Deep Dive - Arista Networks
The Backbone of the Next Generation of Computing Technology? Or Another Networking Bust?
“Cloud computing offers individuals access to data and applications from nearly any point of access to the Internet, offers businesses a whole new way to cut costs for technical infrastructure, and offers big computer companies a potentially giant market for hardware and services.”
I’m sure you’ve heard the old adage about the people who made the real money during gold rushes; it wasn’t the miners, who toiled away in toxic conditions, broke their bodies and spirits, or lost their money and lives in search of that fickle mistress. No, the people who made the most money were sitting back at a trading post and supplying every single one of those miners with the mining equipment they would all require. This is where we get the notion of a ‘picks and shovels’ company, which essentially serves as the backbone to one hype industry or another. Today’s company is a picks and shovels play on a few hype industries, including cloud platforms, data and data centers, and wider use cases for AI/ML.
Every one of these technologies is crucial to the operation of our modern internet and continuing (& accelerating) technological growth as a society. The cloud especially has taken a central role in the functioning of the internet, with financial institutions, government agencies, large enterprises, and streaming platforms all heavily relying on cloud infrastructure for their operations. Data, meanwhile, is said to be the next coming of gold, wherein access to and insight into data will decide the winners and losers in tech for the coming decades. More recently, AI and ML and have taken the hype cycle by storm, threatening to engulf us all in some myriad combination of evil AI robots, an overload of information, the betterment of the entire human race, and everything in between. I’m not getting into that today, or likely ever. But I am interested in what makes these industries tick - what is the picks and shovels play for allowing all these weird and wonderful industries to exist? What will every company making a play in the realms of data, AI, and cloud require as a backbone for their operations? Enter Arista Networks (Ticker: ANET), a company disrupting the networking industry as we have known it to provide a full systems-based approach for data centers, cloud infrastructure, and high-demand AI/ML uses in enterprise operations.
Thanks for reading Hourglass Investing ! Subscribe for free to receive new posts and support my work.
Disclaimer: I am not a professional financial planner, advisor, analyst, or any other word to do with finance. I also own shares in Arista and may subsequently be biased towards the company and its future success. Please take all following information with a (large) grain of salt and do not make financial decisions based exclusively on the information you read here! It is intended for entertainment and education purposes only.
Arista was founded in 2004 by former Cisco employees Ken Duda, Andy Bechtolsheim, and David Cheriton. Cisco was, for a long time, the industry leader in networking hardware (and is an excellent example of why it’s best not to blindly follow the crowd into inflated prices, if you care to go and look at its stock chart). However, Cheriton and Bechtolsheim left positions at the company to start Arista and brought along Duda - Duda and Bechtolsheim are still involved in C-suite positions with the company today, and were successful in poaching another former Cisco-ite, Jayshree Ullal, to head the company as CEO.
Arista serves to solve many of the same problems that Cisco does, only better (remember what I said about possible bias). The company manufactures switches, which are pretty well the most boring things ever, and have put me to sleep many a time while researching this company. Boring though they may be, they are essential to efficient and scalable computing capacity - essential enough that Google essentially threw down the gauntlet in 2004, requesting a networking switch that exceeded the ability of anything that existed at the time. This challenge was the foundation of Arista’s creation - within months of Google’s challenge, the three founders were working towards meeting this demand.
By 2008, they were ready to launch their first product and begin disrupting the networking hardware industry. It was at this time that they brought in Ullal to serve as CEO, who brought business-side leadership to the engineering-oriented trio of founders. Their initial offerings were targeted towards financial institutions, where they found enough success to provide revenues and an industry foothold that allowed them to continue growing. By 2010, just two years later, they had developed a switch with 6x the capacity of the nearest comparable product - then it was off to the races, as Arista began competing for cloud customers, helping them to take the company public in 2014.
Since then, Arista has consistently demonstrated the values that gave them their initial success: innovation, consistent improvement, and a company culture that drives excellence. And the proof is in the pudding - market research on networking switches for data centers shows that Arista has consistently been stealing the lunch of legacy players;
Arista has gained a ~25% total market share in dollars and 19% total market share in networking, while incumbents, such as Cisco, steadily bleed market share. While 19% may not seem like a lot, keep in mind that this is a highly competitive industry, where even 1% market share is difficult to obtain and costly to maintain. It is also not truly representative of Arista’s focus as a business up to this point - the company serves mostly large enterprise customers that require Arista’s high-performance products, like the 200- and 400-gigabit switches. In these categories, they are the #1 player in the industry, proving that Arista is the most trusted player to satisfy the performance requirements of large businesses and that the advantage gained through Arista’s offerings attract and maintain best-in-class enterprises. Given these facts, Arista’s business model is obviously skewed towards the top-end - while thye have begun expanding their offerings to small- and mid-sized players, this area of growth is only just beginning. Considering this, the 19% market share is doubly impressive - Arista needs to have absolutely dominated the top-tier market in order to achieve this. Digging a bit deeper, we can see this is exactly what they’ve done;
Despite Arista’s success at the top of the tech industry, as well as an almost $50bn market cap as of writing, there is still a lot of room for future growth in a burgeoning networking industry. If the company can continue its trend of innovation to further steal market share and maintain a a competitive advantage, as well as take advantage of some growth runways, Arista Networks has the opportunity to become the dominant player across several industries and with companies of all sizes. If it finds success in doing so, Arista may find itself the critical backbone of networking, AI, cloud, data, and future technologies we haven’t even started to imagine yet.
Let me preface this by iterating that I am not a super knowledgeable on the intricacies of networking hardware. This will be a very high-level overview of the offerings and products that Arista offers its customers. If you are interested in a more in-depth look at the specs of individual hardware odds and sods, or need some help getting to sleep tonight, I highly recommend checking out Arista's Product page.
Right, let’s get into it. Before we delve into the business model and offerings, let’s take a glance at Arista’s customer base.
Arista started off selling its networking equipment to financial institutions, where the reliability and low-latency (which just means minimal delay on high volumes) allowed stock market traders a slight advantage over competitors with slower equipment. As time went on, Arista began securing wins within a broader customer base, including Microsoft and Meta, which are now its two largest customers.
Today, Arista sells to 9000+ customers around the world, though their revenue generation still very much revolves around U.S.-based financial institutions and tech giants that require best-in-class networking systems, as well as government agencies. As Arista expands its product offerings and global presence, and cements its reputation within the networking industry (pretty easy to do when you’re partnering with some of the best and brightest companies in the world), the number and types of customers should continue to grow with the company. As of today, Arista’s target customers are large enterprises that need high-performance networking systems (either software, hardware, or both) for cloud, AI/ML, and/or data centers, as well as the more recent additions of small- and medium-sized businesses across various industries in need of secure and reliable WiFi, access points, and compact ethernet switches.
“We were looking for a flexible, open data center switch architecture with multiple vendor products that would scale, and found Arista when we started. Arista has become the de facto standard in the cloud world as the maker of core networking fabrics".”
-Lance Crosby, CIO at StackPath
Arista’s business model revolves around the design of extremely high quality, low-latency, and reliable networking hardware - the company then partners with a third-party organization to complete the manufacturing. This is very critical - while Arista designs its hardware, it is not involved with the minutiae of the manufacturing process. While a significant amount of work and innovation has gone into crafting exceptional and high-performing hardware, which generates approximately 80% of the company’s revenues, Arista’s true bread and butter is its software - the Extensible Operating System (EOS), an operating system engineered by Arista to be programmable and flexible to individual customer’s needs, and CloudVision, which allows insights into and control over the network, as well as automation of various network features. Though combining for only ~20% of revenues, the software portions of Arista allows businesses to use any third-party applications or tools on top of the software so as to be able to offer their unique offering in conjunction with Arista’s products. In addition to being very open to customization, Arista’s software also gives enterprises zero-trust security, operational efficiency, visibility into networks, and the safety blanket of a network operating without outages - this final point is crucial to the businesses that trust Arista to supply their networking systems, as outages can cost them millions of dollars, damage their reputation and customer loyalty, hinder efficiency, and provide openings in cybersecurity that could yet further damage them. The stakes are high enough that businesses are willing to splash extra cash for the best product in the market - Arista provides that best-of-breed product, making the money spent on Arista products non-optional spend for many companies. It is simply not an area they can afford to cheap out on. Between their mission-critical software and high-performance hardware, Arista’s combination of offerings transcends a narrow definition as either a software or hardware company, transcending to become a full network system-based enterprise at the heart of the tech industry.
Again, this is going to be super high level. Keep that in mind - if you are genuinely interested in Arista, dig deeper into the products they offer and how they help differentiate them from competitors.
Switches - The original mission that Arista’s founders set out to accomplish was create the world’s best switch, and that’s exactly what they managed to do. Since then, they have managed to redefine the boundaries of switch capacity several times. But what is a switch? Like I said, I’m no networking hardware expert. So, I asked ChatGPT to explain it to me like I was a 5 year-old, and it was the first time an explanation of a switch managed to click in my brain (I don’t know if this says more about my level of intelligence or the complexity of Arista’s products, and I’m not sure I want to know either). Here’s what I got out of it - switches connect all devices on a network and allow the efficient and speedy transfer of information to the correct destination over a network. While this doesn’t sound particularly impressive, speed is crucial in the tech industry. Since their first product, Arista has continued to innovate to create consistently more impressive switches; now they are the leader in 100, 200, and 400G switches - 400G switches are the most efficient in the market today. 100G switches are primarily used by large enterprises, 400G for the cloud, while 800G switches (which is made possible by leaf-spine architecture that connects multiple switches - see below) are used for AI use cases.
Adjacent to switches, Arista also started designing leaf spine hardware for networking and data centers, which is essentially infrastructure that connects multiple switches into a single capacity system. It allows a significant increase in the amount of data that a network can handle at once without slowing down or crashing - for any company handling vast amounts of data, this is critical infrastructure.
Finally, the R-series switches that Arista designs blends switches and routers into a a condensed hardware, which reduces the amount of space, money, and installation time required by companies to achieve the same effect. To summarize switches in a nutshell, they save time by making networks more efficient. And time is money. And money is good.
“Arista is really virtualizing flexible networking, working in multiple environments with open APIs to provide interoperability. They are building ultra low-latency network platforms which are important for cloud-based applications, and at the same time are optimizing these platforms for space and energy efficiency.”
-Diane Green, Former CEO of Google Cloud
EOS - EOS is the operating system that allows Arista hardware to run and function in conjunction and synchronization. It manages the flow of data and ensures efficiency and reliability in the network. It’s also the portion of the software that network security is built into, which is enabled by Arista’s use of AI and ML in Network Detection Response (NDR) to allow a strong (and critically important) security platform across networks that utilize Arista’s software. This is done by building NDR sensors directly into switches to allow security analytics to be built directly into the network, a feature that has made Arista a leader in Network Detection Response. Cybersecurity is of critical importance in our modern tech age - many companies say that cybersecurity would be one of the very last expenditures that they would slash, as any data leak or infrastructure failure as a result of a cyber attack can, again, mean millions of dollars lost and damage to a company’s reputation. This is another component of Arista’s value proposition that ensures its customers are satisfied, safe, and returning to them for high-end networking system quality.
CloudVision - This is the network management component of Arista’s offering. As described above, it allows network security, visibility, openness, troubleshooting, reliability, and efficiency. It also allows seamless troubleshooting and updating capacity that can be quickly undone if there is an error in the code - for example, if a change is made to the software at the heart of a network and there is an error involved, engineers can revert back to the original system pre-update at the click of a button, keeping networks operating while teams work to flesh out the problem in the code.
Between network security and complete network management capabilities, Arista’s EOS and CloudVision software is the key differentiator against other competitors in the networking space, allowing network insights and efficiencies that hardware alone cannot begin to offer.
Cognitive Campus & Routing - Newer to the Arista product offering is the business’ Cognitive Campus. The campus refers to the Local Area Network (LAN) of a company and organization, including all wired and wireless connections and networks that contain multiple buildings physically separate from each other. The Cognitive Campus is designed to take some of the network features experienced in data centers, like automation and threat detection, and expand it to non-data center organizations and various enterprise sizes that have different needs from data centers. This area of Arista’s business allows it to expand into a larger number of use cases and customers, and focuses on providing WiFi 6e, smooth integration of wired & wireless connections, reliable and agile internet, condensed hardware, simplified network management and security across the entire digital infrastructure of a network environment, and scalability for companies in growth stages. This offering can also be utilized in campus environments like universities or colleges (just to add some confusion to the name) that undergo large traffic volumes while needing to maintain a smooth network experience for users. This portion of the business is growing fast, expanding from ~1500 to ~2000 customers (+~33%) YoY.
Roped in with the 20%ish revenue gained from software is Arista’s Support segment. This section of the business provides supplemental and fairly steady, repeatable income for the company at high margins. As Arista’s technology and networks in general can be extremely complex and requires a lot of know how, this is a crucial offering for ensuring the Arista culture of placing customers first and ensuring product satisfaction is maintained. There is not a dedicated staff for providing this service - customers have full, 24/7 access to the engineering team itself, helping put the people most knowledgeable on the products face-to-face with the real-world issues faced by clients. The support segment offers customers;
Professional Services - consulting on network design and operational efficiency - especially useful for large enterprises with complex networking requirements, or smaller businesses lacking personnel with the technical knowledge required to fully take advantage of Arista’s offering. Also included in the professional services is training for just such situations and teams, so as to make sure businesses are equipped with the knowledge and capability to independently utilize Arista’s products in the long-term.
Software Updates - As with any software, Arista’s EOS and CloudVision software requires constant tweaking. Perhaps even more than other software, as this is a company that is constantly in innovation mode and seeking to deliver better products to customers.
Hardware Replacement - Hardware is not invincible; eventually, individual components or larger sections of hardware will need replacing if they fail, or if a company chooses to upgrade their portions of their networking infrastructure. Arista provides these installment and replacement services in their Support segment.
Throughout its corporate history, Arista has formed several key partnerships with various enterprises that give them a leg-up against their competitors in the networking space;
Microsoft & Arista have been partnered since Arista’s early days to revolutionize the cloud. Beginning in 2008, Arista and Microsoft have worked closely to build out the Azure cloud platform’s data centers with scalable, low-latency, and programmable switches necessary for a constantly-evolving industry. This partnership has worked out to be immensely profitable for both Microsoft and Arista, and speaks volumes to the quality of Arista’s team that this arrangement has continued on for so long.
Broadcom & Arista are partnered to place Broadcom’s high-performance silicon semiconductor chips in Arista’s switches, giving their hardware the capacity needed to handle intensive demand. Arista’s partnership with Broadcom has helped both companies, providing the chip company with steady revenues and Arista with a dependable partner to supply next-gen semiconductor capacity - this allows them to continue innovating at rapid pace with high-quality components and without fear of low-quality components.
Dell & Arista are partnered to supply Arista networking products through Dell’s data center solutions. This arrangement would help to provide Dell’s customers with product satisfaction while accelerating Arista’s growth and market share.
Palo Alto Networks & Arista’s partnership helps to deliver Palo Alto’s network security platform and capabilities through Arista’s network infrastructure, helping to provide an extra value proposition for Arista’s customers, protecting valuable cyber assets, and expanding the scope of Palo Alto’s business operations.
Balance Sheet - The Bad & The Ugly
Customer Concentration - Meta and Microsoft accounted for 42% of Arista’s revenues in fiscal 2022, demonstrating the large dependence Arista has on the tech giants to drive values. While these revenues are choppy and bound to fluctuate based on when the two companies make large purchases, they have been +10% customers for a large portion of Arista’s history.
While Meta and Microsoft are highly unlikely to go under, any reduction in spend on Arista’s products from these companies would have significant impact. Meta semi-recently announced a “year of efficiency” - if that efficiency involved buying fewer of Arista’s products, it would undoubtedly mean a year of much lower revenue, in turn lowering analyst expectations and subsequently the share price as well. Conversely, if Meta does continue to spend on Arista products while focusing on efficiency, it would demonstrate both the value offered by Arista and the ‘required spend’ nature of their products. It will be interesting to keep an eye on Meta and Arista’s relationship over the rest of the year (and possibly several coming years) to monitor the nature of this relationship/
Gross margins - Gross margins have been on a slight decline over the last several years, and are currently hovering around 60%. While this is nothing to sneeze at, it is certainly a decline from the 67% margins they were operating with the year they went public (2014). It is also potentially indicative of a problem faced by many companies in competitive spaces, particularly surrounding manufacturing (again, Arista does not directly manufacture their products, but they still sell hardware that serves as a commodity and has competitors, so the effect is the same) - in order to stay competitive with prices offered by the competition and not lose/continue gaining market share, prices must be slashed concurrently and in turn affect gross margins. If this downslide in margins continued, it could reduce the attractiveness of the stock for investors, the share price, and the overall quality of the business.
It’s not all bad on the margin front - gross margins are reduced partly as a result of the customers that Arista serves. Because they primarily sell to large enterprises, their customers are usually making large orders and are able to negotiate discounts, which in turn reduces gross margins. This is simply a consequence of the niche in which Arista operates, but it is an important trend to keep an eye on. It also serves as a good incentive for Arista to continue expanding their business to small- and medium-size customers to help balance out the lower margins from their current customer base. Furthermore, Arista’s lower-margin (I say ‘lower’ but they are really quite exceptional margins still) sales help to accelerate the company’s growth by stealing market share from networking competitors and shipping their products to a greater number of customers, who are presumably then sold on the quality of Arista’s products and roped into the exorbitant switching costs (more on this later) to change network providers again. As the margins they accept to fast track their growth in this manner are still more than sustainable, I actually don’t view Arista’s margin slide as a bad thing - so long as it is not a constant trend in the long-term.
Balance Sheet - The Good
Customer Concentration - No, you aren’t seeing double. I believe that Arista’s customer concentration is both a good and a bad thing. Previously discussed are the risks associated with having a great portion of your annual revenues wrapped up in just two companies, including the impact on gross margins and potential harm from any reduction in spend. However, there are some advantages to this situation as well, especially when those two companies are as highly successful as Meta and Microsoft. Like I mentioned above, if these companies continue to buy Arista products even in economic downturns, it speaks to Arista’s value proposition. Furthermore, selling to the best of the best gives Arista a highly distinguished name in the tech industry and serves as a selling point for their products - if Meta and Microsoft have trusted them for several years, potential customers can be assured they’re in safe hands with Arista.
ROIC - Typically, any Return on Invested Capital (ROIC) above 15% is a fairly sure sign that the capital a company invests is driving strong returns. Arista’s ROIC is 29%, which is an incredibly attractive figure and speaks volumes on Arista’s capital efficiency. ROIC is one of my favourite metrics to monitor in assessing new/current investments - if it is a metric you aren’t familiar with, I highly recommend this article on the benefits and growth experienced by companies with strong returns on invested capital.
Strong Cash Position - With just under $1bn in cash on the balance sheet, Arista has plenty of dry powder for acquisitions, cover unforeseen circumstances, or whatever else they may choose to/be required to use it for. Their free cash flow generation is impressive too, raking in ~$450m in free cash flow generation in 2022 and showing their ability to fund business operations while still building ample cash reserves. A strong cash position gives Arista a safety blanket to fall back on, pay off debts, or capacity to further expand and grow.
Low Debt-to-Equity - On top of having a lot of cash on hand, Arista has very little in the way of debt that they need to worry about, giving investors an extra sense of security if they choose to put their capital into the company. Their $1bn in cash (slightly different from equity) also far outweighs their $55m in total debt, demonstrating again their commitment to strategic use of capital, with debt that can be easily paid off should the need arise.
Stock-Based Compensation < Cash Flows - While many tech companies struggle to retain talent and adequately manage stock-based compensation at the same time, Arista has somehow found a way, likely aided by their strong company culture. While SBC is certainly a factor and grew from $50m to $62m YoY (+24%), their cash flows from operations grew by 72% to ~$375m in that same period. Arista can obviously afford to pay their engineers well with these kind of cash flows while still maintaining a prudent cost structure. So long as the price of retaining Arista’s talented team of innovators is so heavily outweighed by cash from operations, I am extremely comfortable with how management compensates its team.
Despite a couple minor concerns (which are mostly nitpicky), things are looking pretty for Arista overall. All told, they have a beautiful balance sheet, and investors can be confident that their investment in Arista is going to company that knows how to operate efficiently and responsibly. The balance sheet is one of the strongest selling points for Arista.
Key Performance Indicators (KPIs)
Gross Margins - Keeping an eye on how gross margins are impacted over the next several years will be an important indicator for the business. If margins continue to fall, it could indicate several things (again, keep in mind that lower margins are part of their expansion strategy);
Failure on Arista’s part to expand outside large enterprise customers that are able to negotiate discounts for their products.
A significant portion of their revenues coming from hardware sales, and less so from higher-margin services and software.
Competition is causing Arista to lower prices in order to remain an attractive suitor when compared to competitor’s cheaper options.
Conversely, an increase in margins could mean they are successful in these areas of the business and the investment becomes even more attractive to institutional investors, in which case look for the share value to gain significantly.
Service and Software Revenues - This is the higher margin portion of the business, and the differentiator in Arista’s product. It also provides steady and dependable recurring revenues that is less beholden to the choppiness seen in the hardware portion of the business. Look for how this segment of the business grows or shrinks to monitor the vitality of the business.
Global Revenues - If Arista is successful in expanding more into international markets, this investment becomes far more attractive. The size of their TAM will increase considerably, they will be less reliant on the US tech cycle for income, and they will generate far greater revenues while making a name for themselves in regions that are developing tech hubs of their own. I see this as a major driver for the success of the company; if they are able to continue the strong growth numbers (~80%) in these markets, the Arista of a few years from now will be a behemoth in comparison to the Arista of today.
Adjacent Market Penetration - One final KPI to monitor is the success of their offerings outside their core portfolio of switches, including campus, routing, and software/services. Furthermore, if they create further adjacent markets, it points to even larger market potential and revenues, as well as Arista’s ability to continue innovating to serve a larger number of customers.
Defense & Offense
Geographic Expansion - Offense
Only about 30% of Arista’s revenues are derived from international markets, with the other 70% concentrated in the United States. As the world’s tech giant, this concentration makes sense, but there’s certainly lots of room for global growth as data centers, cloud, and AI continue to play more and more vital roles in countries around the world. CEO Jayshree Ullal recently mentioned that international markets are their largest growth driver, with approximately 80% YoY growth. If they can continue to grow in other countries and deliver their systems approach to a wider global market, they will continue to expand their total addressable market and make boat-loads more cash while they’re at it. Geographic expansion is one of the most attractive runways for growing Arista’s business, as well as helping tech hubs around the world create, deliver, and expand technology solutions in developing countries (which I see as a major driver for improving quality of human life and environmental harmony, but that’s a topic for another time).
Steal Market Share - Offense
As shown in the most recent annual reports, Arista has been successful in growing their company in the shadow of Cisco and other incumbent network players, stealing market share in dollars and ports alike and becoming the #1 player in 100/200/400G ports. This demonstrates the quality offering that Arista provides - if it’s a trend they can continue to deliver on, they have the potential to become one of, if not the biggest player in a massive and mission-critical industry to the tech world. The total addressable market is only so big - unlike, say, Google Search, not everyone needs data center switches and networking software solutions. Most companies that do require infrastructure in this market are already buying from one provider or another, so it’s important that Arista continues to innovate, build a strong reputation, and deliver quality products to further steal customers and market share in the existing market in order to grow.
Adjacent Network Markets - Offense
Perhaps Arista’s greater offensive weapon will be their ability to continue developing products for wider use cases, thus expanding into wider realms within networking and increasing the size of their TAM. They have demonstrated this ability in the past with the R-series switches that blend routing and switches (where before they were exclusively switch oriented) and with their campus offerings. While geographic expansion and the theft of market share will undoubtedly help them, they will only increase their ability to take advantage of those offensive weapons by offering a greater range of products and use cases.
Shred the Tailwinds - Offense
AI, the cloud, data. These are some of the buzzwords in our modern internet, and Arista serves as a foundational player to companies offering these services at scale. This provides a lot of tailwinds to both the company and the industry they operate in. Arista started riding these tailwinds before there even were tailwinds, in the very early days of the cloud. This gave them a leg up over the competition when cloud began to gain steam in the tech industry, and has served them again in the development of wider use cases for AI. Given their ability to innovate to take advantage of tech trends before they even start, I’d say Arista has amply demonstrated their drive and ability to ride tailwinds, as well as profit off them.
Innovation - Defense
Every company and their mother says innovation is their key distinguisher, but Arista is one of very few companies that can really flex this with a straight face. Arista’s ability to redefine networking solutions and systems serves as a strong form of defense against encroaching competitors, new entrants in the space, or the ability of incumbents to catch up to Arista. While innovation would have served initially as more of an offensive weapon, they can now use the advantage it has garnered them as a defense to stay ahead of competition. In their short history, they have made several innovations that changed networking and had competitors playing catch-up. And they don’t show any signs of stopping soon - nearly 80% of Arista’s research and development costs go into their software, which is central to their approach to changing network systems and a true differentiator in distinguishing themselves amongst the larger market. So long as they don’t get complacent and continue their trend of innovation, it will be very hard for anyone to steal market share from Arista, while Arista themselves will continue to grow into ever larger shoes.
Switching Costs - Defense
Massive amounts of money go into the equipment and software that Arista provides, and subsequently is not easy or cheap to replace. If, for example, a data center chooses Arista as their provider for a full system solution (software & hardware), they are pretty locked up with Arista based solely on how much it would cost the company to replace all the hardware for another system. The quality of the product Arista provides should discourage customers from wanting to switch anyways, but the cost associated with doing so provides an extra deterrent that serves as a form of defense for Arista. Competitors would have to provide a significantly better and cheaper option to make this re-investment worth it for companies.
To further the conversation on tailwinds that Arista is riding, here is a graphic they provided in their annual presentation demonstrating the size of the market they’re operating in and the growth it is projected to undergo over just the coming 4 years; a $20 bn increase into 2027 represents a nearly 65% increase, providing lots of room for growth even if Arista’s market share remained stagnant over the coming years. This future growth is just a continuation of the previous success for the high-performance switching industry, which in 2012 was projected to 10x to $2.6bn… we’ve come a long way from there to today’s value, and I don’t see it stopping anytime soon with data centers, AI/ML, and the cloud becoming more and more integrated with the daily lives of an ever-expanding number of people around the globe.
Cisco - Cisco is obvious place to start, not only as the primary incumbent that Arista is trying to displace and largest player in the space, but also as the origin point for the creation of Arista (CEO and founders all being former Cisco employees). A darling of the dot-com bubble, Cisco continues to maintain the advantage in the war against Arista, but has lost battle after battle. Nevertheless, the competition presented by Cisco is significant, and Arista is still just a small fish comparatively. At over $200bn market cap and 40+ years of experience in the industry, Cisco is an absolute giant and a strong company in its own right, boasting a 20% ROIC, 61% gross margins, much more reasonable valuations from earnings, sales, and books perspectives, and with more cash to splash to boot (as well as a healthy dividend, if that’s your jazz)
However, growth is Arista’s claim to fame - as a behemoth, Cisco lacks the agility and innovation that Arista boasts, a difference that has resulted in their loss in the battle for the high performance switch market (100-400G). Arista also has an advantage in software-defined networking, which is revolutionizing the networking industry. Though Cisco faces a loosening foothold on Networking Mountain, they certainly aren’t anywhere close to having the rocks slide out from under their feet and taking a long, slow tumble down the hill. They are watching Arista climb towards them rapidly and are likely well aware of the danger that Arista poses to their business. With so much money to spend and the lead in market share, they will continue to pose a threat to Arista’s business. Nevertheless, I think the advantage that Arista has with their EOS will be difficult for any company to replicate - even if Cisco is working on software solutions for their own networking hardware already, they will be playing catch-up to Arista in that department for a considerable amount of time, all while Arista continues to innovate and improve their product offerings to steal more market share.
Juniper - Juniper Networks was another dot.com boom and bust. Like Cisco, Juniper is priced much more reasonably on the valuation front than Arista, but it is a quarter of the size of Arista, with less attractive margins and a much lower return on invested capital. While Juniper is certainly still one of the biggest players in this industry, it just doesn’t seem to hold a candle to Arista’s EOS. It does, however, have a wider range of product offerings and more industry recognition as a result of having operated so much longer. Juniper has also poached one of Arista’s original founding members through an acquisition (that founder being David Cheriton, but more on this later), which could potentially give the company some insight into Arista’s intellectual property and their business operations. While this is certainly a risk, it has been a number of years since Cheriton was involved with the company, and the technology & offerings have likely evolved far enough past that point so as to minimize this particular risk.
Huawei - No stock chart on this one, as Huawei is not a publicly traded company. This is the final competitor I’ll touch on, and I’ve saved the crappiest… ahem, I mean least threatening company for last (I’m still a little sour on the hunk of junk I used to own that Huawei tried to pass off as a phone). From my research, it seem that Huawei has a fairly significant foothold in the international market for switches and networking hardware. It is also recognized by the Gartner Magic Quadrant for Data Center Networking as a Challenger (as opposed to Juniper, Cisco, and Arista, which are all recognized as Leaders) - if you are not familiar with the Magic Quadrant, it offers an excellent snapshot of an industry and the various players within it, as well as their respective positions;
Huawei II - So, while Huawei is a large company (estimated around $155 bn) with a significant international presence, their interests are spread across a variety of products. They don’t seem to have a true competitor for Arista’s EOS - furthermore, my research gathered from the trustiest of sources, Reddit, revealed an astonishing amount of real networking experts (a group that certainly doesn’t include myself, so take everything I say on this front with a grain of salt) that warned vehemently against purchasing or using Huawei products for networking needs. And if you can’t trust Reddit… well, your trust issues are probably valid there, but as I am a complete rookie in this field, so I still thought it was valid enough to throw in. This is an industry where enterprises are willing to splash for higher-end products that will ensure low-latency and high-reliability - anything below this standard can potentially cost companies millions of dollars, so I don’t see Huawei being much of a competitor to Arista, at least in terms of quality. The other knock against Huawei in this race is the current geopolitical tensions between China and the U.S., where a recent ban of advanced chips to China revealed the States’ ability and willingness to play dirty in the tech race. If this spreads to networking system offerings like Arista’s, Huawei could offer even less threat to Arista.
To summarize the competition front, both Cisco and Juniper Networks have a far broader range of products available, and thus serve a wider range of customers. They have also been around far longer and built a more recognizable brand name for themselves; whereas almost everyone has seen a piece of Cisco hardware at some point in their life, Arista has nowhere near the household recognizability on their brand. Huawei, meanwhile, holds a significant advantage in the international markets that Arista is trying to grow into. While Arista’s EOS and the immense quality of their products certainly gives them an edge against these three competitors, they have a significant amount of work to do in order to fulfill the lofty expectations baked into the current stock value. If competitors are able to stall them in their goal to steal market share, catch up or even leap frog Arista’s software capabilities, or dominate the rest of the market to a degree in which Arista is not able to penetrate adjacent markets, then Arista could become just another player in a saturated and highly competitive space. Though unlikely to go completely under, it could stall growth significantly enough to reveal this as an enormously overvalued stock - this is the biggest threat to Arista as I see it today.
The CEO and management team bear a large influence on whether I make an investment or not - Glassdoor helps to gain an insight into what employees think of the company and management for Arista;
The company as a whole has a 4/5 rating, with an 88% approval rating on CEO Jayshree Ullal. For such a large company (and almost 1000 reviews), this is extremely impressive. Work-life balance was by far the most mentioned pro for working at the company, and a large portion of the reviews also talked about how great the culture at the company is. This is foundationally supported by ‘The Arista Way’, the name the company created for its culture of putting customers, employees, and shareholders first and foremost.
Jayshree Ullal previously worked at Cisco and AMD, and is an engineer herself, though her role as CEO of Arista is far more business- and front-facing oriented, where she brings experience in networking market dynamics and customer relationships to fruit for the company. She achieved profitability very rapidly after taking over the helm and helped the company undergo its IPO in 2014. Ullal has demonstrated an impressive capacity to create an enjoyable work environment while leading a innovative, disruptive company to success against rival S&P 500 enterprises - no easy feat. Her presence in interviews, videos, and other media is remarkable and she manages to avoid giving me the icky, corporate feeling that many CEOs I research do - this is perhaps most impressive. Were I to work at a company like Arista, I would feel much better having a real human in charge of me than a robot-like figure.
My only concern with Ullal is her age - she is 62 (which I almost can’t believe - she must have found the fountain of youth sometime before starting at Arista) and nearing the sort of age where people move onto the next stage of their lives. She certainly doesn’t need the money, at an approximately $2bn net worth. Hopefully she has had a strong enough effect on the company culture to pave the path for a worthy successor to maintain her presence even after she is gone. Whoever does succeed her as the company’s leader will certainly have big shoes to fill - this is not an industry or company that can be headed by an idiot and still succeed, which goes against Buffett’s old adage that you should only invest in companies that can be run by an idiot, because one day they will be (I’m actually not sure that was Buffett, I think most semi-intelligent investing quotes are blindly attributed to him, but someone definitely said it).
We love founder-led teams, and so should you. There is research to suggest founder-led teams outperform the wider market, and there is something to be said about the pride and dedication that comes with starting a company from scratch and wanting to see the very best for it. Kenneth Duda, Andy Bechtolsheim, and David Cheriton, Arista’s three founders, are thankfully still very much involved with the running of this company. I say thankfully because this is one of the most impressive groups of founders I have ever come across. They boast some seriously remarkable resumes, and they are exactly the kind of leaders I want to see heading a company for a variety of reasons;
Ken Duda is an inventor and software engineer that joined original founders Bechtolsheim and Cheriton a few months in to the making of Arista, and now heads the engineering team as CTO. Ken is also a huge nerd. Seriously, I mean this in the nicest way, but watch a video of him explaining some of Arista’s products and you will see what I mean. He is obviously passionate about networking and Arista as a whole, and has to be one of the most instantly lovable nerds I’ve seen in my entire life. He’s exactly the sort of incredibly intelligent and product-focused leader you want to be involved in the founding of a complex and highly technical company that you are invested in. Peter Thiel’s ‘Zero to One’ book on entrepreneurship has a great chapter on what he looks for in founders, and Ken Duda is the prototype example of it. **note: if entrepreneurship or learning more about what sets apart successful businesses interests you, I highly recommend the read**
Andy Bechtolsheim has some extremely high praise from some very intelligent and likely very difficult-to-impress people that includes Pat Gelsinger (Intel’s CEO), who described him as “Michaelangelo - a true artist in the area of systems design”. Andy is the CDO (Chief Development Officer) and serves to continuously drive Arista’s growth and innovation to new heights, including the use of silicon in Arista products to propel use cases in the AI/ML realm. Bechtolsheim and Cheriton first met at university, where Bechtolsheim was already tinkering with networking hardware that he called SUN (Stanford University Network). This later came to be the beginning of Sun Microsystems, a highly successful company that Bechtolsheim helped to found before leaving to - wait for it - start two more highly successful networking startups, this time with the help of David Cheriton, that were sold to Sun Microsystems and Cisco respectively. The fourth time was the charm for Bechtolsheim, as the fourth time was the creation of Arista, where he remains to this day. Interestingly, he and David Cheriton were also the first people to write a check for the founders of Google, an investment that, it is safe to say, has earned them a boat-load of cash. Bechtolsheim is an incredibly wealthy man, and I don’t think Arista is so much a project of wealth-building for him as it is a passion project - investors should see him as very much aligned with their own interests of seeing the company succeed in enormous fashion. If that isn’t enough to convince you, then perhaps the fact that the entirety of his stake in Arista is set aside for his children will.
David Cheriton is not an easy man to find information on. That seems to be how he prefers it. He and Bechtolsheim have a long history of success in the networking industry (and investing, in the case of Google) together, and it was partly thanks to Cheriton’s pocketbooks and software expertise that Arista was able to start without constraints from conventional venture capital. He served as the chief scientist with the company. Unfortunately, Cheriton resigned from Arista’s board the year the company IPO’d and sued the company for a (petty) dispute over the use of a programming language Cheriton had helped develop as head of another company. This case is still ongoing. He later founded Apstra, another company in the networking realm - unfortunately, this later led to him turning traitor (kidding) when Juniper Networks acquired his company and brought him on as Chief Data Center Scientist. I guess there has to be a black sheep in every family. Despite the lawsuit and departure, Cheriton and Bechtolsheim remain friends.
What I find most impressive concerning the three founders was their ability to put Arista first by not only stepping aside for but also actively recruiting Jayshree Ullal to come and head up the business-side of the company. Many founders would have refused outright to give up the top-job, and other companies with three founders may even have had some internal conflict between the founding team on which of the three would take it. Not so with these three, and not so with Arista. Duda, Cheriton, and Bechtolsheim knew where their competencies and strengths lay, and encouraged someone else to come in and take the Chief Executive role to fill in the blanks. I find this extremely admirable, and speaks a lot to the ‘Arista Way’ culture that lays at the foundation of the company to drive such positive views from Arista’s employees.
Here’s a brief snapshot of Arista the stock (as opposed to Arista the business). The capital efficiency, structure, and margin sections look highly enticing to me, as does the juicy stock chart. This overview was taken from Stratosphere.io - I highly recommend this service if you are looking to continue research into Arista or any other company, as even the free version provides quick and easy access to financial data for most companies you could dream about dreaming about.
Everything has seemed pretty aces up Arista Avenue so far. Unfortunately, this is where the avenue takes a southward turn. It’s difficult to keep a company like Arista quiet - with a strong balance sheet, profitability, and rapid growth in an evolving industry, it is bound to be richly valued. And so it is.
Investors today are paying a premium price for Arista, with a price nearly 10x its revenues and book value, as well as a rich P/E ratio. While the P/S ratio and P/B are relatively in line with Arista’s historical average (just slightly higher), the current price relative to earnings is actually lower than the historical average of 37 since 2014, suggesting that current prices may present a bargain for such an attractive investment.
One of my major concerns for Arista’s valuation is when the price is put into the context of the TAM that Arista operates in - estimated to be approximately $51bn by 2027, the size of the total market is roughly in line with Arista’s market cap today. This could suggest a pretty limited ability for upwards momentum from here unless Arista is able to take advantage of some other markets in addition to their core offerings.
One of the few potential bright spots in the valuation realm, beyond a lower-than-average P/E, is the PEG ratio - this metric helps to get a sense of fair value relative to the earnings per share. I far prefer this metric to the basic P/E as it helps to price in growth and any share buybacks the company undertakes as well. The PEG for Arista is 0.67 (1 being fair value, <1 being undervalued, and >1 being overvalued). Now this looks enticing due to the way this ratio is calculated and the fact that Arista had quite a large jump in EPS over the last year, so I calculated a PEG ratio based on the average growth in EPS since 2014 (53%) - using this amount and projecting it out into the coming year, I came to a forward-PEG ratio of 0.44, which is even more attractive.
The other bright spot is the EV/EBITDA ratio, which looks at the value of the company (including debt, which market cap does not figure in) in relation to the earnings before interest, taxes, depreciation, and amortization, a popular metric to consider when analyzing a growing company. While still a high value at 26ish, this is again roughly at or just below the historical average.
Though very richly valued in many senses, particularly in relation to revenues, Arista is potentially undervalued when looking at the earnings they return for every share, and fairly valued from a historical context in several metrics as well. Nevertheless, it’s important to consider that such rich valuations could deliver underwhelming results compared to the rest of the market; if the company fails to perform perfectly (as it is essentially priced to perfection), then analyst’s models are changed, price targets are dropped accordingly, and shareholders may underperform compared to the wider market as a result. Especially given that hype cycles are buzzing on AI right now, the share price may be inflated as a result of investors looking for a way to invest in the space - if true, any die-down in investor excitement could depreciate the price. When considering conventionally overvalued stocks, there is often far more downside potential than upside in the short-term when a stock is priced so richly - in the long-term, this valuation could begin to matter less if Arista continues proving itself as a strong company (think about the fact that Amazon would have been considered conventionally overvalued through much of its history, but that hasn’t stopped people from securing multi-baggers by holding for long periods).
I always take analyst price targets with a pretty large grain of salt, as they are very short-term in focus and many analysts recommending a stock may be biased if they are owners of the company themselves and are hoping to generate more investor interest. Nevertheless, it is good to see what the street is thinking about a company and what they see the future of Arista looking like. Here’s a snapshot of a chart based on recommendations from 23 analysts from the NASDAQ website, where the stock is rated as a Strong Buy:
At a price of $155.86 at the time of writing, analyst projections call for -17%, 11%, and 28% returns at the low, average, and high targets respectively. While a 17% loss is obviously not ideal, even the average price target could generate a +10% return (keep in mind that the 30-year average return on an S&P 500 index fund is ~10% - anything above that is what I generally regard as market-beating returns).
While not anything insane for upside opportunity in the near-term, in the long-term I believe Arista has significant potential to appreciate in share value and begin returning cash back to shareholders, either in the form of a dividend (as Cisco has) or through share buybacks.
Threats to the Share Performance
Supply Chain - Disruptions or constraints on the supply chain have the potential to wreak significant impact to Arista. If they or their manufacturers are not able to obtain the materials they need, it could have a significant impact to their revenue generation, and any unresolved, long-term issues could threaten the company in a more serious and lasting way. Many of the individual components within Arista products are sourced from “sole or limited” suppliers, which furthers their reliance on a healthy and efficient supply chain to maintain steady revenue growth.
Customer Concentration - I won’t get into this too much, as I’ve discussed it a few times above. As established earlier, if there is any significant change to the amounts that Meta or Microsoft (or both) are spending on Arista products, the concentration of their revenues in these two customers will come back to bite them and subsequently impact share value.
Competition in the SDN Space - Arista has essentially pioneered the networking industry towards software-defined networking. This has given them a solid head start when compared to other competitors in the industry, but this is capitalism we’re talking about. Businesses rarely maintain their advantage forever. If Cisco, Juniper, or any other new entrant is able to make significant headway on creating a networking operating system to rival Arista’s, this would spell a serious threat to Arista and could potentially send them into a margin spiral in which they must constantly cut costs in order to stay competitive and not lose customers. While switching costs allow some form of defense against this, if there is a significantly cheaper and better product somewhere in the market, it becomes more enticing for customers to choose competitors and replace their Arista infrastructure. I see this as the biggest threat to Arista’s share performance, and it will be important to monitor the networking industry as a whole to maintain some sense of where competitors are in creating a rival system. **note: competitors in the space definitely have some software already created, but at this point do not hold a handle to Arista’s EOS**
Yep, I’m pretty much going to leave this empty. There is little in the way of red flags on the balance sheet or in the company structure, and I have already delved into the valuation and the possible risks that could stem from such lofty expectations. This industry is especially volatile and cruel to wildly overvalued companies - if you bought Cisco and/or Juniper at the height of their dot.com prices, you still would not have made your money back. For Arista’s closest competitors to be the poster children of the dangers of inflated prices hits a little close to home, but this is the only red flag I can possibly come up with beyond the possible threats to their share price discussed above.
Gartner Magic Quadrant Leader - As discussed in the ‘Competition’ section, Arista was named a leader in the networking industry for the Gartner Magic Quadrant, a position it has held for several consecutive years. This is very much a green flag for me - Gartner is an objective viewer in this race, so for them to consistently name Arista as a dominant player in the industry speaks to their position and progress. So long as they continue to maintain this in future years, I see this as very encouraging sign for the quality of the company.
Trusted & Fastest Growing Network Player - The fact that Arista is the go-to supplier for several top-of-the-line tech companies, the #1 player in the high-performance switch market, and long-standing partners with a prestigious list of companies shows the trust that other enterprises have in Arista’s products. The company also revealed recently that they have a 92% customer satisfaction rate - a wildly impressive clip for such complex systems. As time goes on and this trust expands to a greater number of customers, Arista will become more entrenched in the industry.
I hope this has given you at least a launching pad for some future interest in Arista Networks. While I tried my best to give you a good sense of the investment and company, I struggled immensely to really dig deep on the products they offer. While it is clear their software offerings through CloudVision and EOS are what differentiate them from other networking competitors, I don’t have the industry expertise to delve into how these efficiencies are derived or the strength of the moat these may or may not deliver - I can only expound on the basic features their software and hardware provide: speed, reliability, openness, automation, and security, and the fact that all these features are critical for the customers Arista serves. Hence why we have seen the immense success of Arista, going from a tiny fish in the pond to the fastest-growing player in the networking space in the span of 9 years since IPO.
Beyond their products, the quality of their leadership as a founder-led team is beyond impressive - you’d be hard-pressed to find a more capable and passionate bunch of leaders that are so aligned with shareholders. Their passion for Arista is infectious, and it’s clear to me the culture they have created through the Arista Way has benefitted the company, customers, and owners of the stock immensely. While their valuations are rich today, I believe that if they don’t rest up on the innovation front they can grow into their shoes by expanding into adjacent networking and international markets, stealing market share from incumbents, and riding the coat tails of a rapidly expanding industry that may come to rely heavily on Arista products. By doing so, Arista can demonstrate that there is a very good reason for their high expectations and handily reward those with the patience to stick with them for the long-haul. As a shareholder and technological enthusiast, I am beyond excited for the long-term growth potential of this company.
Now, to hand out a final grade to Arista:
Final Grade: A
Hope you enjoyed this article on Arista. Happy investing folks!
Thanks for reading Hourglass Investing ! Subscribe for free to receive new posts and support my work.