You've likely heard the news by now: Intel and Nvidia have agreed to build a new range of products together. These co-developed chips will fuse Intel's x86 architecture with Nvidia's RTX graphics chiplets. The idea being, if you bring together both companies' tech, it's possible to deliver something more flexible, more desirable than either company can deliver on its own.
The benefits for the two companies are clear enough. Nvidia needs processors for GPU-accelerated server racks, and despite its best efforts to buy Arm, that fell through, dashing any hope of becoming a fully-integrated hardware company with it. Intel needs a helping hand in a few ways. The company is struggling with a falling share price and recently struck a deal with the US government to buy a share of the business, and getting an in with one of the richest companies on the planet isn't a bad get. Though supercharging its processors with more capable and powerful graphics chips has a few big benefits, too. To some degree in gaming, sure, but mostly AI.
"I think these are going to be revolutionary products. I know that all of us working on it are super excited about it," Nvidia CEO Jensen Huang said in a joint conference with Intel CEO Lip-Bu Tan.
"Nothing of its kind has ever been built before."
You might have something to say about that last statement if you work for AMD. The company is able to deliver competitive products across both CPU and GPU in-house. That's been embedded into its DNA since it bought ATI, and only improved on and amplified since with the growing success of its Zen architecture (with admittedly a few ups and downs in between—a story for another time), and delivered many generations of APU. If anyone has shown the capability to consistently deliver a combination of high-performance processors and graphics in a single chip or system-on-chip (SOC), it's AMD.
AMD is dominant in games consoles, through its semi-custom division and partnerships with Microsoft and Sony, for precisely this reason. It's been highly successful here, scoring these console manufacturers as repeat clients for multiple generations, and according to current expectations for at another generation to come. In the PC space, we have products such as the newly-released Ryzen AI Max 300 Series to point to, a processor line-up we've tested multiple times in various forms this year.
You might then consider this move from Intel and Nvidia to be pretty anti-AMD. I'm sure Dr. Lisa Su and other AMD executives have had a few emergency meetings to discuss yesterday's announcement—which saw AMD's share price dip sharply and then largely recover. Though there's definitely a positive way of framing it for the red team:
It just goes to show how much of an advantage AMD has if it plays its cards right.
No matter the promise of integration between Intel and Nvidia, no two competing companies will ever be able to collaborate quite as well as multiple departments within the same company. The sharing of GPU IP, if any such sharing occurs, between Intel and Nvidia would have to be pretty tight to keep either party out of the other's biggest secrets, especially as both companies sort of dabble in similar sectors (Nvidia with Arm partnerships, Intel with Arc GPUs).
AMD's advantage also extends deep into the technological stack. Let's take Intel's Foveros technology, for example. This is a way of lashing chiplets (tiles, in Intel speak) together onto a single interposer to operate as a single chip. We've seen this in action with various mobile architectures, Meteor Lake and Lunar Lake, but also since come into play with desktop chips. It's also the key to these future x86 RTX SOCs.
"Intel has the Foveros multi-technology packaging capability, and it's really enabling here," says Jensen. "And the reason for that is because, as we all know, Nvidia's GPU technology is based on TSMC foundry. And this is one of the extraordinary things that you can do, connecting Nvidia's chiplets with Intel CPUs in a multi-technology packaging capability and multi-process packaging technology."
AMD has similar chiplet technology. It's used chiplets across both of its Ryzen CPU and Radeon GPU products at various points throughout the past decade, even before Intel got around to it, and to huge success. Though the processor I'd point to here is the Ryzen AI Max+ 395 in the Framework Desktop, or to a slightly lesser degree, the Ryzen AI Max 390, a chip I recently tested in my Asus ROG Flow Z13 review. The more powerful of the two here, the 395, comes with desktop-grade components inside an ostensibly mobile part: that's 16 Zen 5 cores and 40 Radeon 3.5 CUs. All connected via an interconnect specially designed for this chip to deliver high bandwidth, low latency, and low power.
Not only does this sound a lot like what I'd expect a high-performance x86 RTX SOC might one-day deliver (noting here we don't actually know what these are yet, they're a long way off), with the exception of being AMD's x86 and GPU components instead, it also has those smaller details in interconnect, packaging and software from the same company. Now, you can argue whether there's always been a fluid relationship between AMD's many teams in the past, but in the future, especially one that relies more and more on AI, it could play an increasingly important role.
This is all about AI, right? The Ryzen AI Max 300-series wasn't really intended for gaming. I mean, it's decent enough for playing any number of modern games, but it's easily defeated in price/performance by a discrete GPU and a lower core-count chip. We've not rated it for gaming in either of the products we've used it in. No, these chips do best when loaded up with a significant amount of memory, allowing them to run AI models with a huge number of parameters, such as the new gpt-oss-120b from OpenAI or Alibaba's 253 billion-parameter model.
AI capabilities are seemingly the endgame for AMD here, as it builds out big bandwidth, big GPU, low-power chips. As much as it looks to be a target for Intel and Nvidia. If the future of AI is an increasing amount of local compute, and local compute powered by a GPU, rather than an NPU, then it's AMD that has the headstart in actually delivering that. Especially on mobile, which is a growing slice of the PC market. AMD also has the benefit of having more time to develop on these products over the next few years, whereas Intel and Nvidia's partnership is only in its nascent stage.
So, none of this spells doom for AMD to me. They say imitation is the highest form of flattery, right?
1. Best overall:
AMD Ryzen 7 9800X3D
2. Best budget:
AMD Ryzen 5 7600X
3. Best mid-range:
AMD Ryzen 7 9700X
4. Best high-end:
AMD Ryzen 9 9950X3D
5. Best AM4 upgrade:
AMD Ryzen 7 5700X3D
6. Best CPU graphics:
AMD Ryzen 7 8700G
You must confirm your public display name before commenting
Please logout and then login again, you will then be prompted to enter your display name.
