The battle for AI PCs has spread from CES 2026 to off-stage activities. Just a few days ago, AMD made a high-profile announcement about the graphics performance of its Ryzen AI Max 395+.Big winIntelCore Ultra Series 3 processors, codenamed Panther LakeFaced with this challenge, Intel clearly has no intention of swallowing it whole.
Tom Petersen, who previously worked on the NVIDIA GeForce team for over 14 years and is now a senior researcher at Intel, recently...When interviewed by foreign mediaThe attack was all-out, directly pointing out the fundamental problems with the design strategy of AMD's Ryzen AI Max processor, codenamed Strix Halo, emphasizing that it would be more practical to pair it with a discrete graphics card instead of going to great lengths to create an expensive "Big APU".
Intel: RDNA 3.5 is just old wine in new bottles; the Xe 3 graphics architecture is the new generation.
Tom Petersen offered a different interpretation of AMD's claimed performance superiority. He pointed out that AMD's currently lauded Ryzen AI Max series still uses an integrated graphics architecture based on RDNA 3.5. Essentially, it's just a mobile version of RDNA 3 with minor tweaks, belonging to the previous generation. In terms of energy efficiency and support for new technologies (such as ray tracing efficiency), it's already struggling to keep up with NVIDIA's RTX 40 series.
In contrast, Intel's Panther Lake processors utilize the new Xe 3 graphics architecture. Tom Petersen stated that Xe 3 has more modern advantages than RDNA 3.5 in terms of architecture design, performance per watt, and AI-added features. Therefore, for mainstream thin and light laptops, Panther Lake can provide a better experience with lower TDP (Thermal Design Power), rather than simply stacking more cores to achieve higher benchmark scores.
Is the "large integrated graphics" issue a false one? Intel admits it won't make such products.
The most interesting part of the interview was Intel's statement regarding its product roadmap. Tom Petersen stated that Intel currently has no plans to launch products like the AMD Strix Halo, which integrates a large-area, high-performance GPU.
Why? The reason is simple: cost and positioning.
Intel believes that the Ryzen AI Max chip design is extremely complex, resulting in high costs. Although it has a huge amount of unified memory (which is very beneficial for running AI models), in the gaming market, for the same budget, consumers who buy a laptop with a "standard processor + NVIDIA discrete graphics card" will often have a better gaming experience than those with integrated graphics.
In other words, Intel believes that AMD's strategy is to make chips into expensive monsters at the "mobile workstation" level, rather than gaming solutions suitable for the mass market.
Handheld Market: Following the Ryzen Z Model in Twitter Regulations?
Besides the laptop market, handheld gaming consoles are also a fiercely contested battleground. This is especially true given Intel's plans to use Panther Lake processors.Entering the handheld marketMarket sources suggest that Intel may not simply use the existing Core Ultra processor specifications directly, but will instead follow the model of AMD's Ryzen Z series and launch a special product line optimized for handheld power consumption and gaming needs, attempting to carve out a niche in a market dominated by competitors such as Steam Deck and Xbox ROG Ally.
Analysis of viewpoints
This war of words between Intel and AMD actually reflects the two companies' drastically different AI PC strategic philosophies.
• AMD's bet is "integration":With Strix Halo (Ryzen AI Max), AMD is attempting to replicate the success of Apple's M-series chips—utilizing ultra-large unified memory bandwidth to allow the CPU, GPU, and NPU to share resources. This is indeed uniquely attractive to developers or video creators who need to run large language models (LLMs), as it can solve the pain point of insufficient dedicated graphics memory.
• Intel's logic is "pragmatism": Tom Petersen's statement was quite "honest." When he said "it's better to pair it with a discrete graphics card," he was, to some extent, admitting that "NVIDIA is the leader in the high-performance graphics computing segment." Intel chose to focus on the performance of the CPU itself and the computing power of the NPU, and left the graphics burden to discrete graphics cards (whether it's its own Arc or its competitors' GeForce).
In my opinion:
• AMD wins in specific scenarios:If you're a creator who needs to edit videos and run AI models on an airplane, the Ryzen AI Max's high energy efficiency and unified memory are a good choice.
• Intel wins in the mainstream market:For most people who buy laptops to play League of Legends or watch Netflix, Panther Lake's low power consumption and battery life are probably more noticeable; while for hardcore gamers, Intel CPU + NVIDIA GPU is still the most stable combination.
While Intel's counterattack was sharp, it also indirectly exposed the fact that it cannot catch up with AMD in terms of "top-tier integrated graphics performance" for the time being, and can only defend itself through "product positioning." As for whether consumers will buy into the "large integrated graphics," that depends on whether the pricing of laptops equipped with Ryzen AI Max is "sweet" enough.
