Arista Networks has crossed its latest revenue milestone, reporting the company’s first $3 billion quarter as AI networking demand remains strong and component supply pressures begin to ease. The company posted revenue of $3.036 billion for the second quarter, a 12.1% increase from the previous quarter and a 37.7% jump from the same period a year earlier. The achievement is even more striking when measured against recent history: just five years ago, Arista recorded $2.9 billion in revenue for all of 2021.

Jayshree Ullal, Arista’s chief executive officer, told investors during the company’s second-quarter earnings call that networking has become the central nervous system for modern infrastructure, spanning client, campus, data center, and AI environments. She noted that Arista’s AI fabric momentum, built around its Etherlink switching products, now exceeds 100 cumulative customers. That represents a significant expansion from the initial four or five customers she described in 2024.

“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.”

Ullal said the company is seeing growth across all major product categories, from back-end AI fabrics to data center front-end switching, and from campus switching to routing. The overall scale-across switching and routing market is expected to reach between $15 billion and $20 billion by 2030, according to Arista’s estimates, positioning the vendor for continued expansion.

Arista also revised its growth forecast upward. “We are now projecting 40% annual growth,” Ullal said, adding that the new projection represents an incremental $2.1 billion over the company’s 2025 Analyst Day goal of $10.5 billion and an incremental $1.1 billion over its May 2026 projection of $11.5 billion.

Supply chain pressures beginning to ease

Arista has had to manage a difficult supply environment over the past year. The company previously warned that shortages of memory, chips, and wafers were pushing up costs and limiting its ability to meet customer demand. While those pressures have not disappeared, the company says it has made measurable progress.

Todd Nightingale, Arista’s president and chief operating officer, said the company has spent the past six months improving its supply chain to match rising product 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 gen AI technologies,” Nightingale said.

Nightingale also highlighted stronger collaboration with silicon suppliers. Memory supply has been secured for 2026, and Arista has extended visibility into 2027 across DDR4, DDR5, and NAND memory. The company has expanded its vendor qualification programs and added optionality to create a more resilient supply base. For printed circuit boards and optics, Arista says it can now build capacity within a 12-month window and has strengthened its commitments with key suppliers.

“We’ve improved our lead times and inventory management of thousands of component SKUs, improving sub-component pipelining and multi-sourcing, and providing increased flexibility with reduced inventory risk,” Nightingale said.

The company has also been preparing for next-generation AI systems that require advanced cooling. Arista has established a liquid-cooling supply chain with cold plate, quick disconnect, and tubing vendors, complete with capacity agreements for new AI technology.

“By focusing on vendor stability and diversity, risk mitigation, and predictable delivery terms, innovation for new AI products, and capacity across our factories, we are making significant improvements and significant capacity increases across our supply chain,” Nightingale added.

Ullal offered a more cautious long-term outlook, warning that industry-wide supply constraints will not disappear quickly. “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, 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.”

EOS innovations

Arista’s technology leadership was a central topic during the earnings call. Kenneth Duda, Arista’s president and chief technology officer, said he has never witnessed the combination of rapid innovation and scale deployment that is now unfolding in AI networks. He pointed to three EOS capabilities that are helping differentiate Arista systems: Smart System Upgrade, or SSU; Multipath Reliable Connection, or MRC; and Segment Routing, also known as SRv6.

SSU is designed to let network operators upgrade switch software without any disruption. Duda argued that frequent upgrades have become a hard reality in modern networks, especially as AI both exposes security vulnerabilities and creates tools that can exploit them. Many competing systems require a full reboot when a security patch is applied, resulting in expensive downtime, while Arista EOS aims to handle those upgrades seamlessly.

MRC addresses a significant performance issue in first-generation AI networks. In those networks, every packet on an XPU-to-XPU flow is forced to take the same path through the fabric. If two flows happen to hash to the same path, they can collide and reduce throughput. MRC allows senders to spray a single flow across multiple paths, while receivers reassemble any data that arrives out of order. That approach helps eliminate the performance penalties caused by fabric collisions.

Duda also discussed SRv6, which is not new but is being used in a new way to load balance AI fabrics. With SRv6, the sender tags each packet with a stack of segment IDs that dictate the exact path the packet will follow. The system can then use real-time congestion signaling to shift packets away from hot spots in the network.

“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. “It gives our customers the combination of high-quality top performance and operational simplicity that Arista is known for.”

The optics

Optical connection technology continues to grow in importance for AI networking, but Arista does not expect a sudden industry transition to co-packaged optics. Ullal said Arista believes most of the market will continue to rely on pluggable optics and direct-attach copper through 2028 or 2029. “There’s very much a philosophy there of copper if you can, optics if you must,” she said.

She predicted that copper will remain common for short distances of two to three meters, especially inside racks, while pluggable optics will remain central for longer connections. At the same time, the industry has seen several proprietary approaches to co-packaged optics emerge. Arista, she said, does not want to embrace a fragmented ecosystem.

“Arista is not a fan of five different proprietary implementations,” Ullal said. The company has been working with partners to develop an open co-packaged optics approach. The idea is to use socketed optical engines and pigtail fibers that can be fully pretested, whether they are soldered onto the board or placed nearby. Arista wants to establish a truly open interface that works with multiple vendors and multiple switch configurations.

Ullal said Arista supports open, socketed optical engines rather than multiple proprietary solutions. Co-packaged optics and near packaged optics are expected to enter trials in 2027, but they will remain a small part of the overall market in the near term, she said.

Arista recently introduced extended pluggable optics, or XPO, a form factor designed specifically for high-speed optical connections. The company has also assembled more than 100 optics module suppliers as part of a multi-source agreement to build and support the XPO ecosystem. That effort reflects Arista’s broader strategy of promoting open standards and supplier diversity while still delivering the performance AI workloads demand.

Source: Network World News