Arista Networks outlined its latest innovations in high-speed networking along with its strategy to capitalize on artificial intelligence (AI) opportunities during its Q4 2023 earnings call on Feb. 12.

The company reported robust Q4 results with revenue of $1.5 billion, up 20.8% year over year. For the full year 2023, revenue hit $5.8 billion, a 33.8% increase. According to CEO Jayshree Ullal, performance was driven by momentum in both the cloud and enterprise sectors. Arista counts cloud titans Microsoft and Meta as major customers, each contributing over 10% of 2023 revenue. The company has also more than doubled enterprise revenue in the past three years.

While artificial intelligence (AI) is not a major contributor to Arista's financial results in 2023, Ullal expects it will be a big driver for the company's future growth.

“We are cautiously optimistic about achieving our AI revenue goal of at least $750 million in AI networking in 2025,” Ullal said during her company's earnings call.

How the Ethernet is being improved for the AI era

Arista is one of the founding members of the Ultra Ethernet Consortium (UEC), which got started in 2023 as a multi-stakeholder effort to improve Ethernet's capabilities for AI and high performance computing (HPC).

“Basically, AI at scale needs Ethernet at scale,” Ullal said. “AI workloads cannot tolerate the delays in the network because the job can only be completed after all flows are successfully delivered to the GPU clusters. All it takes is one culprit or worst-case link to throttle an entire AI workload.”

Ullal noted that among the innovations that UEC is working on is a technique known as packet spraying, which is a method to help accelerate every data flow to be able simultaneously access all parts of the destination.

Flexible ordering is another area that UEC is working on.

“Key to an AI job completion is the rapid and reliable bulk transfer with flexible ordering using Ethernet links to optimally balance AI-intensive operations, unlike the rigid ordering of InfiniBand,” Ullal said. “Arista is working closely with its leading vendors to achieve this.”

UEC is also working on ways to improve network congestion issues. Ullal noted that with AI networks, there's a common congestion problem whereby multiple uncoordinated senders can send traffic to the receiver simultaneously.

“Arista's platforms are purpose-built and designed to avoid these kinds of hotspots,” she said.

400G and 800G are core to Arista's strategy

A key part of Arista's strategy centers on leading the transition to higher network speeds.

Arista now has 800 customers adopting its 400G solutions, up from 600 in 2022. The company believes it holds the no. 1 market share in 100G, 200G and 400G switching.

Looking ahead, Arista is preparing pilots for 800G Ethernet with broad production expected in 2025.

“AI workloads are placing greater demands on Ethernet as they are both data and compute-intensive across thousands of processes today,” Ullal said.

It's not all about AI — campus and enterprise matter too

Ullal also emphasized Arista's continued focus on campus and enterprise deployments for connectivity as well as security.

“The post-pandemic campus is seeking network-as-a-service overlays and zero-trust network embedded in with high availability, observability, and consistency across our OS [operating system] and management domains,” she said. “We are also successfully deployed in many routing edge and peering use cases.”

At the foundation of Arista's networking gear is the company's EOS network operating system (OS). Over the course of 2023, the company had six major EOS updates that Ullal said provided over 600 new features. She also noted that the Arista WAN Routing system, which was released in March 2023, has been successfully providing organizations with encryption and WAN transit routing capabilities.

“While 2023's headline has been mostly about AI, we are pleased with the momentum of enterprise and provider customers as well,” Ullal said. “Arista continues to diversify its business globally with multiple use cases and verticals.”