4nm will have a hard time competing with Intel 18A and TSMC 3nm.
Last week, AMD presented its AI, data center, and consumer plans for the next few years to financial and industry analysts. Despite the huge success in the data center market, partial success in AI, and a solid roadmap ahead, I wanted to share a shortcoming on the consumer roadmap, especially concerning 2026.
Jack Huynh senior vice president and general manager of the Computing and Graphics Group and his team are walking in the right direction; the X3D desktop was a great win for Team Red. However, instant success across some crucial parts of the consumer business, especially consumer and commercial notebooks, is unlikely to materialize in 2026. The company will need more time. Here is why.
No Zen 6 in 2026
Most of the leaked rumors that appeared in the media regarding the consumer business ended up being true. In 2026, AMD plans to launch the Gorgon platform, a refresh of its Strix Point series. The platform will still use a Zen 5-based processor, and I will focus on the mobile client and consumer part of the market.
In 2027, AMD will finally have Medusa, a Zen 6 platform promising good performance per watt and improved NPU capability. Let’s first look at 2026 and Gorgon.
Gorgon Unlikely to Catch Up with Panther Lake and Snapdragon X Elite Gen 2
A shortcoming of AMD’s client strategies is the lack of a power-efficient solution in notebooks. Strix Point and Strix Halo were good products, especially the former. It secured design wins for AMD, and those notebooks were quite decent, but not as good as the increased competition from Intel with Lunar Lake (Core Ultra 200) and the Snapdragon X Elite series.
AMD is suffering from higher thermals and lower performance per watt simply because Zen 5 and all of its predecessors were designed to win in the data center. Engineers close to the matter keep sharing with us that the Infinity Fabric consumes too much power for an efficient design in the mobile space. AMD was also lacking a strong smaller core strategy.
Forrest Norrod’s team in the Data Center Group is thriving because of the data center-first, wide, and powerful core strategy. The Consumer and Commercial Mobile Business Unit has to live with Zen 5’s shortcomings.
The Zen 5 core is simply not great for mobile, and while the smaller Zen 5c core did improve the odds, the Strix Point platform didn’t move the needle. It had the impossible job of unsuccessfully fighting with Lunar Lake (Core Ultra 200), which simply stole the show in 2025. Having Qualcomm coming in with the Snapdragon X Elite—offering incredible performance per watt, battery life, and efficiency—put AMD in a difficult position. AMD only had the Zen 4c core as of May 2023, launching in the Phoenix / Phoenix 2 platform.
The Core Ultra 200 series (Lunar Lake) was a ground-up design focused on efficiency and the successful execution of the Mac Book Air’s thin-and-light design philosophy. Intel was even beating AMD in GPU performance.
Lunar Lake was on 3nm, while AMD and Qualcomm stayed with the more cost-aware but performance-limiting 4nm node.
Strix Halo, the PR Chip
I saw the ASUS ROG Flow Z13 tablet design and a few from HP with Strix Halo at CES 2025, and I could not shake the feeling that launching with only three designs indicated small interest in the biggest GPU ever put on a mobile chip. Personally, I loved the idea of a big chiplet-based CPU + GPU combo and saw tremendous benefits in the AIPC (AI PC) market, just not in gaming, as ASUS was implying. That particular product was running hot and had too small a screen for my gaming taste. Pricing it at $2500 / €2500 didn’t help.
It was clear that a platform with unified memory and a large GPU would do well in what we segment as a Workstation AIPC. Some editors who ended up using this design as a daily driver complained about 6-hour battery life in a standard Windows environment on the HP OMEN MAX design.
This is half the battery life that most competitors are offering, whether it’s the Lunar Lake Core 200 with 15+ hours (even the 15.6-inch Yoga Slim 7 15ILL9), Arrow Lake H with 12-ish hours (16-inch Lenovo Yoga designs), or the ASUS 14-inch Snapdragon X Elite with closer to 20 hours in Windows scenarios. Machines such as the Dell XPS 13 and 14-inch ASUS Zenbook 14 were also providing 20-ish hours with 120 Hz OLED displays.
We ended up calling Strix Halo a “PR chip” because despite its excellent GPU and AI performance advantages, the platform failed to materialize as a successful product with OEMs. Strix Halo is currently getting a second life with Chinese OEMs in the small PC environment, but I don’t see any major OEM jumping on it anymore. In case they do a refresh with Gorgon, let’s hope that one does better. The reality is that mobiles drive most commercial and consumer business sales these days.
I didn’t catch that Kraken got significant traction, either.
High-End Desktop is a Win
Success in high-end desktop with the 9800X3D and 9950X3D kept the gamers happy, capturing significant parts of this popular market. This verifies that the Zen 5 core paired with a lot of cache was the right approach. Again, the data center-first approach paid off in this segment.
The 101 of CPU design, over decades of meeting with engineers, suggests that doubling the cache size on a design alone can bring around a 10 percent performance increase. It just makes the core significantly more expensive.
Gorgon is the Refresh
The year 2026 for AMD is the year of the refresh. A refresh means that AMD is bringing a 4nm core to fight with Intel 18A, a process that is 10%+ better than Intel’s 3nm design. AMD’s Gorgon is supposed to fight with the Snapdragon X Elite Gen 2, another 3nm product. I will let you draw your own conclusion about who has a better chance of achieving superior battery life and performance per watt: a 4nm part or two competitors that have smaller and more power-efficient transistors.
Panther Lake is bringing better graphics to the thin-and-light segment, and having 16 cores with a great GPU in a thin-and-light platform is simply overwhelming OEMs’ attention. With the Gorgon Point refresh platform, we don’t see a clear possibility of doing well in the market.
AMD will win some designs back from Qualcomm, but we don’t see much interest in Gorgon over Panther Lake / Lunar Lake designs.
2027
AMD shared very few details of the 2027 products codenamed Medusa. Medusa will bring another decent desktop part, but without redesigning the core and the Infinity Fabric, I have serious doubts that it can significantly improve the battery life and performance-per-watt metrics, even with the 2027 platform.
The design is well underway, and I am sure that AMD is aware of its shortcomings, but it remains to be seen if they plan to seriously address them. Even in 2027, Medusa is supposed to fight with Nova Lake, and this platform shows a lot of promise in both desktop and mobile.
Conclusion
AMD plans to grow to 40 percent in the consumer market. This is possible over time, but it will most likely happen in less profitable market segments that might hurt margins.
AMD indicated that there will be an increased number of design wins, but let’s see if that translates to any meaningful profit numbers. From where we are standing right now, an increased number of Gorgon designs almost feels like a fail-safe mechanism that is in place to potentially fill supply constraints of Panther Lake designs. Gorgon will likely succeed in filling that gap.
Panther Lake (Core Ultra 300 series) is indicating a high demand due to its dominant performance and efficiency over 2025 designs.
2027, the year when AMD finally makes Zen 6 available, is likely to improve AMD’s ability to compete. The only concern I have is that it might still be lacking what it needs to compete with Intel Nova Lake and the Qualcomm solution on performance per watt and battery life. The odds for the competition are certainly better than with the 2026 Gorgon refresh.
Disclaimer
ACAnalysis operates as an independent analytical service and works with multiple clients across various industries.
The content, analysis, and views expressed in this report are based on our proprietary research and methodology. While we strive to maintain objectivity and impartiality, some of the companies, organizations, or individuals mentioned or referenced within this report may be current or former clients of ACAnalysis.com.
The inclusion of any entity in this report, whether a client or not, does not imply endorsement, and our analysis is not influenced by the existence of a client relationship.
All opinions are those of ACAnalysis.com and are subject to change without notice.
