Arista Networks has reached a significant financial milestone, delivering its first $3 billion quarter amid surging demand for AI networking infrastructure. The company reported revenue of $3.036 billion for the second quarter, representing a 12.1% increase from the first quarter and a 37.7% jump from the same period a year earlier.

CEO Jayshree Ullal said customers now view networking as the central nervous system of modern data center ecosystems. “Customers see networking as the central nervous system for infrastructure from the client to campus to data and AI centers,” she said. To put the latest results in perspective, Ullal noted that Arista brought in $2.9 billion for the entire 2021 fiscal year. One quarter in 2026 has now surpassed that full-year figure, a sign of how much the AI revolution has changed the networking business.

Arista was founded in 2004 with the goal of bringing a software-driven approach to enterprise and cloud networking. Over the past two decades, it has become one of the largest producers of high-performance data center switches and routers. Its EOS operating system was designed to run consistently across hardware generations and has become a staple in large-scale data centers operated by cloud providers and global enterprises. The early focus on programmability and automation is paying off in the artificial intelligence era, where network engineers must orchestrate thousands of switches to support massive training and inference workloads.

AI Fabric Momentum Accelerates

Ullal said Arista’s AI fabric momentum with Etherlink switches now exceeds 100 cumulative customers, up from the initial four or five customers she spoke of in 2024. “Our AI fabrics momentum with Etherlink switches now exceeds 100 cumulative customers,” she said. The company’s Etherlink product line is designed specifically for AI networking, offering high-density 400G and 800G interfaces, advanced congestion control, and support for multiple back-end network topologies.

Growth is happening across every part of Arista’s business, according to Ullal, from back-end AI fabrics to the core data-center front end, plus campus and routing products. Historically, corporate networks were built primarily for client-server North-South traffic. But AI model training is dominated by East-West traffic between tens of thousands of accelerators. That shift is reshaping how networks are designed and has made the switch fabric a critical performance element in the AI stack. Arista’s results are a signal that the traditional switch market remains robust even as specialized AI fabrics emerge.

Supply Chain Pressures Beginning to Ease

Supply chain problems were one of the biggest concerns Arista raised during its previous earnings report. The company cited ongoing shortages of memory, chips, and wafers, along with rising costs. That picture has improved, although Arista leadership remains cautious about how quickly the industry will fully recover.

Todd Nightingale, Arista’s president and chief operating officer, said the company has spent the last six months working to align its supply chain with growing demand. “We’ve secured multiyear agreements with leading vendors of strategic components, qualified new suppliers in key areas to limit risk, and built out supply chains for next generation AI technologies,” Nightingale said. He also reported that Arista’s memory supply for 2026 is now secured and the company has extended visibility into 2027 across DDR4, DDR5 and NAND memory.

For PCs and optics, Arista can now build capacity within a 12-month window, with strengthened commitments from key suppliers. The company has improved lead times and inventory management across thousands of component SKUs, with better sub-component pipelining and multi-sourcing flexibility. Arista also established a liquid-cooling supply chain capable of supporting next-generation AI infrastructure. That includes cold plates, quick disconnects, and tubing vendors under capacity agreements.

Ullal cautioned, however, that the industry-wide supply situation will not be fixed overnight. “I don’t want you to believe that suddenly we waved a magic wand and all our problems have gone away,” she said. “The industry is going to have a two-year problem [with memory and other silicon availability challenges], and I don’t think we get out of it as an industry until 2028.” Arista is taking targeted steps in the first half of the year, with results expected to show up in the second half.

EOS Innovations: SSU, MRC and SRv6

Kenneth Duda, Arista’s president and CTO, said he has never witnessed this combination of rapid innovation and large-scale deployment in networking. He highlighted three EOS technologies that are helping to set Arista’s gear apart: Smart System Upgrade (SSU), Multipath Reliable Connection (MRC), and Segment Routing over IPv6 (SRv6).

SSU enables switching software upgrades without any data plane disruption. Frequent security patches are becoming unavoidable, especially as AI exposes more vulnerabilities and generates new attack tools. Many competing systems require a full reboot to apply patches, which creates expensive network downtime. Arista EOS uses an in-service software upgrade approach to avoid that problem.

MRC addresses a fundamental limitation of first-generation AI networks, where every packet on an XPU-to-XPU flow has to follow the same path. If two flows hash to the same link, each runs at only half speed. MRC allows a single flow to be sprayed across many paths through the switch fabric

Source: Network World News