“Vigorous competitors” AMD and Intel have been slow to the Arm party – where cloud hyperscalers and the likes of Apple have made major headway on their own chip designs – and they have been slow and outflanked by NVIDIA when it comes to the cash-cow GPU and AI software party.
Now they’ve teamed up to lead an “x86 Ecosystem Advisory Group” to deliver “a more unified set of instructions and architectural interfaces” for the still dominant server chip architecture; a move that comes a year after Intel proposed a range of deep changes [pdf] to its x86-64 architecture.
Joining them are Broadcom, Dell, Google, HPE, HP, Lenovo, Meta, Microsoft, Oracle, Red Hat, Linus Torvalds and Epic's Tim Sweeney. AMDs Lisa Su said they will “ensure that the x86 architecture continues evolving as the compute platform of choice for both developers and customers.”
See also: Intel bloodbath: 15,000 jobs slashed, dividend frozen, shares tank
(The changing face of enterprise compute is a unique threat as the cloud providers spend growing sums on overhauling their data centres with GPU-centric racks for AI and Arm-centric racks for traditional CPU workloads – in 2025 cloud CapEx will “equate to the entire Apollo Space program in real terms” as Morgan Stanley noted in a report this month.)
The group is being launched as ASML, manufacturer of the world’s most advanced chipmaking machines saw €50 billion erased from its market cap, with its CEO telling [pdf] investors that "anything related to AI is strong, but other than that there are limited capacity additions..."
x86 ecosystem advisory group: The Big Idea?
So… what is this x86 Ecosystem Advisory Group planning?
Pat Gelsinger explained: “It’s… agreeing on what the system interfaces, application interfaces need to be because often we had differences between us that were slowing, not accelerating, the ecosystem.
“So it's really about how do we collaborate on… what the interfaces need to be at the system level, how we have operating system and system level support for that, but then also how it embodies itself into the application stack” he said October 15, with the example of “things like AVX and AMX.”
AVX and AMX are instruction set architectures that extend the capabilities of x86 processors. Gelsinger told Six Five Media: “It's not, not like one is good or bad, but it is saying, ‘hey, when we're not aligned, we don't accelerate people's ability to bring applications onto the x86 architecture.”
See also: Warren Buffett’s GEICO repatriates work from the cloud, continues ambitious infrastructure overhaul
Dr Lisa Su added in a joint interview: “The truth is, you know, we hear a lot from customers that they appreciate x86, they appreciate all of the technology, but, boy, could AMD and Intel make it easier for them.”
Google Cloud CEO Thomas Kurian added that by “simplifying and standardizing across the x86 ecosystem, we can unlock new levels of performance, efficiency, and ease of use, ultimately accelerating the development and adoption of cutting-edge technologies.”
So when might this group actually, y’know, deliver something?
Gelsinger told Six Five: “I think [we are initially] just sending this message to the ecosystem, I think people will all of a sudden pause and say, huh, can I go partner with them, right? Can I go look at the x86 through a more forward-looking lens than I had before? We're already having those conversations with some of the cloud guys. So say ‘I can influence this and think about, you know, how I shape the x86, whether that's at the application level, whether that's at custom product levels, whether that's at the chiplet level, whether that's, you know, new use cases…”
