Arista Networks' (ANET -5.12%) stock fell 24% after management reported earnings on Oct. 31. And since Arista is a recommendation in both Motley Fool Stock Advisor and Rule Breakers (as well as 11 other Foolish services), we thought it was a good time to do a deep-dive question-and-answer session about the stock.
John Rotonti, a senior analyst at The Fool and an Arista bull, pulled together a list of important questions in early November that he thinks investors should be asking about the company, and then he answered them for the readers on fool.com! Here's his report.
Q. What is Arista's history? How and when did it get started?
According to the book Cloud Networking Transformation: The Making of Arista, Google (today a part of Alphabet (GOOG 0.33%) (GOOGL 0.30%)) put out a request for information in 2004. It was looking for a switch design that could tie together 100,000 servers at 1 Gigabit per second (1Gb/s, or 1G) ethernet speeds at a cost of $100 per server. (At the time, 1G was the standard ethernet speed for data center switches.)
If that 2004 Google request sounded complicated, don't worry. The main takeaway is that such a product did not exist. So Andy Bechtolsheim (co-founder of Sun Microsystems and the first outside investor in Google) put together a team -- including co-founders David Cheriton and Ken Duda -- to deliver on Google's request.
In 2008, the first product was created, and the business was ready to go to market. So the three co-founders appointed Jayshree Ullal, former Cisco Systems (CSCO -0.35%) executive, as president and CEO of Arista.
Arista launched its first switch product in 2010, and that device offered six times the throughput of anything else on the market then. The company first supplied its switches to high-frequency trading firms to showcase their speed and low latency. Arista's success with the Wall Street trading firms proved that the technology was next-gen and highly capable, and Microsoft (MSFT -0.10%) became its first cloud customer.
Together, Arista and Microsoft created the blueprint for hyperscale cloud networking, and other cloud titans soon became Arista customers as well. Arista first helped build Microsoft's cloud infrastructure platform, Azure, on 40G switches and then transitioned to 100G switches.
It is safe to say that the cloud would not exist at the scale that it does today without Arista, a name that means "excellent" in Greek.
Q. You mentioned switches. Before we move on, briefly explain the difference between a switch and a router.
At a basic level, switches connect servers and other devices to a network, and routers connect networks together and connect networks to the internet. Because a router is a device to connect to the internet, it has many more addressable destinations (or routes) than do switches that just connect a network within a single company. But I wouldn't get too bogged down in these terms, because networking equipment is starting to overlap.
For example, Arista offers a flex route license, which turns a switch into a router and increases the number of routes from roughly 256,000 to more than 2 million. So Arista's customers can purchase one piece of networking equipment that serves as both a switch and a router, thereby reducing inventory and complexity.
Q. What is Arista's business model?
At a high level, cloud software is distributed across hundreds of thousands of servers, and those servers are connected to each other (and connected to the Internet) with high-speed switches and routers. Arista sells software-driven networking equipment (the switches and routers) that allows the cloud to function.
Arista's switching and routing platforms have high capacity, low latency, port density, and power efficiency. Arista's key innovation is its advanced networking operating system (called Extensible Operating System, or EOS) that works across different cloud networks (public, private, and hybrid) and is based on an open architecture (open Linux), allowing for integration with third-party software providers. Arista's software-infused hardware works seamlessly on Amazon Web Services, Microsoft Azure, Facebook (META -0.70%) Cloud, Google Cloud, Oracle Cloud, and IBM Cloud. For security, Arista partners with several leading providers such as Palo Alto Networks, Zscaler, and VMware. Having an open architecture also allows Arista's customers to integrate whichever cybersecurity systems they think are best for protecting their networks.
Arista also uses the same software version across all its network switches and routers, which improves reliability and reduces costs and complexity (meaning fewer switches are needed to achieve massive scale, and fewer switches means lower cost of ownership). This is different from legacy network equipment vendors that use different operating systems for different items of hardware in the network, adding layers of complexity and making software updates and maintenance difficult. Arista's use of one software version across the whole network makes software upgrades and maintenance simple, rapid, and automated.
Arista's operating system allows for robust data analytics and diagnostics and automatically gets to work fixing small problems (bugs) before they become big problems (this allows Arista to prevent network outages). In this way, it is self-healing. Arista's automated diagnostics (self-healing capabilities) are much more efficient than having a quality-assurance team manually working to find and fix a bug. In other words, Arista's network technology saves time and money (because fewer network personnel are required to be on call to monitor thousands of servers). Because Arista's software allows for easier maintenance and updates, tighter security, and fewer switching tiers and personnel, it provides its clients with a lower total cost of ownership compared to legacy vendors.
Arista also uses a wide variety of best-of-breed, off-the-shelf merchant silicon architectures (from leading providers like Broadcom and Intel) as opposed to legacy providers that spend lots of time and money making custom chips. Arista has found that the off-the-shelf products perform just as well as (and possibly better than) the custom chips, and this approach allows Arista to:
- Focus all its resources on engineering the best networking software in the world (rather than trying to make custom semiconductors)
- Cut costs
- Bring new products to market faster
Arista's use of many different merchant silicon chips reinforces its approach to an open system and allows it to create a broader range of products more quickly.
Q. That sounds pretty technical. So what does all this mean?
It means that Arista has a unique business model and engineers best-in-class networking technology. Arista has been recognized by Gartner as a Magic Quadrant leader for the fifth year in a row and is also a technology leader, according to Forrester.
Q. But do its unique business model and proprietary technology culminate in superior financial performance?
Yes, most definitely. Arista's product and customer obsessions have resulted in rapid market share gains and extremely high (and rising) returns on invested capital (ROIC) and free cash flow (FCF) margins.
Q. Wow, Arista generates ROIC of more than 100%. How does it do that, and will it be able to sustain high returns in the future?
Arista is able to generate such high ROICs because it is highly profitable (the numerator in the ROIC calculation) and has an asset-light business model (with an average capital expenditures-to-sales ratio of roughly 2%) and cash-rich balance sheet (net cash accounts for a whopping 65% of Arista's total assets).
New Constructs subtracts excess cash to calculate invested capital (which is the denominator in the ROIC equation). So a high numerator and smaller denominator result in ROIC of more than 100%. That is the accounting answer to the question.
But it should also be said that Arista is able to generate such high profits and cash flows because it benefits from several sources of competitive advantage, including its best-in-class proprietary technology and "the best networking software team in the industry." This brand makes Arista a magnet to attract the best and brightest networking talent. If you are a networking buff, you most likely want to work alongside Bechtolsheim and the Arista team!
Arista also has very close relationships with the cloud titans and provides top-notch customer service. The company literally engineers and develops products right alongside its biggest customers, providing Arista with a feedback loop to continually innovate to meet customer expectations. Arista is such a critical component (and partner) that it is helping to plan out its customers' long-term technology and cloud roadmaps. Arista's customers have 24/7 direct access to Arista's technology development team, meaning that the same engineers that code the product are the ones testing it, troubleshooting it, and repairing it.
Finally, Arista has a culture of innovation and adaptation, something that I really try to zero in on using the complexity investing framework. This means that Arista is committed to finding new sources of growth and new sources of moat over time, which should help it sustain its fantastic financials for some time to come.
In the forward to The Making of Arista, Jayshree Ullal writes:
Great leaders and their companies never rest on their laurels. They constantly question themselves, regardless of past glories. They don't adapt to change; they are the change. They create the inflection points that new start-ups take advantage of and existing companies use to reinvent themselves.
The book closes by saying:
We are not standing still. Each year, our engineers reinvent our software stack, utilizing the latest developments in computer science. ... In the making of Arista, every aspect of traditional technology has been made smarter, challenged or eliminated. This dynamic and on-going transformation is core to Arista's entrepreneurial culture and is central to delivering Arista's customers with the opportunity to drive the leading edge of innovations. And we won't stop.
Q. You mentioned Arista has a ton of cash. Does Arista have a strong balance sheet, and do you agree with its balance sheet strategy?
Arista has net cash of $2.3 billion, which accounts for 65% of its total assets and 16% of its market cap (of $14.7 billion as of Nov. 11, 2019). That's quite the treasure chest, and it is largely a manifestation of Arista's copious free cash flow generation.
Through 2018, I thought that Arista was holding onto excess cash partly because of an outstanding lawsuit filed by Cisco claiming IP infringement. But Arista paid $400 million to settle that lawsuit back in August of 2018, and both sides agreed that "no new litigation will be brought over patents or copyrights related to existing products, for five years."
But now that there seems to be a low risk of future litigation with Cisco, I still agree with Arista's strategy of maintaining a large net cash position. The company operates in the hardware industry with relatively short product life cycles and a lack of revenue visibility, so the cash can provide the fuel to maintain investment spending during an industry-specific or broader economic downturn. In short, the cash provides Arista with the optionality to find and deliver on its next stage of growth.
Q. You mentioned growth. Where will Arista's future growth come from?
Generally, Arista's growth is powered by the transition to a digital economy and cloud computing. Companies are continuing to shift computing infrastructure, applications, and workloads to the cloud, and technologies such as artificial intelligence (AI), machine learning, 5G telecommunications, and the Internet of Things (IoT) are all driving the demand for more switching and routing. Basically, anything that drives bandwidth and interconnectivity is a driver of demand for Arista. And as computing moves closer to the "edge" of the cloud, that also benefits Arista, because those connections are still done with networking equipment, and Arista's 400G ethernet switches (more on this below) should facilitate edge computing because of faster speeds and lower latency.
In the medium term, Arista's growth is powered by:
- The transition from the current standard 100G ethernet switches (developed around 2016) to 400G, which is still in its infancy. The rollout is really supposed to ramp up in the second half of 2020 and into 2021. Basically, 400G switches deliver four times the network scale and density, two times power efficiency, but at two times the price. But over time, I suspect the price will come down dramatically (just as we saw with the switch of standards from 40G to 100G).
- Growth should also come as Arista starts to sell more products to "campus" cloud networks. Think of a data center as a room (or building) packed full of servers and related equipment, and think of the campus as all of the connectivity that happens outside of the server room in the rest of the office building or office campus. Another way to think about it is that the data center is where the servers connect to the network, and the campus is where employees (and their related devices) connect to the network. Arista is on track to reach $100 million in campus sales in its first full year of shipments. Over the last 10 years, most of the innovation in networking and tech took place in the data center, so that is where the opportunity started for Arista. Before now, the biggest innovation in the campus was wireless connectivity, but now the campus needs to play catch-up, so Arista is helping to extend the data center cloud technology throughout the campus. Ullal said that Arista is currently sowing the seeds for an eventual $500 million to $1 billion-a-year campus segment but said it could take some time to get to that point.
- International growth also presents an attractive opportunity. In the third quarter of 2019, only about 20% of sales came from outside of the Americas. But the cloud titans are aggressively building data centers all over the world, and Arista's customers will typically deploy Arista products internationally for easy setup, monitoring, and maintenance.
Q. Why did the stock sell off 24% after the company reported its third-quarter 2019 earnings?
Arista's stock did fall 24% after reporting of third-quarter 2019 results on Oct. 31. But the stock was under pressure even before that. At $191 a share, Arista's stock is down 42% from its 52-week high of $331 a share.
Here is the short version of a long story about the drop. One very large customer (presumably Microsoft) paused its spending with Arista toward the end of the first quarter of 2019, but that customer has since resumed spending with Arista. Then, more recently, a second cloud titan (presumably Facebook) notified Arista that it would reduce spending in the fourth quarter of 2019 and in 2020. Facebook accounts for at least 10% of Arista's sales, so its spending actions have a large effect on Arista's overall results. So Arista significantly lowered guidance for this year's fourth quarter, sending the stock down 24%.
Here's what might be going on. It's not inconceivable that some of the cloud titans shift their capital expenditures to other parts of cloud infrastructure as they wait for the rollout of 400G switches in the second half of 2020 (and really into 2021). The cloud titans need to update their servers before they update to 400G switches that connect those servers in the cloud. So that timeline makes sense to me.
Investors should note that this is not an Arista-specific problem. Industry leader Cisco also reported that its customers have paused their spending on networking equipment. The effect of the pause is just more pronounced at Arista because of its large-customer concentration.
As for Facebook, I suspect (really a guess) that the social media network is shifting all resources (time, energy, focus, and even money) to protecting the platform from cyberhacks and other interference during the 2020 presidential campaign. I think Facebook believes it can delay some spending on the network for a year but that it absolutely cannot mess up another presidential election. A whole lot for Facebook rides on it getting this right. That's why Facebook suddenly shifted its spending pattern specifically in 2020 with no warning.
And something else probably caught investors' eyes. Arista is going to have to increase sales and marketing (operating expenses) to expand into the campus market, which could put modest pressure on operating margins. So at the same time that sales are slowing (see the chart below), operating expenses as a percentage of sales are likely going to rise, and that double whammy has investors questioning the multiple they want to pay for Arista's stock. From 2015 through 2018, Arista traded at an average forward P/E ratio of 33, according to S&P Global. Today it trades at a forward P/E of only 23.
Although some high-growth investors dumped the stock after recent earnings, I bet some other thoughtful investors were thinking: "OK, so here's a business that is literally providing the software-driven infrastructure that is powering hyperscale cloud growth and the digital economy but whose growth rate has shifted down a few gears, and that slowdown may only be temporary, with a P/E that is trading 10 points lower than its historical average. And this is a company with $2.34 billion in net cash and an estimated 80.3 million diluted shares outstanding by the fourth quarter of this year, so it's got $29 in net cash per share. If I take the current stock price of $191 and subtract the $29 (to get $162) and use the next-12-month consensus EPS estimate of $8.57, then we're looking at one of the great tech companies of the last 10 years, run by the best networking management team in the world, trading at a forward P/E (ex-cash) of only 19 times. Great businesses find ways to grow profitably, and when growth eventually slows, they find new growth curves. So in hindsight, Arista at 19 times earnings may prove to be quite the bargain."
A P/E (ex-cash) of 19 is still not considered "cheap" enough for some, but I bet they are still taking a long, hard look at Arista at these prices.
Q. What should investors monitor going forward?
First, I'll be checking to see if Arista maintains its track record of market share gains (which signifies that it's innovating at a high rate and that its products remain highly relevant and in demand and that its sales strategy is optimized for the campus opportunity). I'm hoping that these market share gains translate into overall top-line growth of at least 9% (and hopefully as high as the mid-teens, percentage-wise).
Second, Arista made its first two acquisitions in 2018 (Mojo and Metamako) to expand its networking cloud product offering. Combined, it spent $97 million on these two acquisitions. I would like to learn more about its acquisition strategy and the financial criteria (and hurdles) that it will use to measure success. Arista is relatively inexperienced with acquisitions, but I do understand their importance for adding key talent and technologies. And, of course, acquisitions can be used to enhance the product or service and can drive long-term organic growth. Arista's management team is very good, so I suspect they will proceed methodically and prudently.
Third, there is no doubt that Arista's culture is based on innovation, and its long-term success largely depends on its innovation engine. I am in no way questioning the networking genius of Bechtolsheim, Cheriton, and Duda or the awesome leadership of Ullal. But I will point out that its ratio of research and development (R&D) as a percentage of sales has fallen every year since 2014 and fell to as low as 16% in the third quarter of 2019. R&D is a prime driver of stock returns, so I would just like to learn more about why Arista's R&D-to-sales ratio is falling given that innovation is in the company's corporate DNA.
Metric | 2014 | 2015 | 2016 | 2017 | 2018 | Average |
---|---|---|---|---|---|---|
Sales (in millions $) | $584.1 | $837.6 | $1,129.2 | $1,646.2 | $2,151.4 | 38.5%* |
YOY sales growth | 43% | 35% | 46% | 31% | ||
R&D (in millions $) | $148.9 | $209.4 | $273.6 | $349.6 | $442.5 | |
YOY R&D growth | 41% | 31% | 28% | 27% | ||
R&D/sales ratio | 25.5% | 25% | 24.2% | 21.2% | 20.6% | 23.3% |
Capital expenditures (in millions $) | $13.1 | $20 | $21.4 | $15.3 | $23.8 | |
Capital expenditures/sales ratio | 2.2% | 2.4% | 1.9% | 0.9% | 1.1% | 1.7% |
Q. What are the risks to the story?
I think the biggest risk for the company right now is personnel related. It's not a stretch to say that Andy Bechtolsheim is possibly the best networking engineer on the planet. And David Cheriton, Ken Duda, and Jayshree Ullal each play critical roles. Separately, they are each special talents and high performers and would be difficult to replace. Together, they create a networking corporate dream team.
Customer concentration is also a real risk, as we saw with the third-quarter sell-off. Microsoft and Facebook each account for at least 10% of Arista's total revenue. This gives Microsoft and other large customers leverage over Arista when it comes to negotiating pricing, and losing any one of these large customers would hurt Arista's business severely.
Related to customer concentration, Arista doesn't have any long-term purchase commitments from customers, and customers can cancel at any time. Arista's sales are surely driven by long-term trends, but visibility in the short term is very low, and investors could be in for occasional shocks (again, as we saw in the recent quarter).
A final related point is that Arista already has a lot of business from the hyperscale cloud players, so I worry that there are not enough large customers left to continue fueling high sales growth. Supplier concentration risk and the ever-present risk of product obsolescence are also out there, but these don't concern me as much.
Q. You've laid out why Arista is a very high-quality, growing business. How do you rate management?
As I've already noted, this is a corporate dream team, with Andy Bechtolsheim as the spiritual leader. Each key person brings a unique skill set that compliments the others. There is so much to like, but I'll point out two things that I love.
First, Arista's mission is to reinvent the network, and Arista's founders and management team have spent the last decade putting in place an organizational structure, processes, and corporate culture that are 100% geared in that direction. Arista is a relatively young company with an entrepreneurial and innovating spirit, and rather than being a large, bloated conglomerate, it is one innovative and agile company focused on trying to make networks better over the long term.
Next, Arista's management team is focused on driving long-term profitable growth. The emphasis here is on profitability. The company is investing to drive long-term revenue growth, but it does so in a way that balances profits and ensures that the company is self-funding, with a rock-solid balance sheet. And Arista is vocal about its laser focus on profitable growth.
- Jayshree Ullal starts every earnings call by highlighting Arista's "profitability growth combination."
- The company's proxy statement says that the executive compensation program is designed to "drive the development of a successful and profitable business."
- In the very first paragraph of the Business Description in the company's 10-K, Arista says, "We have been profitable and cash flow positive for each year since 2010." This is clearly a company that prioritizes profitable growth.
Of course, there are always areas for improvement. Arista's business is highly technical, and Arista's management team often explains it using technical jargon. I think some investors would appreciate it if Arista would take the time during earnings calls or presentations to explain some of the more technical stuff in a way that nonengineers can understand and incorporate into their analysis on the company. I think this may invite and welcome a larger investor base.
Also, because visibility into its customers' buying patterns is only one or two quarters, I think that Arista should consider abandoning quarterly guidance. This could help to train investors to focus more on the long-term trajectory and opportunity of this great business rather than on the quarterly bumps in the road.
Finally, I would love to see Arista publish its first Sustainability Report and spend more time discussing its environmental, social, and governance (ESG) initiatives.
Q. So what is Arista's stock worth?
I'm going to pass on pretending that I can use a discounted cash flow (DCF) model to estimate Arista's fair value. Here's why: From 2014 (the year of Arista's IPO) through 2018, Arista's revenue increased at a compounded annual growth rate (CAGR) of 38.5%. Then more recently, Arista said that it thinks it can grow revenue at a mid-teens CAGR over the next three years if cloud spending remains muted and thinks it can increase revenue at a high-teens CAGR if cloud spending picks back up. (And Arista thinks it can achieve an operating margin of around 35% over that time.)
So Arista is expecting a top-line growth rate that is roughly half (or even less than half) of its historical average. But these mid- to high-teens revenue growth estimates do not apply to 2020 anymore. The company pulled back on that guidance for 2020 because Facebook (we presume) surprised Arista by significantly cutting spending with Arista for the fourth quarter of 2019 and throughout 2020. In fact, Jayshree Ullal said a mid-single-digit top-line growth figure for 2020 is not out of the question. There is just too much uncertainty in Arista's top-line growth rate going forward for me to have any conviction in a DCF-based valuation. But that's OK, because a DCF is not my preferred method of valuation anyway.
Arista generated $706 million in free cash flow over the last 12 months, based on New Construct's rigorous calculations. Arista has a market cap of $14.7 billion and $2.34 billion in net cash, so its enterprise value is about $12.4 billion. That means that Arista's FCF yield is nearly 6%. Arista's 6% yield is 3 times the yield on the 10-year U.S. Treasury (which is a proxy for the risk-free rate) and is even more attractive than the roughly 5% FCF yield of the S&P 500.
Also, at a 6% FCF yield, Arista only has to grow organic revenue at 9% per year to generate expected annualized returns of 15% (for simplicity's sake, I'm assuming that Arista's FCF grows roughly in line with its organic revenue growth). Remember that Arista's revenue increased by 31% in 2018 and has increased at a 21% rate over the last 12 months. And if we put 2020 aside (which we know could be a low-growth year), 9% growth is also below the low end of Arista's long-term guidance. Arista's business is still powered by hurricane-force long-term tailwinds, and not a lot has to go right for Arista to generate 9% organic top-line growth over the medium term. That's a bet I'll take all day!
I'll offer one last thought. Arista generated trailing-12-month revenue of about $2.5 billion and has a market value of $14.7 billion. Its prime competitor and industry leader, Cisco, generated trailing revenue of about $52 billion and has a market value of $205 billion, roughly 14 times larger than Arista. Both are innovating on the leading edges of technology and selling into large growing markets. In 5 or 10 years, I think Arista will be multiples larger than it is today.