Skip to content
    ANET
    Earnings call· Dec 2024(Q4 FY24)

    Arista Networks, Inc. ANET

    Feb 18, 2025 Source

    Executive summary

    Arista Networks Q4 FY24 — Strong AI Momentum Drives Revenue Growth and Increased FY25 Outlook

    Arista Networks concluded FY24 with robust performance, driven by strong demand from cloud and AI titans, and an expanding enterprise footprint. The company raised its FY25 revenue outlook, reflecting confidence in its differentiated EOS software and hardware portfolio for data-driven networking across campus, cloud, and AI centers. Management emphasized the strategic importance of AI back-end clusters and the continued diversification of its business.

    Highlights

    5

    • Achieved record annual revenue of $7 billion in FY24, representing almost 20% growth, exceeding initial guidance of 10-12%.

    • Delivered Q4 FY24 revenues of $1.93 billion, up 25.3% year-over-year and above the upper end of guidance.

    • Reported strong non-GAAP operating margin of 47.5% for FY24 and 47% for Q4 FY24.

    • Increased FY25 revenue growth outlook to approximately 17%, targeting $8.2 billion, up from initial 15-17%.

    • Surpassed $1 billion revenue mark in 2024 for software and subscription services category.

    Concerns

    3

    • International revenue contribution decreased to 16% in Q4, down from 17.6% in Q3, due to increased domestic mix.

    • One of five AI clusters mentioned in prior calls is stalled due to awaiting GPUs and funding.

    • Gross margin guidance for FY25 is 60-62%, lower than Q4 FY24's 64.2%, primarily due to mix and China tariffs.

    Guidance & targets

    13
    CategoryTargetConfidence
    FY25 Revenue Growth
    approximately 17%
    high materiality
    High
    FY25 Revenue
    $8.2 billion
    high materiality
    High
    FY25 Gross Margin
    60% to 62%
    medium materiality
    Medium
    Q1 FY25 Gross Margin
    approximately 63%
    medium materiality
    High
    FY25 Operating Margin
    43% to 44%
    medium materiality
    Medium
    Q1 FY25 Operating Margin
    approximately 44%
    medium materiality
    High
    FY25 Structural Tax Rate
    21.5%
    low materiality
    High
    Q1 FY25 Revenue
    $1.93 billion to $1.97 billion
    high materiality
    High
    Q1 FY25 Effective Tax Rate
    approximately 21.5%
    low materiality
    High
    Q1 FY25 Diluted Shares
    1.285 billion
    low materiality
    High
    AI Revenue Goal
    $1.5 billion in AI centers, including $750 million in AI back-end clusters
    high materiality
    High
    Campus and Routing Adjacencies Revenue
    over $1 billion
    medium materiality
    High
    CapEx for Santa Clara Facility
    approximately $100 million
    low materiality
    High

    Segment performance

    8
    SegmentRevenueYoYQoQMargin
    Cloud and AI Titans
    Contributed significantly to annual sector revenue for 2024. Oracle is a new member of this category. Microsoft and Meta are greater than 10% concentration customers.
    Microsoft concentration: approximately 20%Meta concentration: approximately 14.6%
    		approximately 48%———
    Enterprise and Financials
    Strong contribution to annual sector revenue for 2024.
    		approximately 35%———
    Providers
    Includes Apple, contributed to annual sector revenue for 2024.
    		approximately 17%———
    International
    Quarter-over-quarter decrease driven by the relative increased mix of domestic revenue from large global customers.
    Percentage of total revenue: 16%
    		$311.1 million—down from 17.6% last quarter—
    Americas
    Super strong for the quarter.
    Percentage of total revenue: 84%
    		————
    Core Cloud AI and Data Center Products
    Drove approximately 65% of revenue in 2024, deployed across 10, 25, 100, 200, 400 and 800 gigabit Ethernet speeds.
    Percentage of revenue: approximately 65%
    		————
    Campus and Routing Adjacencies
    Includes investments in cognitive wired, wireless zero-touch provisioning, network identity, and sensors for threat mitigation.
    Percentage of revenue: approximately 18%
    		————
    Network Software and Services
    Based on subscription models like Arista A-Care, CloudVision, DMF observability, and advanced security sensors. Exceeded $1 billion revenue mark in 2024.
    Percentage of total revenue: approximately 17%
    		————

    Operational metrics

    25
    Services and subscription software revenue
    18.3%up from 17.6% in Q3
    Q4 FY24

    Percentage of total revenue.

    Operating expenses
    $332.4 millionup from $279.9 million last quarter
    Q4 FY24

    Non-GAAP operating expenses.

    R&D spending
    $226.1 millionup from 9.8% last quarter
    Q4 FY24

    Non-GAAP R&D spending, matching expectations for timing of engineering costs for next-gen products.

    R&D spending
    11.2%
    FY24

    Non-GAAP R&D spending as percentage of revenue, demonstrating continued focus on product innovation.

    Sales and marketing expense
    $86.3 millionup from $83.4 million last quarter
    Q4 FY24

    Non-GAAP sales and marketing expense, driven by continued investment in headcount and channel programs.

    G&A costs
    $19.9 millionup from $19.1 million last quarter
    Q4 FY24

    Non-GAAP G&A costs, reflecting continued investment in scaling the company.

    Non-GAAP operating income
    $907.1 million
    Q4 FY24

    Strong Q4 finish.

    Non-GAAP operating income
    $3.3 billion
    FY24

    Impressive achievement for the fiscal year.

    Other income and expense
    $89.3 million
    Q4 FY24

    Favorable result.

    Effective tax rate
    16.7%
    Q4 FY24

    Lower-than-normal quarterly tax rate due to release of tax reserves and favorable changes in state taxes.

    Non-GAAP net income
    $830.1 million
    Q4 FY24

    Result for the quarter.

    Diluted share count
    1.283 billion
    Q4 FY24

    Post-split basis.

    DSOs
    54 daysdown from 57 days in Q3
    Q4 FY24

    Reflecting timing of shipments and strong collections performance.

    Inventory turns
    1.4xup from 1.3x last quarter
    Q4 FY24

    Reflecting diligent inventory management.

    Inventory
    $1.83 billionmarginally increased
    Q4 FY24

    Reflecting diligent inventory management across raw and finished goods.

    Purchase commitments
    $3.1 billionup from $2.4 billion at the end of Q3
    Q4 FY24

    Represents purchases for chips related to new products and AI deployments. Expected to maintain a healthy position for key components.

    Product deferred revenue increase
    $150 millionover the last quarter
    Q4 FY24

    Due to new product introductions, customers, and expanded use cases leading to increased customer trials and contracts with acceptance clauses.

    Accounts payable days
    51 daysup from 42 days in Q3
    Q4 FY24

    Reflecting the timing of inventory receipts and payments.

    Total employees
    approximately 4,465
    2024

    Rooted in engineering and customer investments.

    Installed base
    100 million
    cumulative

    Cumulative ports of installed base since going public.

    Net Promoter Score
    87
    current

    Highest in the industry, reflecting quality and support.

    400-gig customer base
    approximately 1,000
    2024

    Increased customer base for 400-gig products.

    AI back-end clusters GPUs rolled out
    100,000
    2025

    Cumulative GPUs expected to be rolled out by 3 of 4 active AI customers.

    Market share in switching category
    >40%
    current

    Attained #1 position according to industry analysts.

    Total Addressable Market (TAM)
    $70 billion
    2028

    Total market TAM for 2028, with approximate segment breakdown.

    Industry KPIs

    12
    MetricValueDetails
    EPS$0.65 (Q4 FY24); $2.27 (FY24)USD
    Gross margin64.2% (Q4 FY24); 64.6% (FY24)%
    Operating margin47% (Q4 FY24); 47.5% (FY24)%
    Operating expenses$332.4 million (Q4 FY24)USD
    Operating cash flow$1 billionUSD
    Revenue growth rate$1.93 billion (Q4 FY24); $7 billion (FY24)USD
    Customer account countExceeded 10,000customers
    Cash investments balance$8.3 billionUSD
    Inventory channel inventory$1.83 billionUSD
    Rpo backlog bookings orders$2.79 billionUSD
    Share buyback capital return$123.8 million (Q4 FY24); $423.6 million (FY24)USD
    Segment end market revenue mixCloud and AI titans: ~48%; Enterprise and financials: ~35%; Providers: ~17%%

    Orderbook & backlog

    2
    Total Deferred Revenue$2.79 billionQ4 FY24

    up from $2.51 billion in the prior quarter

    Majority is services related, linked to timing and term of service contracts. Product deferred revenue increased by $150 million over Q3.

    Purchase Commitments$3.1 billionQ4 FY24

    up from $2.4 billion at the end of Q3

    Represents purchases for chips related to new products and AI deployments. Expected to maintain a healthy position related to key components.

    Product announcements

    2
    ProductTypeDetails
    SWAG (Switched Aggregation Group)launch
    EOS Software Releasesmilestone

    Deals & partnerships

    1
    MetaJointly engineered product for distributed Ecolink switch

    Co-developed the distributed Ecolink switch, which provides leaf spine combination with both FBOSS and EOS options.

    Capital programs

    1
    Santa Clara Facility Constructionunderway
    Period spend: approximately $100 million
    Start: October 2024

    Benefit: expanded facilities

    Initial construction work began in October. Expected to incur approximately $100 million in CapEx during fiscal '25 for this project.

    Risks & headwinds

    3
    Stalled AI cluster deployment2025

    1 out of 5 AI clusters

    Mitigation: Management is focusing on the remaining 4 clusters, with 3 expected to roll out 100,000 GPUs. The stalled cluster is awaiting GPUs and funding.

    China tariffsFY25

    absorbed a little bit of the specific tariffs

    Mitigation: Management is monitoring the fluid tariff situation and being thoughtful for both short- and long-term outcomes. Planning for China ahead of time has been done.

    Market volatilitysince the beginning of the year

    recognition of the volatility

    Mitigation: Reflected in the combined outlook for cloud, AI, enterprise, and cloud specialty providers, leading to increased FY25 revenue guidance.

    What to watch in Q1 FY25

    5

    4th AI customer production status

    next year
    CurrentIn pilot, migrating from InfiniBand to Ethernet
    TargetGo into production

    Why it matters

    Indicates successful migration to Ethernet for a historically InfiniBand customer, validating Arista's solution and expanding its AI market.

    On the fourth one, we are migrating right now from InfiniBand to proving that Ethernet is a viable solution, so we're still -- they've historically been InfiniBand. And so we're still in pilot and we expect to go into production next year.

    Q&A highlights

    6

    How does the timing of AI back-end switch rollouts align with the deployment of next-gen GPUs, and is GPU availability a gating factor?

    Arista is committed to 4 of 5 AI clusters, with one stalled. Three customers are expected to roll out 100,000 GPUs cumulatively in 2025. The fourth customer is migrating from InfiniBand to Ethernet, with production expected in 2026.

    “3 out of the 4 customers are expected to this year rolled out a cumulative of 100,000 GPUs.”

    asked by Michael Ng · answered by Jayshree Ullal

    2 min read6 chapters

    Detailed Narrative

    01

    AI Back-end Cluster Deployments

    Arista remains committed to 4 out of 5 previously mentioned AI clusters, with one stalled due to awaiting GPUs and funding. Three of the active customers are expected to roll out a cumulative 100,000 GPUs in 2025. The fourth customer is currently in pilot, migrating from InfiniBand to Ethernet, with production anticipated in 2026, demonstrating Ethernet's viability as a solution.

    02

    White Box vs. Arista in AI Networks

    The AI network market is vast, allowing for coexistence with white box solutions. Arista's EOS-based products typically dominate the AI spine due to the need for rich routing, scale, and advanced features. While AI leads can be hybrid (using SONic or FBOSS), Arista's differentiators include superior scale, routing capabilities, cost and load balancing, real-time AI visibility and analytics, personal queuing, congestion control, and smart system upgrades, which are crucial for expensive GPU deployments (costing $25,000 each).

    03

    Evolution of AI Accelerators and Networking

    The company foresees a shift towards more vertically integrated rack solutions for AI, combining processors, scale-up, scale-out, and software for unified control, particularly in 2026 and 2027. This trend will drive the demand for higher network speeds like 1.6T Ethernet. Arista aims to be GPU agnostic, connecting to various accelerators including NVIDIA, AMD, and in-house designs, anticipating a more diversified market beyond NVIDIA's current dominance in 2-3 years.

    04

    Enterprise and Routing Strategy

    Arista's enterprise strategy, especially within the Global 2000, is a key investment area, benefiting from a trickle-down of high-end data center products. The focus is on go-to-market execution, increased sales and marketing headcount, and expanded partner programs. Routing, evolving from a software enhancement, is now a critical offering for service providers and large enterprises, with a comprehensive feature set (VXLANSEC, MPLS, BGP) enabling dedicated hardware solutions like the 7280 platform.

    05

    Co-packaged Optics (CPO) and Copper (CPC)

    Co-packaged optics (CPO) has historically faced challenges with field failures and serviceability, leading customers to prefer flexible pluggable switches and optics. Arista supports co-packaged copper (CPC) and linear drive optics (LTO) as alternatives. While CPO could offer benefits like higher channel counts, its widespread adoption at 224 gig or 448 gig will depend on improvements in reliability and manufacturing, as customers currently prioritize modularity.

    06

    Speed Transitions in AI Networking

    AI is accelerating network speed transitions, shortening cycles from 3-5 years to approximately 2 years. 2024 was characterized by 400 gig deployments, with 800 gig expected to dominate 2025-2026. The introduction of 1.6T Ethernet for production is anticipated in late 2026, with significant deployment in 2027, contingent on the availability of new GPUs and advancements in liquid cooling technologies.

    AI-generated summary of the company’s earnings call. Not investment advice.