Arista Networks has released their financial results for the second quarter of 2023. Liz Stine, Arista's Director of Investor Relations, and Jayshree Ullal, Arista Networks’ President and Chief Executive Officer, and Ita Brennan, Arista's Chief Financial Officer, are present on the call. During the call, Arista Networks management will make forward-looking statements, such as financial outlooks for 2023 and beyond, customer demand trends, supply chain constraints, component costs, and product innovation, which may differ from anticipated results due to risks and uncertainties.

The company reported $1.46 billion in revenue and a non-GAAP earnings per share of $1.58 for the quarter. Services and software support renewals contributed 15.2% of revenue, with international contribution at 21% and the Americas at 79%. Cloud CapEx has increased significantly in the past few years to meet the customers' needs, but one customer has signaled a slowdown in CapEx. Despite this, the company expects to grow in excess of 30% annually, with the AI opportunity presenting an exciting potential.

Arista is adapting to changes in cloud and AI networking plans, investing in the Ultra Ethernet Consortium to build open multi-vendor AI networking. Arista is developing EOS features such as intelligent load balancing and advanced analyzers to report and rebalance flows for predictable performance. AI Analyzer monitors and reports traffic counters to detect and address microbursts. Arista is in the middle of trials for back-end AI networks, leading to pilots in 2024 and larger clusters and production deployments in 2025. They are also experiencing good momentum in the non-cloud enterprise category.

Arista presented a highly optimized cognitive campus solution with a single binary EOS image, universal API, and built-in automation features to modernize the customer's legacy campus. Arista also provided an EVPN L3 VPN over VXLAN routing architecture with distributed AVA sensors to provide a zero-trust approach and real-time streaming telemetry and visibility. Finally, Arista helped a large public sector customer upgrade their MPLS routing with a single spine for both LAN and WAN, resulting in an 80% reduction in total cost of ownership. Arista's platform innovation, quality, and support enabled these customer wins with a low TCO and a single cloud vision and EOS software stack.

Arista's Chief Financial Officer, Ita Brennan, is set to retire in 2024 after a successful 8-year tenure. During her time at Arista, Ita displayed the Arista way, prioritizing customers, employees, and shareholders while also delivering growth and profitability with a very small G&A investment. In Q2 of 2021, Arista's total revenue was $1.46 billion, up 38.7% year-over-year, and services and subscription software contributed 15.2% of revenue. Additionally, international revenue was $304.4 million, or 20.9% of total revenue, up from 17.5% the previous quarter.

In the second quarter, gross margin was 61.3%, in line with guidance, and up from 60.3% in the previous quarter. Operating expenses were $287.3 million, with R&D spending at $188.5 million, sales and marketing expense at $79.6 million, and G&A costs at $19.1 million. Operating income was $606.5 million, resulting in net income of $501.2 million and a diluted earnings per share of $1.58. The company also repurchased $30 million of its common stock at an average price of $137.2 per share, leaving $145 million available for repurchase in future quarters.

In the second quarter, cash from operations was $434.1 million, DSOs decreased to 49 days, inventory increased to $1.9 billion, purchase commitments decreased to $2.2 billion, and deferred revenue balance decreased to $1.085 billion. Going forward, global supply chain disruptions have resulted in shorter planning horizons and customer demand signals, particularly for cloud titan customers. These customers have had elevated purchases in the past year, and are now changing technology roadmaps and priorities before providing visibility to future demand.

Arista expects to continue shipping against previously committed deployment plans while gradually improving their 2023 outlook of over 30% year-over-year growth. Gross margin is expected to improve through the end of the year, and investments will be prioritized. Cash flow will be focused on supply chain and working capital optimization, and there will be an increase in inventory. Third quarter guidance is expected to be $1.45-$1.50 billion in revenue, 62% gross margin, 41% operating margin, 21.5% effective tax rate, and 318 million diluted shares.

Jayshree Ullal and Liz Stine from Arista discussed the outlook for revenue growth, noting that while cloud growth is moderating, it is still strong. Ullal noted that enterprise momentum is getting stronger and that the company is doubling down on AI. Stine asked for the next question from Tim Long with Barclays. Long asked Ullal to discuss AI further, including if the move to AI would expand or diversify cloud customers and the InfiniBand versus Ethernet debate.

Jayshree Ullal states that Arista is participating in the front end of the network, but for the back end, there are three classes of networks: small networks with PCIe or CXL, medium clusters with generative AI and inference, and large clusters with trillion parameters. Currently, InfiniBand is the only technology available for customers, but Arista is part of the Ultra Ethernet Consortium to make Ethernet the right long-term technology for AI networks.

Cloud customers have been prioritizing AI deployments in recent quarters due to the competitive battle between the largest titans in the world. However, there are signs that this is slowing down and cloud customers will eventually return to refreshing their standard computer infrastructure. Lead time improvements have allowed them to wait longer than usual, and there will be a mix of AI and classic cloud networking over time.

Ita Brennan spoke about the lead times being mixed across products and the goal to get back to a six-month lead time. She discussed the need to be diligent in understanding customer demand and putting it into reasonable deployment schedules and deployment plans. She emphasized the importance of understanding customer needs and timing in order to prioritize and manage inventory levels.

Arista's executives discussed the company's opportunities with hyperscalers beyond their two largest customers. Anshul Sadana said that they are doing well technologically with them but the opportunity has not yet materialized. Jayshree Ullal added that they are investing in AI opportunities with cloud titans and other high-end Tier 2 cloud providers and enterprises. Liz Stine concluded the discussion by acknowledging the fear of cloud digestion in calendar '24.

Jushree Ullal discussed how Arista has become a stronger company since three years ago, with a diversified portfolio and a larger TAM. She noted that AI opportunities have been added to the use cases, and there is a greater cohesion between the front and back ends. Liz Stine then asked Amit a tough question about the growth coming from backlog drawdown and order trends.

Ita Brennan and Jayshree Ullal discussed the sharp deceleration in growth for 4Q and how deployments into 2024 are planned. They noted that as lead time shorten, customers don't need to place orders until they get back into the lead time. They also mentioned that it will be difficult to repeat the last two years of exceptional cloud CapEx for cloud networking and that they will know more as time goes on.

Tom urged everyone to think of the business as a three-year CAGR with double digits and good numbers. Jayshree Ullal then discussed the visibility across multiple areas, with enterprise having 6-12 months visibility, cloud having less than 6 months, and AI having greater visibility due to the early cycle and joint development. Samik Chatterjee then asked about sustaining growth rate in enterprise into 2024.

Jayshree Ullal discusses the success the company has had in diversifying their enterprise business globally in different verticals. Anshul Sadana then explains the product cycle cadence for 800 gig silicon, which is being pulled by AI clusters. He predicts that there will be trials and studies in 2024, leading to volume in 2025.

Anshul's team, made up of Mark Brillhart and John McCool, have optimized the supply chain, leading to improvements in the quarter-to-quarter gross margins. This improvement is due to optimizations in freight and expedite costs, as well as a healthy cloud mix.

Jayshree Ullal explains that Arista is winning in the enterprise market due to their alternative perspective and the fact that customers have not had it for a long time. She also mentions that Arista is gaining market share as a result of their enterprise traction.

Arista has been chosen for its cloud operating model architecture that provides a high-quality, high supportive, and very friendly software experience. This shift in the enterprise is the number one reason for their success. Additionally, Arista's product depth and breadth has also been key to their success in the enterprise, and they have been engaging with customers for the last three to five years. This has made them the gold standard and given them a seat at the table.

Ita and Jayshree discussed the potential for double-digit growth in the coming year and promised to discuss more details at the Analyst Day. Erik then asked Anshul to explain how the cloud titans are buying large volumes of GPUs.

Arista CEO Jayshree Ullal discussed the company's plans for participation in AI by 2023. She noted that the purchase of GPUs is the first step, and it can take a couple of months to a year for the cluster to be fine-tuned and tested before it is released to production. For larger networks, there are different technologies that can be used, such as Ethernet or a non-networking technology, such as an IO or a bus.

Jayshree Ullal and Anshul Sadana are discussing the planning and preparation for deploying GPUs in 2025. They are focusing on clusters larger than 1000 GPUs, and the planning includes acquiring the GPUs, the size of the cluster, the language model datasets, and the network foundation. In 2024, they expect to deploy 4,000-8,000 GPUs in 400 gig type clusters, and in 2025 they expect to deploy 30,000-100,000 GPUs. Ullal emphasizes the importance of testing and taking out the kinks of the GPUs and networks in the next two years, as an efficient network is pivotal to getting the most out of expensive GPUs.

Ackerman asked Ullal about investor concerns that hyperscale customers may focus more on white box solutions for 800 gig WAN in the 400 gig cycle. Ullal commented that while there will always be an element of white box, cost savings are far outweighed by the total operational expenses needed to make the box work. She believes that Arista will continue to coexist with white box in some of their cloud titan customers.

Jayshree Ullal of Arista Networks was asked by Ben Reitzes of Melius Research if the demand for their enterprise products had picked up, as their outperformance had been a surprise. Ullal responded that their enterprise demand had always been strong, but had been overshadowed by their cloud performance. This concluded the Arista Networks second quarter 2023 earnings call, with a presentation providing additional information available on their Investors section of their website.

The call has now concluded and participants may disconnect their lines.

This summary was generated with AI and may contain some inaccuracies.