Arista Networks delivered a record-breaking fiscal year 2025 with revenue of $9.0 billion, representing 28.6% year-over-year growth, driven by robust momentum in AI and cloud networking. Q4 revenue reached $2.49 billion, up 28.9% YoY, with non-GAAP gross margin for the year at 64.6% and operating margin at 48.2%. The company achieved significant milestones in AI networking, exceeding $1.5 billion in revenue, and set an aggressive goal to double this to $3.25 billion in 2026. Despite rising supply chain costs and memory constraints, management raised fiscal 2026 revenue guidance to $11.25 billion (25% growth) and maintained a strong outlook for operating margins around 46%. The balance sheet remains robust with $10.74 billion in cash and significant share repurchases, underscoring the resilience of the business model amidst unprecedented demand.
| Metric | Value | Change |
|---|---|---|
| Q4 Revenue | $2.49 billion | +28.9% YoY |
| FY 2025 Revenue | $9.0 billion | +28.6% YoY |
| Q4 Non-GAAP Gross Margin | 63.4% | -0.8 pp YoY |
| FY 2025 Non-GAAP Gross Margin | 64.6% | N/A |
| Q4 Non-GAAP EPS | $0.82 | +24.2% YoY |
| FY 2025 Non-GAAP EPS | $2.98 | +28.4% YoY |
| Cash & Equivalents | $10.74 billion | N/A |
| Purchase Commitments | $6.8 billion | +$2.0 billion QoQ |
Arista is aggressively positioning itself as the 'gold standard' for AI networking, moving beyond traditional cloud switching. The company announced a strategic target to double AI networking revenue to $3.25 billion in 2026, driven by the deployment of 800G and upcoming 1.6T Ethernet solutions. This shift is supported by their leadership in the Ultra Ethernet Consortium and ESUN standards, allowing them to interoperate with diverse chipmakers like AMD and Broadcom, reducing reliance on a single vendor.
The company is executing on its 'Arista 2.0' strategy to expand adjacencies, specifically targeting $1.25 billion in revenue from Campus and Branch networking in 2026. The integration of the VeloCloud acquisition is central to this strategy, enabling a unified 'client-to-cloud' solution. Management highlighted strong traction in routing with the new 7800R4 spine, indicating success in penetrating new market segments beyond the data center.
Management is navigating a complex supply chain environment characterized by severe memory shortages and cost inflation. Rather than retreating, Arista is leveraging its balance sheet to secure supply, evidenced by purchase commitments increasing to $6.8 billion. They signaled a strategic shift to pass on some costs to customers via price hikes on memory-intensive SKUs, a move peers have already made, to protect gross margins.
A significant strategic shift is the diversification of the customer base. While 'Customer A and B' (Microsoft and Meta) still account for 16-20% of revenue, management anticipates adding one or two new 10% customers in 2026. This includes emerging 'Neo Clouds' and AI specialty providers like Anthropic and xAI, reducing reliance on the traditional hyperscalers and broadening the total addressable market.
Severe supply chain constraints and cost inflation pose a tangible risk to margins. Management admitted memory prices are 'horrendous' and 'exponentially higher,' forcing a price hike strategy. While guidance for gross margin (62-64%) was maintained, the commentary suggests a volatile cost environment where absorbing further increases may not be sustainable, potentially pressuring profitability if demand elasticity changes.
Customer concentration remains a critical risk, with two customers driving 16-20% of the business. Although management expects to diversify, the near-term reliance on a few hyperscalers for massive AI buildouts creates execution risk. Any pause or shift in capital expenditure priorities from these key accounts could significantly impact revenue growth rates.
The volatility of deferred revenue, which spiked to $5.4 billion, introduces unpredictability into quarterly results. Management acknowledged that large AI deals have 12-18 month acceptance criteria, making the timing of revenue recognition lumpy. This 'ballooning' balance sheet liability could lead to future quarters where revenue beats or misses depend heavily on when these acceptance clauses are triggered rather than underlying demand trends.
Analyst scrutiny regarding the growth profile outside of AI and Campus revealed potential weakness in the core legacy business. When pressed, management did not refute the characterization that the non-AI, non-campus portion of the business (roughly 60% of revenue) is guiding to flat growth, suggesting that the massive AI tailwinds are effectively masking stagnation or softness in other areas of the portfolio.
Overall: Management exhibited a highly confident and assertive demeanor throughout the call, celebrating a 'defining year' and expressing strong enthusiasm for the future. CEO Jayshree Ullal was particularly emphatic about the company's positioning in the AI market, pushing back against analyst characterizations of the guidance as cautious. While acknowledging supply chain headwinds, the tone remained optimistic regarding the company's ability to navigate costs and maintain market leadership.
Confidence: HIGH - Management raised full-year guidance significantly just months after Analyst Day, citing 'unprecedented networking demand' and 'increased visibility.' Executives spoke with certainty about product roadmaps and customer adoption rates, even when discussing external variables like memory supply.
$11.25 billion (25% growth)
$3.25 billion
$1.25 billion
62% - 64%
~46%
~$2.6 billion
Hedging & Uncertainty: Management used specific hedges primarily around supply chain variables and the timing of large customer deployments. Phrases like 'variables will decide that final number' and 'if we can ship' were used to qualify the 2026 outlook regarding new 10% customers. However, hedging was notably absent regarding the overall demand trajectory, where language was assertive ('unprecedented demand'). There was also hedging on the exact timing of revenue recognition from deferred revenue, with Chantelle Breithaupt stating it is a 'wild guess' to predict the specific quarters large balances will release.
I do not think I am being cautious. I think I went all out to give you a high dose of reality. - Jayshree Ullal, CEO
Memory is now the new gold for the AI and automotive sector. - Jayshree Ullal, CEO
We are having to smile and take it just about at any price we can get. - Jayshree Ullal, CEO
The network lags a little [behind CapEx]. - Jayshree Ullal, CEO
We do not allocate to our customers. It is first in, first served. - Jayshree Ullal, CEO
Analyst Sentiment: Analysts were broadly positive and impressed by the strong beat and raise, though they probed aggressively on the sustainability of non-AI growth and the specifics of supply chain costs. There was skepticism regarding the conservatism of the guide relative to Q1 performance, which management firmly rebutted.
Management Responses: Jayshree Ullal was direct and occasionally combative, particularly when pushed on the 'cautious' nature of the guide or the math behind segment growth. She emphasized the 'law of large numbers' and the lag between CapEx and network shipments. Chantelle Breithaupt provided measured, detailed responses on financial mechanics, particularly regarding deferred revenue and margin levers.
Supply Chain & Memory Costs: Analysts sought to quantify the impact of memory shortages and price hikes on margins. Management confirmed they are absorbing some costs but will implement selective price increases.
Customer Concentration & Additions: Questions focused on the potential for new 10% customers. Management confirmed visibility for 1-2 new additions but noted variables like acceptance criteria and supply.
Segment Growth: There was intense focus on why the core business (ex-AI/Campus) appears flat in the guide. Management argued it is too early to tell and emphasized that AI growth lifts the entire fleet.
Deferred Revenue: Analysts asked about the 'stacking' deferred revenue and when it would recognize. Management stated it depends on acceptance criteria and is difficult to predict quarterly.
Arista Networks is executing at a high level, successfully pivoting from a cloud-only vendor to a dominant player in the massive AI networking infrastructure market. The 25% raised guidance for 2026 demonstrates strong visibility and demand that outweighs near-term supply chain noise. While margin pressure from memory costs is a valid concern, Arista's pricing power and operational excellence (48% operating margins) provide a buffer. The strategic expansion into Campus and Routing via the VeloCloud acquisition and new routing platforms opens a multi-billion dollar adjacency opportunity. The company's deepening relationships with hyperscalers and emerging AI labs position it as a primary beneficiary of the ongoing AI capex boom.
The networking and server industries are facing severe constraints in DRAM and silicon fabrication, with prices rising exponentially. This is forcing vendors to hold inventory longer and implement price hikes, potentially dampening margin expansion if demand remains elastic.
Demand for AI networking remains 'unprecedented' with a TAM expanding to $100+ billion. The build-out is shifting from pilot to production scale, requiring complex scale-up, scale-out, and scale-across architectures that favor Ethernet standards.
Enterprise adoption of AI and network modernization is accelerating, with Arista noting strong traction in the $5M-$10M customer category and international growth north of 40%, indicating broad-based recovery outside of hyperscalers.