The Post Earnings Drop In Arista Networks Is An Opportunity (ANET) - $135-$137

Even after small beats on earnings and revenue for Q3 and guidance for Q4, and 2025, Arista dropped significantly from its all time high of $165. Analysts and investors didn't seem to like the margin contraction for next year. But, I believe it is a long-term positive to continue gaining traction in a massive growth industry.

I plan to add this week. This company can still return about 15-18% annually from these levels.

The premium is worth it: The stock may look expensive at 40x earnings, growing 22-24%, and 16x sales, growing 20-22%, but given its strong operating margins of 45% and operating cash flow margin of 50%, and its pole position in data center connectivity, I believe the premium valuation is justified. I also believe that it will remain the market leader in AI data center connectivity, and ride its growth for the next 3-5 years. The total addressable market for Arista can grow to $100Bn in the next 5 years, given the massive AI data center buildout. In 2025, AI has moved to inference as the primary growth driver instead of training, and demand for more inference tokens will lead to more data in transmission, leading to the need for more networking capacity. Arista will increase its share of AI-oriented revenue to 24-28 % of net sales in the next 2 years.

The strategic partner for hyperscalers: As I’ve mentioned for several years, Arista planned its business as a strategic partner for cloud service providers and hyperscalers such as Microsoft (MSFT) and Meta (META), collaborating with them from the blueprint and design stage, integrating its rack-scale infrastructure, of cabling, co-packaging, power supply, cooling systems, and the interconnects across semiconductors. Arista has the largest market share among ethernet providers in this segment, with the closest competitors being Nvidia and Broadcom.

Growth catalyst: Arista’s Ethernet for Scale-Up Networking [ESUN], is a collaboration with major AI players and developers for common Ethernet integration for "scaling up". Data center "scale out", is already ethernet reliant - scaling out is horizontal, more servers connected together, wider (horizontal). "Scaling up" is more intensive, i.e getting more out of existing clusters, vertical, increasing connectivity between servers to function better as a cohesive unit. Scaling up is dominated by Nvidia's Infiniband. Arista is going to make inroads in this market, simply because the market is big enough for several players and it will continue growing.

Sacrificing near term margin for long-term customer loyalty: Crucially it starts from the ground up as a blue box model, a customized approach for hyperscalers, instead of a white box, plain vanilla solution. Arista will likely face higher costs because of this impacting margins in the short-term. But I think the markets made a mistake worrying about the near term impact of margins from a blue box model, instead of focusing on the high level of customer loyalty from the longer-term approach, where switching costs would become a huge competitive advantage. This is their playbook after all, and the drop in its stock price is an opportunity to buy.