Arista Networks has reached a significant milestone, reporting its first-ever $3 billion quarter as demand for AI networking infrastructure continues to surge. The company posted 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. The results underscore how deeply AI workloads are now reshaping data center, campus, and wide-area networking requirements.

Arista's CEO Jayshree Ullal highlighted the company's rapid ascent, noting that just five years ago Arista generated roughly $2.9 billion in revenue for the entire 2021 fiscal year. "Customers see networking as the central nervous system for infrastructure from the client to campus to data and AI centers," Ullal said during the earnings call. The company also reported that its AI fabrics momentum with Etherlink switches now exceeds 100 cumulative customers, a sharp rise from the initial four or five customers Ullal spoke of in 2024.

Ullal said the company is seeing growth in virtually every sector, from back-end AI fabrics to the core data-center front end, and extending into adjacent campus and routing businesses. The scale-across switching and routing market is forecast to reach between $15 billion and $20 billion by 2030, positioning Arista well for continued expansion, she added.

Financial performance and aggressive growth forecast

The $3.036 billion quarterly figure represents more than just a milestone; it signals a fundamental shift in how enterprises and cloud providers are investing in networking. According to Arista's management, the company is now projecting 40% annual growth, which equates to an incremental $2.1 billion over its 2025 Analyst Day goal of $10.5 billion. That projection also marks an increase of $1.1 billion over the $11.5 billion target the company laid out in May 2026.

"We are now projecting 40% annual growth, and the market opportunity is broadening," Ullal said. The company's confidence comes from the sustained buildout of AI infrastructure, which requires high-bandwidth, low-latency networks capable of supporting massive distributed training and inference workloads. Unlike traditional data center traffic, AI workloads generate east-west traffic patterns that demand highly resilient, lossless fabrics.

Arista's financial results also reflect its ability to capture share in a market increasingly dominated by a few large players. The company has long competed against Cisco, Juniper, and Huawei, but its early focus on software-driven networking and telemetry has given it an edge in AI-era deployments. Analysts note that Arista's EOS operating system, which runs consistently across switching and routing platforms, reduces operational complexity and appeals to organizations running large-scale AI clusters.

AI networking momentum with Etherlink and beyond

Arista's Etherlink switch portfolio has become a cornerstone of its AI strategy. The company said it now has more than 100 cumulative customers using Etherlink platforms for AI fabric deployments. These customers range from large cloud service providers to enterprises building private AI infrastructure. "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 noted.

The growth is not confined to a single vertical. Arista is seeing adoption across financial services, healthcare, higher education, and the public sector, as organizations seek to build high-performance networks for AI model training, inference, and data processing. The shift toward AI-ready networking has also driven demand for higher-speed Ethernet, including 400G and 800G, as well as for technologies that improve utilization of expensive GPUs and XPUs.

Ullal emphasized that networking is now viewed as a strategic enabler of AI success rather than a commodity layer. "AI workloads are creating unprecedented pressure on network infrastructure," she said. "Arista's ability to deliver scalable, programmable, and observable networks is why customers are choosing us for their most demanding environments."

Supply chain pressures beginning to ease

One of the key questions hanging over Arista's recent quarters has been supply chain tightness. Last quarter, the company warned that shortages of memory, chips, and wafers were driving up costs and constraining its ability to meet customer demand. This quarter, management reported measurable improvements, thanks to a series of strategic actions taken over the past six months.

"Arista has spent the last six months improving our supply chain to meet growing product demand, and we're seeing significant improvements," said Todd Nightingale, president and COO of Arista. "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 elaborated that relationships with strategic silicon vendors remain strong, with close collaboration on both supply chain and technical engagements. Memory supply for 2026 has been secured, and the company now has extended visibility well into 2027 across DDR4, DDR5, and NAND memory. "We've also increased our resiliency through optionality and expanded vendor qualification," he added. For printed circuit boards (PCBs) and optics, Arista can now build capacity within a 12-month window and has strengthened commitments from key suppliers.

The company has also improved lead times and inventory management for thousands of component SKUs. "We're improving sub-component pipelining and multi-sourcing, and providing increased flexibility with reduced inventory risk," Nightingale said. In addition, Arista has established a liquid-cooling supply chain designed to support the next generation of AI infrastructure. This includes cold plate, quick disconnect, and tubing vendors with capacity agreements for cutting-edge AI technology.

Ullal, however, injected a note of caution. "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."

EOS innovations: SSU, MRC, and SRv6

On the technology front, Arista's president and CTO Kenneth Duda highlighted three innovations within the company's EOS operating system that are helping differentiate its platforms in AI networks: Smart System Upgrade (SSU), Multipath Reliable Connection (MRC), and Segment Routing (SRv6).

SSU in an automated software upgrade feature designed to eliminate network disruption during OS updates. "Frequent upgrades are a hard reality today, especially as AI both uncovers security vulnerabilities and creates tools to exploit them," Duda said. "While many competing systems require a full reboot to address these issues, Arista EOS handles these upgrades seamlessly." This capability is particularly valuable in large AI clusters where downtime can stall expensive training runs and delay product development.

MRC addresses the problem of fabric cache collisions in AI networks. In first-generation AI networks, every packet on an XPU-to-XPU flow typically takes the same path. "That means if two flows hash to the same path, they both run at half speed," Duda explained. "MRC enables senders to spray a single flow across many paths through the fabric, where receivers reassemble any data that arrives out of order, eliminating the performance hit from fabric cache collisions." This capability is critical to maximizing the utilization of expensive accelerators, which often sit idle while waiting for data.

SRV6, a form of segment routing over IPv6, provides the control plane intelligence needed to direct traffic dynamically across the fabric. "It's not new, but using it to load balance an AI fabric—that's the game changer," Duda said. "The sender tags each packet with a stack of SRV6 segment IDs dictating the exact path the packet will take. The system then uses real-time congestion signaling to dynamically shift packets away from hotspots." This approach allows Arista to offer both high performance and operational simplicity.

Because Arista EOS is a single unified operating system running across the entire portfolio, the company can support SRv6 intelligence from the scale-out fabric to long-distance scale-across routing. "It gives our customers the combination of high-quality top performance and operational simplicity that Arista is known for," Duda added.

Optics and copper: a balanced approach

Optical connection technology continues to move into the AI networking environment, but Arista believes pluggable optics and copper will remain the dominant connectivity options through 2028-2029. Ullal expressed the company's philosophy succinctly: "Copper if you can, optics if you must."

Inside a rack, where distances are only two or three meters, copper cabling remains cost-effective and energy-efficient. For longer distances, pluggable optics are necessary. "I think you're going to see a lot of copper at those short distances, well within a rack," Ullal said. The company is less enthusiastic about proprietary implementations of co-packaged optics (CPO), where optical engines are integrated directly into the switch package. "Arista is not a fan of five different proprietary implementations," she stated.

Instead, Arista has been developing an open approach to co-packaged optics. "We don't think open CPO is going to happen overnight, but the idea here is to use socketed optical engines, pigtail fibers, and allow these modules to be fully pretested," Ullal explained. "And whether they're soldered on the board or nearby, the idea is to have a truly open interface that can operate with multiple vendors and multiple switch configurations."

Arista's support for open, socketed optical engines rather than multiple proprietary solutions is intended to give customers greater flexibility and reduce vendor lock-in. CPO and near-packaged optics (NPO) are expected to enter trials in 2027, but Ullal indicated they will remain a small portion of the market near term. In the meantime, Arista recently unveiled extended pluggable optics (XPO), a form factor designed specifically for high-speed optics. The company assembled more than 100 optics module suppliers as part of a multi-source agreement to build and support XPO, ensuring a competitive ecosystem and supply resilience.

Industry outlook and the road ahead

The AI networking boom shows no signs of slowing, and Arista is positioning itself to capitalize on the multiyear growth cycle. The company's raised guidance reflects confidence that its Ethernet-based AI fabrics will continue to displace older InfiniBand deployments, which have historically been favored for high-performance computing but lack the openness and scalability of Ethernet.

Ullal also pointed to the broader expansion of networking use cases, from campus and branch to data center and AI. "The scale-across switching and routing market is forecast to hit between $15 billion and $20 billion by 2030," she said. This includes not only high-speed data center switches but also the routers and aggregation devices that connect distributed AI infrastructure across multiple sites.

As AI models grow larger and training data sets expand, networks must evolve to handle unprecedented bandwidth demands. Arista's investments in supply chain resilience, software innovation, and open optical ecosystems are aimed squarely at that challenge. While supply chain constraints won't disappear overnight, the company's proactive steps and long-term vendor agreements provide a measure of stability that many competitors lack.

The road to 2028 will likely bring continued volatility in component availability, but Arista's record quarterly revenue and robust pipeline suggest that the company is managing the current environment effectively while positioning itself for the next wave of AI-driven networking growth.

Source: Network World News