Arista Networks Raises 2026 Guidance to $12.6 Billion as Purchase Commitments Nearly Triple to $9.7 Billion
Q2 2026 earnings call, reported August 4, 2026
Arista Networks delivered its first-ever $3 billion revenue quarter and used the results to raise its full-year 2026 guidance for the third time this year, now targeting $12.6 billion in revenue, or 40% annual growth. That figure sits $2.1 billion above the company's original Analyst Day target of $10.5 billion and $1.1 billion above the raised guidance issued in May. CEO Jayshree Ullal framed the moment in stark historical terms: "Arista is experiencing significant demand to achieve our first $3 billion quarter in revenue. To put this in perspective, just 5 years ago, our entire year was $2.9 billion in 2021."
Total revenue for the quarter came in just over $3 billion, up 37.7% year-over-year and above the $2.8 billion guide, according to CFO Chantelle Breithaupt. Operating income hit $1.5 billion, a 49.9% margin, and diluted EPS rose 39.7% to $1.02. Guidance for the third quarter calls for approximately $3.3 billion in revenue with operating margin between 48% and 49%.
Supply Chain Fix Is the Real Story Behind the Raise
The most consequential disclosure on the call wasn't the topline number itself but the mechanics behind it. Arista's multiyear purchase commitments have nearly tripled in a year, from $3.6 billion to approximately $9.7 billion by the end of Q2 2026, up from $8.9 billion just one quarter earlier. This is the clearest signal yet that Arista is buying its way out of the component shortages that have constrained shipments over the past two quarters.
Ullal was unusually direct that the fix is neither instant nor complete. "I don't want you to believe that suddenly we waved the magic wand and all our problems have gone away. The industry is going to have a 2-year problem. I don't think we get out of it as an industry until 2028." That is a materially specific and sobering timeline from a management team that has otherwise sounded increasingly confident. COO Todd Nightingale detailed the operational response: memory supply is now secured through 2026 with visibility into 2027 across DDR4, DDR5 and NAND; PCB and optics capacity can now be built within a 12-month window; and the company has stood up an entirely new liquid cooling supply chain, including cold plate, quick disconnect and tubing vendors, to support next-generation AI infrastructure. Arista also hired Eugenia Corrales, a 35-year manufacturing and supply chain veteran, to lead the effort, and now operates three contract manufacturers and three distribution facilities across the U.S., Asia and Mexico.
Where the Extra $1.1 Billion Is Actually Coming From
Analysts pressed repeatedly on the composition of the raised guide, since Arista left both its AI target of at least $3.6 billion and its campus target of at least $1.25 billion technically "unchanged" even as the total guide rose sharply. Ullal's answer, in response to UBS analyst David Vogt, was candid about the uncertainty: "If you had to ask me whether our AI number or campus number will go up, I think Chantelle, Todd, Ken and I absolutely believe it will. The question is not whether it will go up. The question is what is that number?" Management is deliberately not pinning down the split by product line, preferring to let the supply chain dictate what actually ships. Notably, when pressed by Raymond James analyst Simon Leopold, Ullal pointed to Arista's core front-end data center and enterprise campus business, not just AI, as a source of incremental growth that had previously looked flat. "I fully expect there to be 1, maybe 2 10% customers," Ullal added regarding Microsoft and Meta remaining above the 10%-of-revenue threshold, while declining to commit to a precise number.
Scale-Across Emerges as a Distinct, Fast-Growing Category
Arista quantified for the first time how large its newest AI networking category, "scale-across," already is and how fast the total addressable market is expanding. The segment, which connects geographically distributed AI clusters given constraints on power and physical space, is expected to represent roughly 30% of Arista's $3.6 billion AI target in 2026, or about $1.2 billion. Management sized the broader scale-across switching and routing TAM at $15 billion to $20 billion by 2030, up from an estimated $3 billion to $4 billion today. Ullal was blunt about why this market exists at all: hyperscalers can no longer assume they can co-locate all their compute. "This will be temporary because once people get capacity of compute and power, they'll go back," she said, paraphrasing a skeptic, before rejecting that view outright: "The answer is no, they won't because we're going to be in a constant state of scarcity for power, space and compute." The company said scale-across today is concentrated among cloud and AI titans but is already extending into Neocloud customers.
Networking Architecture Debate: SRv6, MRC and the Case Against White Box
CTO Ken Duda used the call to lay out three specific technical innovations underpinning Arista's AI fabric pitch: Smart System Upgrade (SSU), which allows software upgrades without downtime; Multipath Reliable Connection (MRC), which allows a single data flow to be split across multiple network paths to avoid the performance penalty when two flows collide on the same link; and the use of Segment Routing over IPv6 (SRv6) to give senders explicit control over packet paths with real-time congestion rerouting. Duda called the application of SRv6 to AI fabric load balancing "not new, but... the game changer."
On the perennial white box versus proprietary systems debate, Ullal offered a more nuanced answer than in past quarters, conceding real ground to commodity hardware in simpler use cases while defending Arista's system-level approach for complex deployments. "White box is certainly a tactical solution that we tend to see more in use cases that are simple, scale-up or scale-out, where the actual amount of software and system requirements are low," she said, adding that box-by-box approaches "can slow down the job completion significantly" in more complex topologies. She also disclosed that customers value EOS for three specific reasons: lower operating costs relative to added CapEx, proprietary AI features that are difficult to replicate, and reliability given rising security vulnerabilities. Still, she acknowledged an ultra-low-price white box market "exists in every large market" and "will be there in AI, too."
NVIDIA, Open Co-Packaged Optics and the Multi-Year Roadmap
Asked about competitive dynamics around scale-up switching, Ullal was unusually forthright about NVIDIA's dominant position: "We live in an NVIDIA world... it's a high percentage of the GPUs we connect to," acknowledging that in NVLink-based deployments "there's very little participation from Arista or anybody else's scale up." She drew a direct parallel to the InfiniBand-to-Ethernet transition, suggesting migration away from NVLink toward open alternatives will take two to three years. On optics architecture, Ullal said Arista is backing an "open CPO" standard using socketed, pretested optical engines rather than proprietary co-packaged optics implementations, but cautioned adoption is early stage, with trials starting next year and broader deployment likely limited to 2028-2029, with pluggable optics and copper remaining dominant in the near term. On next-generation 1.6T switching, she said the ramp will mirror the 800G cycle, which took roughly a year from trials to volume: 1.6T trials begin in the second half of 2026 with a handful of large customers, but real production volume isn't expected until 2027.
Margins and Visibility Remain Constrained
Gross margin came in at 63.4% for the quarter, down from 65.6% a year ago on customer mix but up from 62.4% in the prior quarter, aided by tariff refunds and mix. Breithaupt quantified the tariff refund benefit at 20 to 30 basis points versus guidance for the quarter. Management reiterated that pricing increases have not yet been reflected in results and won't meaningfully affect financials until late 2026 or into 2027, given the company is still working through backlog orders. Full-year gross margin guidance was held at 62% to 64%, while operating margin guidance was raised to a range of 48% to 49%. On forward visibility, Breithaupt noted the company still operates with roughly two quarters of visibility into customer demand, and that current guidance reflects supply the company is confident it can secure, leaving room for upside "if the supply was to release a little more."