Arista Networks has crossed the $3 billion quarterly revenue mark for the first time, driven by sustained demand for AI networking infrastructure and a measurable improvement in the component supply chain. The company reported revenue of $3.036 billion for the quarter ended June 30, 2026, up 12.1% from the first quarter and 37.7% from the same period a year earlier. CEO Jayshree Ullal noted that Arista generated just $2.9 billion in total revenue during all of 2021, underscoring how sharply demand has increased as AI moved from lab experiments to production data centers.
“Customers see networking as the central nervous system for infrastructure from the client to campus to data and AI centers,” Ullal said. “Our AI fabrics momentum with Etherlink switches now exceeds 100 cumulative customers from the initial four to five customers I spoke of in 2024.”
According to the company, growth is now coming from every major part of the business. Back-end AI scale-out fabrics are expanding, front-end data center switching remains active, and adjacent campus and routing portfolios are also contributing to the revenue gains. Arista estimated that the broader “scale-across” switching and routing market could grow to between $15 billion and $20 billion by 2030, giving the company a larger arena in which to compete as service providers and enterprises upgrade their networks.
That demand has led Arista to raise its outlook. Ullal said the company is now projecting 40% annual growth, an incremental $2.1 billion over the $10.5 billion target established at its 2025 Analyst Day and $1.1 billion above the $11.5 billion projection it shared in May 2026.
The record quarter did not happen by accident. Just three months ago, Arista warned that memory, chip, and wafer shortages were creating rising costs and forcing customers to wait longer for networking equipment. President and COO Todd Nightingale said the company has spent the last six months attacking those bottlenecks.
“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 gen AI technologies,” Nightingale said. Memory supply is now secured for 2026, with visibility extending well into 2027 across DDR4, DDR5, and NAND devices. For printed circuit boards and optics, Arista says it can add capacity within a 12-month window and has strengthened commitments from key suppliers. The company is also managing inventory more tightly across thousands of component SKUs.
The work extends to newer infrastructure requirements as well. Nightingale said Arista has established a liquid-cooling supply chain that includes cold plate, quick disconnect, and tubing vendors with capacity agreements for next-generation AI technology. As AI clusters push power and thermal limits, liquid cooling is becoming a standard design element in high-performance network switches.
Ullal cautioned that the industry is not out of the woods. “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. But Arista is taking individually and specifically steps in the first half of this year that we believe will have results in the back half of this year.”
On the product side, Arista pointed to several EOS software features that are giving customers reasons to upgrade. President and CTO Kenneth Duda said the combination of rapid innovation and deployment at scale in AI networks is unlike anything he has seen in his career.
One highlighted technology, Smart System Upgrade, is designed to let network operators upgrade switch software without disrupting traffic. Frequent upgrades are becoming harder to avoid as AI workloads expose security vulnerabilities and create new tools for attackers. While many competing systems require a full reboot, Duda said Arista EOS handles the process seamlessly.
The second technology, Multipath Reliable Connection, addresses a performance problem in first-generation AI networks. Without it, each packet in an XPU-to-XPU flow must follow a single path; when two flows hash to the same path, they can run at half speed. MRC lets senders spread a single flow across many paths, with receivers reassembling data that arrives out of order. The result is better utilization of expensive XPUs.
The third area, Segment Routing over IPv6, is not a new concept, but Duda says its use to load-balance an AI fabric is a game changer. The sending device tags each packet with a stack of SRv6 segment IDs that determine the exact path through the fabric. Real-time congestion signaling then shifts packets away from hotspots.
“Because Arista EOS provides a single unified operating system, we support this SRv6 intelligence all the way from the scale-out fabric to the long-distance scale-across routing,” Duda said. That lets customers combine high performance with operational simplicity, he said.
In the optics segment, Arista is making a deliberate bet on open technology. Ullal said the majority of the AI networking market will continue to use pluggable optics and copper through 2028 or 2029. “I think there’s very much a philosophy there [of] copper if you can, optics if you must,” she said.
Copper remains practical for connections of two or three meters, especially inside racks, she said, while pluggable optics will handle longer distances. When it comes to co-packaged optics, which integrate optics closer to switching silicon, Arista does not want to see a fragmented market. “Arista is not a fan of five different proprietary implementations,” Ullal said. Instead, the company is working with partners on an open approach using socketed optical engines and pigtail fibers that can be fully pretested before deployment.
Ullal said open CPO will not happen overnight. Arista supports the idea of a true open interface that works across multiple vendors and switch configurations. Co-packaged optics and near-packaged optics are expected to enter trials in 2027, but they will remain a small portion of the market in the near term.
The company recently unveiled extended pluggable optics, or XPO, a form factor designed for high-speed optics. More than 100 optics module suppliers have joined a multi-source agreement to build and support XPO, giving Arista and its customers a more standardized path to the high-speed connections needed in AI clusters.
The combination of record revenue, improving supply, and expanding AI customer adoption suggests Arista has found a rhythm in the AI networking cycle. The company’s next challenge will be navigating the broader industry shortage that Ullal expects to last until 2028 while continuing to scale the technologies that customers need for the next wave of AI infrastructure.
Source:Network World News