

Management stayed upbeat. In Q4, AMD is aiming at a revenue of about $9.6 billion (approximately 30% of a year-over-year increase) and a gross margin of approximately 54.5%, despite the use of a knockout part of China shipments associated with export controls.

In retrospect, AMD’s growth trajectory of AMD has been stable and massive in the past five years. The yearly income changed between approximately $9.8 billion and $25.8 billion between 2020 and 2024, respectively, a compound growth rate of the mid-twenties to high-twenties percent. Ryzen and EPYC took off in 202122, then PC and crypto were down in 202324, and again in 2025. Projected final 12 months leading to Q3 25 will bring the trailing revenue to near $32 billion, with AI and data-center demand being the core drivers.

The AMD Story: Breakthrough Chips, Hyperscale Deals, Valuation Risks
The AMD Story: Breakthrough Chips, Hyperscale Deals, Valuation Risks

The increase in market share is broad-based. Desktop CPU share is 39% or so by mid-2025, which is a significant improvement over the low-teens a few years earlier. AMD has approximately 40% of the CPU revenue in servers, and has been making a reported plurality of shipments, which have hurt the long-established Intel. NVIDIA continues to be a leader in AI/data-centers chips in GPUs, but AMD is venturing further into the gaming sector with new Radeon models.

AMD has a ton of big things planned in 2025 and beyond: although competition remains stiff, the company has that so far.

Enormous generative-AI tailwind. The new Instinct GPUs (MI300/MI350) have been manufactured in bulk. In 2025, AMD recorded more than $1 billion in quarterly revenue from data-center GPUs, a first for the company, and as early as 2026, could launch an MI350X (CDNA4) with significant performance gains. The momentum supports AMD’s expansion to recent wins: AMD is collaborating to deliver six vis-a-vis sem’s of GPU power to OpenAI, and the first 1-GW rollout of MI450 units is coming by late 2026. Oracle (ORCL) is launching AMD Helios rack design with 50,000 Instinct GPUs, and Amazon, Meta, Microsoft, IBM, and others are deploying AMD GPUs in large quantities to AI/HPC tasks. These alliances and continuous ROCm software modifications leave AMD in an excellent position to take over a larger share of the AI infrastructure market.

Share Gains in CPU Servers

EPYC continues to beat Intel in terms of price to performance, and major clouds, such as Oracle (ORCL) OCI and Google Cloud, are launching new instances powered by AMD. Mercury Research and PassMark indicate that the amount of server CPU revenue AMD shares has surged to approximately 40% by mid-2025. Each new generation of the EPYC cores, Genoa, Bergamo, and the upcoming Genoa-X, scheduled to be released in 2025, has had an increase in cores and power efficiency. That allows AMD to enter into hyperscale data centers and HPC clusters. Along with the acquisition of Pensando Systems, a smart-NIC champion, and the addition of Xilinx networking IP, this can create additional data-center opportunities. The upcoming generation of chipsets, such as the Venice EPYC (Zen 5), will also drive the mid-cycle upgrades.

All in all, the AI push in enterprise and cloud represents the largest near-term catalyst for AMD. It is a roadmap, a GPU that aims at seizing that wave. The CEO continues to profile that by pursuing innovation in high-performance and adaptive computing, AMD finds itself in a position to grow in the long term.

Advanced Micro Devices (NASDAQ:AMD) is trading at premium levels that show just how much optimism investors have about its long-term position in AI and data centers. But when you dig into the numbers, it looks like a lot of that optimism might already be baked into the price.

The AMD Story: Breakthrough Chips, Hyperscale Deals, Valuation Risks
The AMD Story: Breakthrough Chips, Hyperscale Deals, Valuation Risks

As of late 2025, AMD trades near $250 per share, with trailing twelve-month earnings of about $2 per share. That translates into a trailing P/E ratio of about 125 times, which is an elevated valuation, by most standards, particularly against its past records and the rest of the chip industry.

In the future, analysts believe that AMD will report about $5.00-$5.50 per share in total profit for fiscal year 2026. Taking the halfway approach, which turns out to be approximately a 48 P/E forward. For comparison, Nvidia (NASDAQ:NVDA) has a price-to-forward earnings ratio of approximately 34, whereas Intel (NASDAQ:INTC) has a significantly lower ratio due to its weaker performance. AMD is currently overvalued relative to Nvidia, by approximately 40 on future expectations; that is, investors are willing to pay an extra dollar for future expected profits.

The same can be said about other valuation metrics. The company’s EV/EBITDA is approximately 69x, higher than Nvidia’s (48x) and Intel’s (20x). It has a price-to-sales ratio of approximately 13x and a price-to-book ratio of approximately 6.9x, which are high in comparison with chipmakers. Concisely, investors are overpaying by far for the growth potential of AMD.

By putting in a forward P/E of 47.6 and a PEG ratio of approximately 1.2 , the market is in effect pricing in aggressive growth assumptions on the future growth of earnings of approximately 40% outside the next few years. That’s an ambitious target. Analysts now anticipate revenues per year increasing 28% to 35% through 2027, primarily as a result of demand in artificial intelligence and data centers chips. If AMD continues operating at a 5254% margin and becomes more efficient, its EPS growth could increase to 35-40%, which is still slightly lower than the implied growth in the stock price.

The stock is currently priced expensively, although not irrationally. It is priced in advance, and the firm will be required to perform to the letter to explain the prices at which the shares are trading.

In spite of this bullish case, there are a number of risks that mitigate the thesis of investment:

Supply and Geopolitical-Hiccups: AMD is fab-less, and thus relies on such foundries as TSMC or Samsung. In case those plants become full capacity or sluggish or become the target of an earthquake, supply may fail. Besides, there is a concern about U.S.-China trade tensions. One of the things that the New export rules prevented was the sale of a portion of Instinct GPUs to China- AMD went as far as to claim that its Q3 performance does not have any MI308 units sold to China. This is in addition to an earlier inventory write-down this year of an equivalent export law of an amount of $800M. Either of the trade bans may cut off a part of the AMD market or may require costly fixes.

Competition / Price Pushback: It is not that Intel and Nvidia are going to give AMD a free hand, as favored by the market. Intel may reduce the costs of EPYC or hand out favors. Each year, Nvidia is claimed to release a new chip and may beat AMD on the design or software side. Assuming excess supply of GPUs enters the market (some analysts estimate this to occur in the year 2024), there will be a low price of GPUs and CPU, and hence they will squeeze the AMD average selling price. Massive discounts on AMD and Intel server CPUs have already been observed, which is damaging the margins. And do not forget the big money-makers: such hyperscalers as Microsoft, AWS, and Google have immense purchasing power, thus they can demand their suppliers to provide them with better terms, as they rapidly gain momentum on AI.

To the point, the long-term narrative of AMD, which involves AI crunching, multi-cloud data centers, and edge stuff, is very impressive, yet all those risks should be tracked by investors. Such a high valuation of the stock provides the stock with a small leeway in terms of error.

Advanced Micro Devices has experienced all the large-scale technology trends in the recent past, including data-center AI and edge computing, and its most recent figures demonstrate that it is carrying them out so well. To long-term viewers, the attraction is obvious: AMD is no longer a niche producer of processors, it is one of the industry leaders in the production of server hardware and AI chips, and the revenue and profits are growing at double-digit percentages. Having an array of EPYC CPU, Instinct GPU, Ryzen/AI PC, and Versal adaptive chip, it can sail over a host of secular waves. And on top of that, AMD collaborated with the giants, Microsoft, Google, Oracle, OpenAI, and others and received major contracts such as OpenAI multi-gigawatt corporate implementation and Oracle AI supercluster, which confirms the fact that the industry has considered its advances.

Yet all that has driven the price of AMD significantly high and thus there can be very little buffer should the fortunes change. At this point, its valuation is quite giant by mere standards. The new long-term investors have to consider the volatile growth potential of AMD against the risk of economic cycles, supply chain nightmares, and stiff competition. Assuming the bullishness persists, then AMD might continue to realize that high-teens+ revenue growth and beat margins and the price would continue to increase. However, when the executions lose momentum or the economy loosens its grip, returns could disappoint if execution weakens.

Continue Reading