It’s hard not to get the “FOMO” feeling when reading through a Motley Fool teaser ad, because they always start with their past blockbuster winners — here’s how the latest tease begins:

“Can you imagine if you could go back to 2002, when cloud computing burst onto the scene…

“And invest $10,000 in Amazon at $0.77 a share?

“Or travel back to 2005 and invest the same amount in Nvidia at $1.63.

“Can you imagine how much money you’d be sitting on right now?

“I’ll tell you…you’d be sitting on over $4.5 million today.

“And I know this because we made those stock recommendations.”

Well yes, I guess technically it’s “we” that made those picks — but really it was David Gardner back then, and while I’m sure he has taught the latest generation of “Rule Breaker” analysts, he was a pretty singular stock picker… and he’s not doing it any more. Both David and his brother Tom have mostly stepped back from headlining the Motley Fool’s newsletters and portfolios, so the newsletter recommendations are generally chosen and written up by teams of Fool analysts now.

And the newsletter they’re selling today is the flagship, Motley Fool Stock Advisor, which still follows the same basic strategy (half “rule breaker” ideas, half “everlasting” ideas, the letter started as a showdown between David and Tom, with each picking a stock a month).

The basic spiel is about the drive to build “AI’s Invisible Goldmines”, which are just the hyperscale data centers being built willy-nilly by Microsoft et al as they try to meet the demand for cloud computing and find places to put (and power) all those NVIDIA chips they’re buying to run AI models. From the ad…

“Amazon recently announced plans to invest $35 billion in new data centers in Virginia by 2040.

“Google just announced plans to build a new $1.2 billion data center in Nebraska.

“Meta just announced plans to break ground on a new $1 billion data center in Arizona.

“Microsoft is working to bring a sixth data center to a city in Iowa…

“…And it has plans to build 50 to 100 new data centers a year for the foreseeable future.

“And even Tesla recently announced that it will build a new “1st of its kind” data center to grab its share of this new hyperscale boom.”

So… how does one profit from all this spending by the big tech companies? Obviously buying NVIDIA would have worked absurdly well to this point, but the Foolies are teasing something different…

“… our analysts have found a way.

“They’ve identified one stock that they believe directly benefits from this opportunity.

“It’s the main supplier of a next gen AI hardware that’s needed to run these goldmines….

“It returned 95% in 2023, and a whopping +519% since we first recommended it in Sept. 2017….

“With an estimated total addressable market of $51 billion, this company has only tapped into 10% of its market potential so far.”

Are you getting our free Daily Update
"reveal" emails? If not,
just click here...

And the Fool also had a slightly different version of this ad running recently which was headlined, “Microsoft’s New AI Secret Weapon: Why this obscure stock could be key to Microsoft’s AI domination”… and that version of the ad also included the same hints above, but added a little extra for context. So we’ll cheat a little by combining the two teases into one answer:

“According to Yahoo, this stock provides what’s called network switches that help speed up the communication datacenters need to process up to 25.6 terabits of data per second.

“And these switches are absolutely crucial in harnessing enough power to keep AI datacenters running smoothly.

“Even Forbes has reported that when it comes to AI networking, experts say this company will continue to be one of its biggest beneficiaries.”

Which makes it pretty easy to confirm the Thinkolator’s first answer on this tease: The Motley Fool is (again) teasing Arista Networks (ANET), their longtime favorite data center hardware company. Arista has been a big favorite of Tom Gardner’s for years, and he has both bought the stock for the Fool’s own portfolio and teased it many times since 2017 (we first covered his recommendation back in September of 2017, so he used it as a teaser pick right around when he first recommended it to subscribers). It has also been teased as a “double down” stock and an “all in” stock many times, most recently their teaser ads said they had recommended the stock 14 times across all their services (and the number is probably actually higher than that).

Here’s a mild update of what I shared in the Quick Take the last time I covered a reader question about one of the Fool’s Arista teasers, just a month or so ago:

Arista was one of the stocks Tom Gardner at Motley Fool Stock Advisor was most vocal in extolling in the pre-pandemic era, from 2017-2019, and the questions have started to pile up again recently about this “$1.5 million bet” from the Motley Fool as the ads roll again. Arista Networks is a networking equipment company, they’ve been taking share from Cisco Systems (CSCO) for years with their software-driven switches and more rapid product improvement, by most counts, but they’re also tiny in comparison to CSCO, still, and have suffered some from the vagaries of depending on huge mega-cloud operators for a large portion of their order flow, since Microsoft and Meta alone typically make up more than 40% of their revenue, and those big companies sometimes slow projects or change priorities or otherwise order fewer Arista switches than expected.

I’ve owned ANET in the past, but it’s been several years since I sold on a stop loss during a bad patch from 2019 into 2020 (which turns out to have been a big mistake). Late in 2021, things really started to improve again for ANET, with data center upgrades and spending increasing and driving some good “beat and raise” quarters, though there was also quite a bit of volatility in 2022 as spending plans for the tech giants looked less certain… which brought the stock down to a pretty rational valuation for a while, but then things picked up pretty dramatically with the surge in data center investment to meet AI demand, ANET posted a few more “beat and raise” quarters, and the stock soared past what has been more of an average valuation multiple for them over the years, roughly 35X trailing earnings.

That’s still arguably reasonable, as long as they can keep their earnings growth up above 15% or so for a long time (it has averaged 26% over the past five years, the long-term forecast was 25% to start this year… but the average analyst forecast is now for 10% earnings growth over the next five years), but at that kind of multiple they face the same big picture risks as the other pricy tech stocks. I’m probably a little biased against this one, having lost money on it years ago, and it has had a good run this year that makes the valuation much more challenging than it was when we last checked in, about a year ago, as is now the case with most of the tech/data center/AI stocks, but it’s a strong company that has often shown great growth and it’s a pretty reasonable idea if they can keep growing their market share and finding new customers, and, importantly, if Microsoft and Meta (and others) continue their heavy spending on data centers. Likely to be a bumpy ride, but both Cisco and Arista are saying that demand remains high amid data center and AI upgrades, even if some of the telecom companies might be pulling back a little, and there’s certainly plenty of talk about this being still just the beginning of another upgrade cycle for data centers as AI projects demand more and more processing capacity.

The share price hasn’t moved much in the past month or so, despite a solid earnings report in mid-February. Here’s the take from a free Fool.com article last week, for a little context on how challenging the valuation is now, with Arista trading at an $85 billion market cap these days:

“… things could finally be starting to cool down in 2024, mostly due to the big data center customers moderating their spend on infrastructure (all together, including Meta and Microsoft, “Cloud Titans” accounted for 43% of total sales last year). Management said other enterprise customers should still grow, but the end result is a forecast for just 10% to 12% revenue growth in the year ahead…..

“Growth is growth, so it’s no big deal. After all, Arista is holding onto all of its big gains from the last few years. But the problem with a slowdown is that Arista stock now trades for a whopping 40 times trailing-12-month EPS and 42 times trailing-12-month free cash flow.”

Arista is a very impressive company, and I should have held on to the shares I owned years ago — they are getting more involved in security, including through a partnership with Zscaler, so that helps to expand the business a bit beyond the core hardware and software they’ve offered for years… and they’re clearly very involved in helping to engineer switching hardware and data center equipment to better handle the vastly increased data demands of big AI models, including for their tentpole customers Meta and Microsoft, and they still seem to be eating Cisco’s lunch. They had wild acceleration in revenue and earnings in 2022 and 2023, which has helped erase the memories of several years of pretty flat numbers as everyone worried about the capital spending plans of their big customers and as investors were hit with some “surprise” bad years when capex budgets from Microsoft were cut and ANET had to announce big declines in orders from time to time.

Here are the charts they’re probably most proud of from their latest Investor Presentation, showing their consistent success in taking market share from Cisco (though the different scales mean that these would probably — fairly — be convicted of “chart crimes” in the court of Twitter):

And that stands out in the financial results, too — here’s the past five years of revenue and free cash flow growth (or non-growth) for Cisco (CSCO), which has provided a total shareholder return (blue line) of almost exactly 0% over five years:

And here’s that same chart for Arista — Cisco has acquired a bunch of firms over the years and become more diversified, so it’s not exactly an apples-to-apples comparison, but they are the two most important data center switching hardware companies, and the difference is stark:

Those aren’t the only two players, to be clear, and some of the other companies in this space have been somewhere between ANET and CSCO in recent years when it comes to growth and market share gains/losses, like (relatively tiny) Juniper Networks (JNPR), so there won’t necessarily be just one winner, and there’s no guarantee that Arista will keep taking share, which presumably becomes harder as they grow and as competition becomes more intense.

But yes, that’s why Cisco, the largest company in this space, trades for 13X forward earnings with a 3%+ dividend yield, and Arista, the second-largest (at least in the US), trades at 37X forward earnings. If you’re not growing, investors believe you’re dying, and that judgement call can shift pretty quickly.   So if you’re a more “value” minded investor, and want to be more cautious about betting on future growth, you might find Cisco substantially more appealing — it takes all kinds, and some value investing leaders, like the Oakmark funds, have been loading up on Cisco shares over the past year or so.

The risk is still in the dependency on some huge customers and the rich valuation, as everyone now assumes that the AI spending boom will continue for many years. If the spending on data center expansion and upgrades remains very high, particularly for the largest hyperscale cloud companies, then Arista is probably still a good buy even at these all-time highs… if the AI spending firehose is turned down a notch after a few more quarters, then Arista’s share price could easily drop by 50% if the Microsoft and Meta order flow pauses and their revenue growth stops.

On balance, I’d say the outlook should be reasonably solid for Arista, if only because they’re not just driven by AI spending, and they do have plans to keep expanding into adjacent product areas and taking share in non-cloud, non-hyperscale data centers and enterprise “campuses,” and in routers as well as their core switches, and building in more software sales on top of their hardware to generate growth — their own presentation says that they believe their addressable market will grow from $37 billion last year to $60 billion in 2027. And data centers are still in the heat of the 400G upgrade, and probably looking at 800G investments picking up substantially in the next year or two, so there should be a pretty continuous upgrade cycle and growth opportunity.

But yes, to probably be too repetitive here, at this point buying ANET would mean that you have to make a judgement call about future growth — which is essentially true for any of the popular AI or AI-adjacent stocks these days. They have plenty of surplus cash to survive whatever slowdown might come, and they’ve only been diluting shareholders by about 1% per year through stock-based compensation, with adjusted earnings now not too wildly far away from real GAAP earnings, and with real operating cash flow also pretty close to earnings (they’re valued at a little over 40X trailing free cash flow, and also 40X trailing adjusted earnings). They’re in fine shape. They’re just a bit expensive if they’re only going to grow their earnings and cash flow at less than 10% in 2024, even if that growth picks up to 10-15% in the next few years (that’s the current analyst forecast)… and they’re an easy buy if they’re going to continue growing earnings at anything like the 55-60% growth rate of the past two years. If you think the analysts who follow NVIDIA and Arista are going to be accurate — which, to be fair, is very unlikely, NVDA analysts in particular have never gotten the “turn” even close to right when growth has collapsed or accelerated in the past — then ANET is actually more expensive than NVIDIA, growing more slowly, and more reliant on its largest couple of customers… but I’d be marginally more likely to buy ANET at this valuation than NVDA, even though I own the latter, if only because they have some opportunity to build their business with Amazon and Google Cloud and to extend their leadership into other areas of networking equipment and software, and probably face less competitive pressure from next-generation or homegrown hardware than NVIDIA might in the next few years.

That’s just my take, though, and I’ve been wrong about Arista in recent years, and can’t predict the future very well. If you’ve got an opinion about Arista, or other investments that play off of the continuing investment in data center upgrades as the AI mania continues, please let us know with a comment below. Thanks for reading!

Disclosure: Of the companies mentioned above, I own shares of NVIDIA, Amazon and Google parent Alphabet. I will not trade in any covered stock for at least three days after publication, per Stock Gumshoe’s trading rules.