plus69free - Fotolia
Intel quarterly results reveal tactics to address ARM in the datacentre
The chipmaker aims to make use of standard chip libraries and offer its customers different ways to package and optimise Intel silicon
Intel has reported revenue of $18.6bn for the first quarter of 2021, which it said was led by strength in its PC business, as well as the first signs of recovery in the enterprise and government portion of its datacentre business.
These signs of recovery have meant the company’s datacentre business delivered revenue of $5.6bn, down 20% year-over-year. Intel also said the datacentre business experienced a drop in operating income to $2.2bn year-over-year due to the lower revenue, transition to 10-nanometer and increased R&D investment in its Xeon roadmap.
In a transcript of the earnings call posted on the Seeking Alpha financial blogging site, the company discussed the recent news about its foundries. In March, Intel announced it would be expanding its manufacturing capacity, beginning with about $20bn of investment to build two fabrication plants in Arizona. The company also announced Intel plans to become a major provider of foundry capacity in the US and Europe to serve its customers globally.
When asked about the competition coming from ARM designs powering datacentre computing and whether Intel would consider licensing its x86 technology in a similar way to ARM to entice hyperscale cloud providers, CEO Pat Gelsinger said: “X86 cores will be available on our foundry services and available for people to design with them, and it will include the cloud service providers to take advantage of that.
“So, the simple answer is yes. And we 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.”
He said this will enable Intel’s customers to design their own x86-based chips, in a similar way to how ARM licensees build processors based on the ARM chip.
Gelsinger admitted that Intel has previously not catered particularly well for organisations that wanted to do unique design work. “They didn’t have a good x86 choice,” he said. “We gave them our standard products, which have lots of capabilities but particularly for the cloud.”
Read more about datacentre chips
- CPUs have evolved to meet ever-increasing technology demands. We look at the way performance and power characteristics have shifted in recent years.
- Paul Finch, CEO of Harlow-based colocation provider Kao Data, sets out how datacentre designs are having to change to accommodate evolving chip densities.
But other optimisations for networking or memory hierarchy were not possible with the standard x86 server chips. This has now changed, said Gelsigner.
“We are saying absolutely come on in and we are opening the doors of our IP, the doors of our leading process and packaging technologies, to be able to say let’s do this together or let them do their own designs in our foundry as well,” he said. “It’s a very powerful strategy that I think will be a meaningful shift.”
Gelsinger said the company was also seeing “extreme interest” in Intel packaging technologies. As an example, he said customers may be looking to integrate a tile from Samsung or TSMC with a tile from Intel on a single semiconductor device.
Intel chief financial officer George Davis suggested Intel’s future roadmaps would evolve to encompass standard semiconductor libraries.
“We have said over time that, it’s very clear that we are going to be adopting more and more of the industry ecosystem IPs and libraries, because it makes sense to do it,” he said during the earnings call. “And it makes sense in terms of cycle time, in terms of taking advantage of capabilities that already exist out there.”