With the Neoverse N1 core design, Arm proved its metal in the data center arena, spawning chips like Amazon’s Graviton2 and Ampere’s Altra.

And true to Arm’s heritage, this first generation of data center chips neatly balanced performance, core density, and efficiency to maximum effect. But what happens when efficiency is thrown out the window in favor of sheer unadulterated performance? You get Arm’s newest core design, Neoverse V1, according to Chris Bergey, SVP and GM of Arm’s infrastructure line.

“V1 is all about performance. How do we get the highest single threaded performance you can get,” said Bergey, who oversees Arm’s push in the high-performance computing, hyperscale, 5G, and edge arenas.

Announced last September, Neoverse V1 is designed to go toe to toe with Intel’s and AMD’s best data center chips. And unlike the first generation Neoverse N1 or the upcoming N2 variant, V1 prioritizes per-core performance.

“Looking back at Graviton2 … I think people weren’t necessarily surprised that we could create a performant core, what they were surprised about is that … AWS could put 64 of them in a single die and just have totally linear performance right across memory.”

According to Arm, V1 will deliver a 50% performance uplift compared to the previous generation while the more balanced N2 will deliver up to 40% higher performance.

“This is really the widest architecture we’ve ever shipped,” Bergey said. It’s a “pretty incredible uplift from a performance point of view”

However, V1 is more than a core design, Bergey explained. It’s paired with the ArmV9 architecture and will be the first implementation to support its new scalable vector extension, which was developed in collaboration with Fujitsu for the Arm-based Fugaku supercomputer.

“The goal was to show that N1 is not a one-trick pony. It was a disruptive entrance, but we know that we have a lot more to put on the table,” he said.

With the company’s second generation of CPU designs, Arm aims to deliver chips that are at least as performant as those offered by AMD and Intel, but a much more compelling value.

“There’s quite frankly a lot of demand. People are waking up and saying ‘wow that’s possible,’” Bergey said.

Marvel is now planning to deploy Neoverse N2 in its next-generation Octeon infrastructure chips used in equipment like switches, routers, secure gateways, firewalls, base stations, and smartNICs. And India's Ministry of Electronics and Information Technology announced plans to license the Neoverse V1 core design for its latest supercomputing project.

In the cloud, Neoverse N1-based chips are quickly gaining a foothold. Both Cloudflare and Oracle have deployed Ampere’s Altra CPUs in their respective content delivery networks and public clouds, while Amazon is already offering cloud instances running on its internally developed Graviton2 platform.

"We believe Arm is going to be everywhere – from edge to the cloud. We are seeing N1-based processors deliver consistent performance, scalability, and security that customers want from cloud infrastructure,” said Bev Crair, SVP, of Oracle Cloud Infrastructure compute. “Partnering with Ampere Computing and leading [independent software vendors], Oracle is making Arm server-side development a first-class, easy and cost effective solution.”

In China, Alibaba Cloud has begun trialing on Arm, and Tencent is making investments both in hardware and software to adopt Neoverse in the near future, Arm claims.