Arista Networks, Inc. Q4 FY2024 earnings call

Key Points:

• 2024 was a memorable year for Arista, achieving nearly 20% revenue growth to a record $7 billion with a non-GAAP operating margin of 47.5%.
• Q4 2024 revenues were $1.93 billion, non-GAAP EPS $0.65 (adjusted for 4-to-1 stock split), non-GAAP gross margin 64.2%.
• Strong performance in cloud and AI titans (approx. 48% of annual revenue), enterprise and financials (approx. 35%), and providers (approx. 17%).
• Core cloud AI and data center products grow 65% of revenue, with market share over 40% in 100, 200, and 400 gig ports. Increased 400-gig customer base to ~1,000 in 2024.
• Campus and routing adjacencies contribute ~18% of revenue, with network software and services (e.g., CloudVision) adding over 350 customers in 2024.
• Introduced 6 EOS software releases in 2024 with over 600 new features across core and adjacent offerings.

In Q4 2024, Arista delivered revenues of $1.93 billion. The core cloud AI and data center products grew approximately 65% of revenue. The campus and routing adjacencies together contribute approximately 18% of revenue. Arista's subscription-based network services and software contributed approximately 17% of total revenue. International contribution for Q4 2024 was 16% with the Americas at 84%. For fiscal year 2024, Arista achieved 19.5% revenue growth, with a non-GAAP operating margin of 47.5%, and non-GAAP gross margin of 64.6% for the year.

Forward-Looking Guidance:

• 2025 revenue outlook revised to approximately 17% ($8.2 billion), up from initial guidance of 15%-17%.
• Q1 2025 revenue guidance: $1.93 billion to $1.97 billion.
• Gross margin for FY 2025 expected to be in the range of 60%-62%, with Q1 2025 above the range.
• Operating margin outlook for 2025 remains 43%-44%.
• Effective tax rate expected to return to usual historical rate of 21.5% in 2025.

Risks Discussed:

• Supply chain constraints, component costs, and manufacturing output uncertainties.
• Inventory management and inflationary pressures on the business.
• Risks related to forward-looking statements, with actual results possibly differing materially from anticipations.
• Tax rate variability, as seen with the lower-than-normal quarterly tax rate in Q4 2024 due to the expiration of the statute of limitations on favorable changes in state taxes.

Q: Michael Ng from Goldman Sachs asked about the timing of AI switch deployments and gating factors related to next-generation video chips or hyperscale custom ASICs.

A: Jayshree Ullal responded that Arista is committed to four out of five AI fosters, with three out of four customers expected to roll out a cumulative 100,000 GPUs in 2024, and the fourth in pilot with expectation to go into production next year.

Q: Amit Daryanani from Evercore inquired about the impact of white box vendors on revenue growth and differentiators for Arista in AI back-end networks.

A: Jayshree Ullal stated that white box and Arista will coexist, with Arista differentiating through scale, routing features, cost balancing, AI visibility, analytics, and SmartSystem upgrade capabilities.

Q: Tim Long from Barclays asked about Meta's year-over-year revenue decline in cloud titans and other cloud titans' performance.

A: Jayshree Ullal mentioned Meta's numbers were influenced by 2023 CapEx being down 15%-20%, and all other cloud titans are performing well with demand.

Q: Ben Reitzes from Melius asked about gross margin decline and its relation to cloud titan mix.

A: Chantelle Breithaupt responded it's mix-driven, with some absorption of China tariffs, and it's a mix-driven conversation continuing to update through quarterly guidance.

Q: Meta Marshall from Morgan Stanley asked about DeepSeek and the 1:1 ratio of back end vs. front end in AI.

A: Jayshree Ullal saw DeepSeek as positive, noting AI will become more distributed across various CPUs and GPUs beyond just training.

Q: Aaron Rakers from Wells Fargo asked about Stargate and Sovereign AI opportunities.

A: Jayshree Ullal discussed vertical rack integration in AI accelerators and upcoming next-generation projects with increased performance needs.

Q: Atif Malik from Citi asked about AI back-end sales upside and TAM for 2028.

A: Jayshree Ullal stated 20-25 billion of the $70 billion TAM in 2028 is AI, and three customers deploying 100,000 GPUs will contribute to the $750 million AI back-end sales in 2025.

Q: Samik Chatterji from JPMorgan asked about the value of U.S. software layer in the back end and competition with white box players.

A: Jayshree Ullal emphasized mission-critical network functions, high availability, and resilience provided by Arista EOS in AI and cloud use cases, with difficulty in achieving such in non-U.S. alternatives without large staff.

Q: Ben Bollin from Cleveland Research asked about enterprise strategy within G2000.

A: Jayshree Ullal mentioned strong portfolio and investment in go-to-market, with focus on international growth.

Q: Ryan Koontz from Needham & Company asked about co-package optics and its place in the road map.

A: Jayshree Ullal discussed challenges with co-package optics like field failures, preferring pluggable options currently, but seeing potential with improvements.

Q: Simon Leopold from Raymond James asked about cognitive adjacencies and its growth expectations.

A: Jayshree Ullal stated cognitive adjacencies (routing and campus) contribute 18% of revenue, expecting over $1 billion in 2025, with traction in enterprise and service providers using routed backbones.

Q: Tal Liani from Deutsche Bank asked about enterprise growth drivers and routing differentiation.

A: Chantelle Breithaupt discussed enterprise growth through coverage, sales/marketing investment, and AI conversations, while Jayshree Ullal highlighted Arista's comprehensive routing portfolio with dedicated hardware and software features.

Q: Antoine Chkaiban from New Street Research asked about service providers' AI demand.

A: Jayshree Ullal noted limited activity from classic service providers but significant activity from Neo Clouds and specialty providers.

Q: Matt Niknam from Deutsche Bank asked about cash prioritization.

A: Chantelle Breithaupt stated priority on investing cash, repurchasing stock, organic investment in R&D, sales, and possible inorganic activity.

Q: David Vogt from UBS asked about speed deployment evolution from 400G to 1.6T.

A: Jayshree Ullal discussed faster speed transitions in AI, with 2024 as 400G year, 2025-2026 as 800G, and 1.6T coming in late 2026/early 2027 with new GPUs driving demand.

Q: Karl Ackerman from BNP Paribas asked about services outlook.

A: Chantelle Breithaupt stated services growth is influenced by timing, with trend-based guidance as piece parts aren't guided separately.

Q: Sebastian Naji from William Blair asked about Arista's opportunity across general purpose GPU clusters and custom ASIC solutions.

A: Jayshree Ullal stated Arista is GPU agnostic, expecting to connect to various accelerators, with market share potentially 50-50 in two to three years.