A maker of cloud and AI networking gear, Arista builds the Ethernet switches, EOS operating system, and CloudVision management platform that keep data centers, campuses, and wide-area networks running. It was founded in 2004 by Andy Bechtolsheim (a Sun Microsystems co-founder), David Cheriton, and Kenneth Duda, and launched in 2008 under the name Arista — a nod to the Greek word "aristos," meaning "best."
Revenue rose 38% to $3.0B but gross margin fell 2.3 points to 62.9% as large-customer mix shifted.
fell to its lowest in over a year as large-customer sales dominated the mix. rose 37.7% to $3.04 billion and widened 0.7 points to 45.4%, but the on products sold to the company's biggest buyers accelerated. The balance grew again, leaving $5.4 billion in unrecognized AI-trial revenue still on the balance sheet.
Key takeaways
fell 2.3 points to 62.9%, driven by a greater proportion of sales to large customers who receive higher discounts, even as rose 37.7% to $3.04 billion.
Product rose 38.8% on higher switching and routing shipments, while service revenue grew as the installed base of support contracts expanded.
widened 0.7 points to 45.4% because growth of 37.7% outpaced a 17.7% increase in operating expenses, with R&D up 17.4% on personnel and new product introduction costs.
Section summaries
Management's Discussion and Analysis
Revenue surged 38% to $3.0B in Q2 FY2026, but gross margin fell to 62.9% due to a higher mix of large-customer sales.
⌄
Total grew 37.7% to $3.04B in Q2 FY2026, driven by a 38.8% increase in product revenue from higher switching and routing shipments.
declined to 62.9% from 65.2% a year ago, primarily due to a greater proportion of sales to large customers who receive higher discounts.
reached $2.8 billion for the first half of FY2026, boosted by a $1.5 billion increase in tied to customer contracts with acceptance terms.
The company held $13.3 billion in cash and marketable securities against $9.7 billion in non-cancellable , reflecting elevated commitments to manage supply constraints and AI network demand.
What changed
The balance, flagged repeatedly as a watch item, grew by another $1.5 billion in the first half of FY2026 rather than converting to recognized , signaling that AI-related customer trials with acceptance clauses continue to delay recognition.
fell to 62.9% from 65.2% a year ago, confirming the risk that a shift back toward higher-volume, lower-margin large-customer purchases would pressure profitability — the 61.9% figure in Q1 was the lowest since Q1 FY2023, and Q2's 62.9% remains well below the 64%-plus levels of the prior year.
Customer concentration intensified: two end customers accounted for 42% of FY2025 , up from 35% in FY2024, and the Q2 suggests their purchasing mix is now flowing through cost of goods sold at the higher discount rates flagged in prior filings.
Operating expense growth moderated to 17.7% in Q2 from the 36% peak in Q3 FY2025, but R&D spending continued to rise on new product introduction costs and headcount, keeping the expense base elevated.
What to watch
Whether the $5.4 billion balance begins converting to recognized in the second half of FY2026, or whether AI-related customer acceptance clauses continue to push recognition further out.
The trajectory of if the mix of large-customer sales remains elevated — the 62.9% Q2 figure is the second straight quarter below 63%, and further shifts toward these customers would pressure it further.
The impact of the February 2026 Supreme Court ruling on IEEPA-based tariffs and any new trade actions on cost of goods sold and , as management has flagged tariff escalation as a material risk.
The pace at which the $7.6 billion in due within one year flow through cost of goods sold, and whether they trigger excess charges as they did in prior cycles.
Operating expenses rose 17.7% to $532.3M, with R&D up 17.4% on higher personnel and new product introduction costs, and sales and marketing up 18.8% on headcount growth.
reached $2.8B for the first half of FY2026, boosted by a $1.5B increase in tied to customer contracts with acceptance terms.
The company held $13.3B in cash and marketable securities and had $9.7B in non-cancellable purchase obligations, reflecting elevated commitments to manage supply constraints and AI network demand.
Quantitative and Qualitative Disclosures About Market Risk
We are exposed to financial market risks, including changes in foreign currency exchange rates and interest rates, and strategic equity investments. Our exposure to market risk has not changed materially since December 31, 2025. For quantitative and qualitative disclosures about…
⌄
We are exposed to financial market risks, including changes in foreign currency exchange rates and interest rates, and strategic equity investments. Our exposure to market risk has not changed materially since December 31, 2025. For quantitative and qualitative disclosures about market risk, refer to Part II, Item 7A, Quantitative and Qualitative Disclosures About Market Risk, in our Annual Report on Form 10-K for the year ended December 31, 2025.
21
Table of Contents
The information set forth under the “Legal Proceedings” subheading in Note 5. Commitments and Contingencies of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1, of this Quarterly Report on Form 10-Q is incorporated herein by reference.
⌄
The information set forth under the “Legal Proceedings” subheading in Note 5. Commitments and Contingencies of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1, of this Quarterly Report on Form 10-Q is incorporated herein by reference.
Customer concentration, tariff escalation, and merchant silicon dependence are the most material risks to Arista's business.
⌄
Two end customers accounted for 16% and 26% of total in 2025, and any loss, delay, or reduction in their purchases could materially harm results.
Escalating U.S. tariffs and retaliatory measures, particularly involving China, Mexico, and other manufacturing countries, may increase costs and disrupt supply chains.
Reliance on Broadcom as the predominant vendor creates risk of supply shortages, innovation gaps, or competitive threats from white-box solutions.
Tightening supply in memory and silicon markets, driven by AI infrastructure demand, is causing extended and could prevent on-time customer shipments.
Rapid evolution of the AI Ethernet market introduces demand forecasting difficulty, as customers may overestimate AI build-outs and cancel or delay orders.
New or newly emphasized risks include the February 2026 U.S. Supreme Court ruling invalidating IEEPA-based tariffs and the uncertain regulatory environment for AI.