Co-designed x86 chips are happening, according to Intel's CEO

Intel Core i9-11900K review
Intel Core i9-11900K review (Image credit: Harish Jonnalagadda / Windows Central)

What you need to know

  • Intel has been asked how it will handle the growing Arm versus x86 dynamic.
  • Intel says it will allow for co-designed x86 chips.
  • The company says this will especially benefit its cloud service providing customers.

Plenty of companies are interested in Arm technology, ranging from Microsoft to NVIDIA. The latter is even attempting to acquire Arm in its entirety because of the value it sees. And that rampant interest is putting Intel in a bit of a hot seat.

For a long time, Intel was restrictive with its x86 licensing practices and not super big on the idea of hybrid designs built around a customer's preferences rather than its own. However, based on Seeking Alpha transcripts documenting the company's latest earnings call, it seems Intel CEO Pat Gelsinger is certain that Intel's most inflexible days are behind it (via PC Gamer).

In the earnings call, Gelsinger was asked if, in response to Arm's flexible IP licensing practices, Intel would consider allowing "hyperscale" customers to co-design their own products with its foundry services. Gelsinger's answer? Yes. He mentioned this would especially benefit cloud service providers, then framed it within the context of the market duel with Arm:

... we do believe that the ability for our customers to take advantage of x86 this way will be a meaningful shift in how people think about Arm versus x86. Because part of it was we weren't giving them the flexibility to design to comingle IP as I have described it.

So, there you have it. Intel's new $20 billion plan to fight the semiconductor crisis is just the tip of the iceberg when it comes to how the company plans on positioning itself for future relevance and dominance. How far Intel's willing to let customers go when it comes to hybrid designs remains to be seen, but one thing's for certain: It's further than the company's allowed in the past.

Robert Carnevale is the News Editor for Windows Central. He's a big fan of Kinect (it lives on in his heart), Sonic the Hedgehog, and the legendary intersection of those two titans, Sonic Free Riders. He is the author of Cold War 2395. Have a useful tip? Send it to robert.carnevale@futurenet.com.

8 Comments
  • So this will be the first time Intel has licensed their x86 architecture since the days of VIA and Cyrix, interesting. This will definitely place pressure on ARM and Qualcomm since Microsoft could now potentially have another option for their custom processor. Intel and Windows go together like peanut butter and jelly. But I wonder how exactly would this work out if AMD is the one who holds the patents for the x64 architecture, unless Intel were granted leeway in sublicensing or something.
  • I think in college I put a Cyrix 486 clone chip into the same motherboard that had housed an Intel for a speed boost without needing to replace the MoBo…. if I'm remembering correctly (not sure that I am). In today's world, I don't believe Intel needs any patents from AMD. Yes, they're competing and each is developing new IP, but I don't see Intel needing AMD's to create x86 chips.
  • If they don't get the rights from AMD then they would be forced to only make x86 CPUs instead of the existing lineup of x86-64 chips
  • If my memory is correct, Intel got more than leeway in licensing. I remember reading somewhere that the terms of Intel's license for the x86 instruction set require that any innovation to the instruction set (e.g., 64-bit extensions) be licensed back to Intel at no cost, while Intel retained the right to charge additional licensing fees for any innovations that they implemented (e.g., SSE, AVX, etc.). This is how Intel was able to release x64 hardware so quickly after AMD released the first x64 chips. AMD was forced to license their tech back to Intel for free and Intel's only cost was losing the first-mover's advantage but they also saved all of the R&D costs. I also believe any patents around the initial invention of x64 should have expired, though likely quite recently. The initial x64 instruction set was released in 1999. Patents in the US (since 1995) last 20 years from the date of filing and the filing date must be within 1 year of the public release of any information relating to invention for which the patent was filed (e.g., publication of the instruction set).
  • Too late. Too many investments were already made in Arm technology...
  • and too little for x86?
  • Depends on the uses and the required modifications.
    The cloud types rely strongly on virtual machines and sandboxes and Intel has done massively multicore non-x86 chips in times past. They had an 80-core 1 teraflop chip back in 2006, using 65nm tech. They didn't migrate the tech to x86 because of the focus on the x86 software. Remove that focus and license it for special uses (and newer processes) that don't need full x86 compatibility and they might be better suited to the task than ARM.
    They also have lots of experience building supercomputers so they have a lot of high end tech on tap.
    Don't write them off just yet.
  • ARM energy efficiency is more profitable than x86 in cloud services providers, thin and always connected laptops with all day battery life