Arista Networks (ANET) Q2 2026 earnings review
Accelerating Hypergrowth Erases Margin Concerns as Arista Crosses $3 Billion
Arista Networks delivered a stunning blowout in Q2 2026, crossing the $3 billion quarterly revenue mark for the first time while accelerating YoY growth to 37.7%. The fundamental story is pure operational leverage driven by the AI megatrend: while gross margins compressed YoY due to supply chain costs and heavy Cloud Titan mix, massive OpEx leverage pushed operating margins up to 49.9%. Guidance for Q3 implies revenue growth accelerating even further to ~43% YoY. Arista has solidified its position as the de facto networking backbone for the AI era.
🐂 Bull Case
Revenue growth has accelerated for four consecutive quarters, from 27% in 25Q3 to an implied 43% in 26Q3. The rollout of 1.6 Tbps Etherlink platforms positions the company perfectly for the next wave of AI cluster buildouts.
Despite a lower gross margin profile from cloud titans and expedite costs, Arista achieved a 49.9% non-GAAP operating margin by driving total OpEx down to just 17.5% of revenue, demonstrating extreme scalability.
🐻 Bear Case
Non-GAAP gross margin fell to 63.4% from 65.6% a year ago. Arista is paying a premium to secure scarce components (optics, silicon), permanently shifting the margin profile lower as cloud concentration increases.
Total deferred revenue hit $6.87 billion, up from $4.06 billion a year ago. While this represents massive backlog, it also signals extending qualification and deployment cycles for complex AI fabrics, which could create revenue recognition volatility.
⚖️ Verdict: 🟢🟢
Highly Bullish. The sheer velocity of revenue acceleration and the ability to expand operating margins despite gross margin headwinds proves that Arista is executing flawlessly against the AI infrastructure super-cycle.
Key Themes
AI Scale-Out and Scale-Across Adoption
AI networking remains the absolute growth engine. The sheer volume of demand for connecting massive GPU clusters is pulling Arista's top line forward. The introduction of multi-planar leaf-spine designs and the open Multipath Reliable Connection (MRC) protocol demonstrates Arista is solving the unique microburst and resilience challenges of massive AI workloads, winning against proprietary InfiniBand alternatives.
1.6 Tbps Platform Cycle Kickoff
Arista officially introduced its 7060XE7 Series, a 1.6-terabit portfolio for AI fabrics. This is a critical technological leap. By supporting Linear Pluggable Optics (LPO), the new Etherlink platforms cut interconnect power consumption by approximately 60% compared to traditional pluggable optics—a massive TCO advantage for power-constrained AI data centers.
Enterprise Campus Resilience (Arista 2.0)
The Arista 2.0 strategy continues to bear fruit. The company was named a Leader in the 2026 Gartner Magic Quadrant for Enterprise Wired and Wireless LAN. The unified, single-EOS architecture is successfully expanding Arista's footprint beyond the data center into campus edges, providing a higher-margin buffer against the heavily concentrated cloud titan business.
Gross Margin vs Operating Margin Contradiction
A key contradiction in the rosy profit narrative: Non-GAAP operating margins hit a stellar 49.9%, but non-GAAP gross margins actually compressed heavily, dropping 220 basis points YoY (from 65.6% to 63.4%). The company is compensating for structural COGS headwinds (expedite fees, memory costs, cloud titan mix) by throttling OpEx growth. If OpEx needs to catch up to support the $12B+ run rate, operating margins could suddenly compress.
Deferred Revenue Implies Massive Fulfillment Frictions
Total deferred revenue surged to $6.87B (up from $6.2B last quarter and $4.06B a year ago). While bullish for future visibility, management previously cited extended 6-8 quarter qualification cycles for complex AI fabrics. This bottleneck in customer readiness (power, facilities) creates a risk of lumpy revenue recognition and trapped working capital.
Severe Supply Chain Allocation
The demand environment is constrained not by customer budgets, but by silicon, wafer, and optics availability. Arista is essentially operating in an allocation environment, forcing them into multi-year, non-cancellable purchase commitments to secure components with 52+ week lead times, escalating balance sheet risk if AI demand suddenly normalizes.
Other KPIs
Accelerating. Generated $2.78B in the first six months of 2026, up substantially from $1.84B in the first half of 2025. This robust cash generation easily funds expanding inventory requirements and working capital needs without tapping debt markets.
Up 17% YoY, significantly trailing revenue growth of 37.7%. This massive operational leverage is the primary driver of the operating margin expansion to nearly 50%, highlighting the efficiency of the single-EOS software development model.
Guidance
Accelerating. The guidance implies ~42.9% YoY growth (up from $2.308B in Q3 2025). This is a dramatic sequential and annual acceleration, signaling that Arista is successfully converting its massive backlog and securing the necessary component supply to fulfill AI network demand.
Stable to slightly decelerating. Sequentially down from Q2's exceptional 49.9%, indicating that management expects gross margin pressures (expedite fees, mix) to persist, or plans to step up hiring/OpEx in the second half of the year.
Accelerating. The midpoint of $1.07 represents a ~42.6% YoY increase from the $0.75 reported in Q3 2025, perfectly tracking top-line growth.
Key Questions
Deferred Revenue Normalization
Total deferred revenue is approaching $7 billion. When do you expect the 6-8 quarter customer qualification cycles for complex AI fabrics to compress, and what will the resulting revenue release look like?
1.6T Deployment Mix
With the introduction of the 7060XE7 Series, are initial 1.6 Tbps platform deployments skewing more toward scale-out back-end networks, or are you seeing them adopted for scale-across DCI routing sooner than expected?
Gross Margin Floor
Non-GAAP gross margins compressed to 63.4% despite immense volume. Has the company reached the trough of component expedite costs, or will the cloud titan mix continue to drag this metric closer to 62% in the second half?