Arista Networks delivered third-quarter revenue well above its own guidance, with sales reaching US$3.66 billion and GAAP net income hitting US$1.56 billion. The networking specialist is now guiding to as much as US$4.0 billion in fourth-quarter revenue as it expands its AI infrastructure portfolio.
- US$3.66 billion Q3 revenue, up 58.5% year on year
- Revenue exceeded US$3.3 billion guidance by roughly 11%
- GAAP diluted EPS rose to US$1.22 from US$0.67
- Fourth-quarter revenue guidance of US$3.9 billion to US$4.0 billion
- New Ethernet architectures target large-scale AI networks
AI networking demand pushes Arista beyond guidance
Arista Networks, Inc. (NYSE:ANET) delivered its fastest year-on-year revenue growth since its initial public offering in the third quarter, with sales climbing 58.5% to US$3.66 billion. The result was roughly 11% above the company’s US$3.3 billion guidance and marked a 20.5% increase from the second quarter.
The performance translated into US$1.56 billion of GAAP net income, up from US$853.0 million a year earlier, while diluted GAAP earnings per share rose to US$1.22 from US$0.67. GAAP operating margin also expanded to 44.9%, compared with 42.4% in the same quarter of 2025.
Arista’s product revenue did most of the heavy lifting, reaching US$3.19 billion versus US$1.91 billion a year earlier. Service revenue rose more modestly to US$467.6 million from US$396.6 million. The figures point to a quarter dominated by hardware demand, although the filing does not identify the contribution from individual customers or programmes.
Fourth-quarter guidance points to continued expansion
Arista expects fourth-quarter revenue of US$3.9 billion to US$4.0 billion, alongside a non-GAAP operating margin of 48% to 49% and non-GAAP diluted earnings per share of US$1.26 to US$1.30. Those margin and earnings figures are adjusted measures, excluding items including stock-based compensation and intangible asset amortisation; the comparable GAAP outlook was not provided.
The quarter was not uniformly stronger on every measure. GAAP gross margin fell to 63.6% from 64.6% a year earlier, while research and development spending rose to US$486.9 million from US$326.0 million. Arista also reported US$962.8 million of inventory cash outflow during the first nine months, as inventory increased to US$3.21 billion at September 30 from US$2.25 billion at the end of 2025.
Arista used the results to highlight new open Ethernet architectures for AI fabrics, developed with companies including AMD, Arm, Broadcom, Meta, Microsoft and Qualcomm. Its liquid-cooled Etherlink SU-144 design is described as supporting up to 144 accelerators in a single rack and scaling to 1,024 XPUs across racks, while its 7800 AI Spine targets distributed AI networks across data centres. These are product capabilities and strategic initiatives described by the company, rather than reported revenue contributions.
Cash reserves rise as infrastructure investment accelerates
Arista generated US$3.98 billion of operating cash flow in the first nine months, up from US$3.11 billion in the prior-year period. Cash and cash equivalents stood at US$2.29 billion at quarter-end, while marketable securities totalled US$12.07 billion. The company made no share repurchases during the period, compared with US$983.0 million of repurchases in the first nine months of 2025.
The immediate question is whether this pace can persist as AI networking demand broadens beyond a small group of large buyers. Arista’s own risk disclosures point to customer concentration, supply constraints, reliance on a predominant merchant silicon vendor, export restrictions, tariffs and the possibility of more volatile technology spending. The next test will be whether the new rack-scale products begin to support growth without putting further pressure on gross margin.
Bottom Line?
Arista’s revenue momentum is substantial, but inventory growth, softer gross margin and dependence on large AI infrastructure buyers remain important execution tests.
Questions in the middle?
- How much of Arista’s growth is coming from AI networking compared with its established data centre and routing businesses?
- Can the company maintain margins as rack-scale systems become a larger part of its product mix?
- Will inventory continue to build ahead of demand, or convert into revenue and cash flow at the expected pace?