rewrite this content using a minimum of 1000 words and keep HTML tags
The massive amount of money pouring into data center buildouts is trickling down everywhere you look. Plenty of story stocks promise growth tomorrow, but the investments are happening today. Everyone is trying to find “the next bottleneck” and it’s led to an absolute mess where people are stretching the boundaries of what constitutes an “AI stock.” If everything is an AI stock, nothing is an AI stock. It also creates lots of noise for retail investors.
The Paradox of Choice
The only way to navigate the AI stock frenzy is to establish rules in advance of investing in any company. We’re in the middle of the largest corporate investment effort mankind has ever seen, and that money is going straight into data centers. That means that any data center stocks you are invested in should be showing strong revenue growth today. That is a must. No MOUs, no “strategic partnerships,” no NVIDIA joint press releases, just revenues that are growing like mad. That’s rule one.
Rule two is that we only invest in leaders. That’s because as the AI trade matures, leaders will be more likely to capture market share from runners-up. More importantly, they’ll be better able to survive any rapid downturn that may accompany a shift in AI funding tides. The larger you are, the easier it is to raise capital. It also results in the “nobody ever got fired for buying IBM” effect.
Rule three is that we don’t chase hype. When NVIDIA’s Jensen can move shares of a trillion-dollar company with a few words, that’s called irrational exuberance. Competent investors have rules in place, guardrails that protect us from our own greed and fear. Before buying or adding shares of any stock, we make sure it’s not ridiculously overvalued.
With these rules in mind, let’s look at four datacenter stocks listed in our tech stock catalog as a “like.”
Four Data Center Stocks
ETFs are tough to replicate for retail investors which is why we pay fees or expense ratios. However, we can also create “mini ETFs” that provide the same sort of diversified exposure without the fees. In fact, we can select only leaders and they should offer even better concentrated exposure than an ETF might. We’ve done this with semiconductors, electricity infrastructure, and now data centers. In our disruptive tech stock catalog are four data center stocks which are leaders in their respective spaces, each with a simple valuation ratio (SVR) below our cutoff (presently around 24):
This “mini ETF” has a blended return of +150% against a Nasdaq return of +54% over the earliest time frame listed above – May of 2024. However, one of these names is not like the others. Everpure has meaningfully underperformed over that time frame and sports a “cheap” valuation relative to its peers.
A Big Data Company Making Hay While the AI Sun Shines
We’ve been “liking” Pure Storage – err, Everpure – since well before the AI mania. While revenue growth stalled as the company pivoted into “data storage as a service,” it’s since rebounded, and we’re now seeing solid acceleration. Last time we looked they were guiding to 14% growth in Fiscal 2026. Actual growth came in at 16% and now they’re expecting 22% for the coming year.
Our last piece noted a simple valuation ratio (SVR) that had doubled to around seven. It’s now around five which means this big data growth stock can be bought well below the average of our tech stock catalog (about eight). Does this represent value or a value trap? And what’s up with the 30% price drop?
Everpure is a volatile stock so we’re always best served focusing on valuation, not share price. When a company has strong growth and a low SVR it usually comes down to perceptions of profitability. How can Everpure have consistent 68-70% gross margins yet they’re only able to barely realize a positive operating margin?
Everpure’s non-GAAP operating margins remove share-based compensation (SBC) and move towards the 15–21% range with an upward trajectory, which means SBC is putting a damper on the company’s bottom line. This also dilutes current shareholders who slowly own a smaller piece of the business over time.
Then there’s the memory shortage. Everpure sells a product called a “flash array” which is a cluster of a bunch of solid state drives – as opposed to spinning disc drives – used for data storage. The appeal here is that solid-state hardware is faster, lasts longer, and is more energy efficient than spinning disc alternatives. The problem is that these flash storage components have seen an astronomical increase in prices thanks to a massive imbalance between supply and demand. While Everpure was able to eat the costs temporarily, they finally raised prices as much as 70% in 2026. This is a double-edged sword as it improves their margins but stifles demand from price-sensitive customers. This also explains the lack of increasing profitability with increasing revenues. In other words, Everpure’s growth is being eaten up by rising input costs.
We still “like” Everpure’s reasonable valuation and strong growth, and we see the memory shortage as a temporary headwind for the company. The fact that the business is diversified between hardware and software acts as a natural hedge during these turbulent times.
As for leadership, Everpure is not a dominant force in enterprise storage systems, landing in fourth place amongst its peers with a 6.8% market share in what is clearly quite a fragmented market.
Everpure had Q3-2025 revenues of $964.5 million. Of that, subscription revenues were $429 million. Subtract the two and you get the above number, so this all makes sense. Our last piece on Everpure talked about the competition not offering viable ways to play this theme. The competitors above don’t provide pure-play exposure to the growth of big data since they also serve the consumer electronics market. Investors will need to believe Everpure can propel their data subscription business into a leadership position which should eventually help resolve those margin concerns. Now it’s easier to understand why the stock appears relatively undervalued.
ANET Stock – A Pure Play on Data Center Growth
The next three stocks in our mini ETF are all involved in “networking,” or helping computers talk to each other. That might include transmitting information across a data center or to the outside world via the internet. Arista Networks is a leading provider of switches, devices that help facilitate the connection of multiple computers. They’re a data center pure-play with over 90% of revenues coming from this segment.
Our conclusion two years ago was simple: As long as Arista continues to deliver strong growth with a credible plan then stay on for the ride. That is, if you’re willing to accept the customer concentration risk Arista presents. Two customers accounted for 42% of the company’s revenue in 2025, up from 35% in 2024. While not disclosed, it’s speculated that these customers are Microsoft $MSFT and Meta $META, both of which have deep pockets and plans to continue investing heavily in data center projects. Growth has continued unabated, so there’s not much else to do but wait for it to subside.
If you’re sitting on the sidelines eyeballing the rich valuation – an SVR of about 18 – you have two options. You can accept the high valuation due to the fact that this is a highly profitable company with roughly 40% net profit margins, or you can set a valuation target that you won’t exceed. Then when volatile shares of this company inevitably hit your target, you can add slowly and methodically without FOMO. Just beware of the big green gorilla hot on your tail.
A recent report by IDC shows that the Ethernet switch market was worth $55 billion in 2025. About 18% of that or $10 billion came from datacenter networking hardware. Of that, 19% belongs to Arista while 15% belongs to NVIDIA. Sure, there’s room for both to grow right now, but who is in a better position once the massive data center investment push starts to lose momentum?
One company that seems well-suited to survive the demand destruction would be Broadcom with its unique corporate structure.
Is Broadcom Just Getting Started?
That’s the question we asked several years ago and it seems that they were. Growth has continued unabated, accelerating in recent years for this company that does an exceptional job of anticipating what’s going to be hot technology, acquiring it, and integrating it. The result is exceptionally strong growth being driven specifically by two AI use cases – networking and custom silicon.
Fortunately Broadcom breaks out their AI revenues for investors into a bucket called “AI Semiconductor Revenue” which has moved from 28% of total revenues in Q1-2025 to nearly half of total revenues today. Next quarter alone it’s expected to increase 45% sequentially. For 2027, they’re expecting AI revenues to eclipse $100 billion. Below you can see the growth of AI semiconductor revenues over time as a proportion of total semiconductor revenues and total company revenues.
Last quarter, about 40% of AI revenues came from networking, though Broadcom is more of a peer to Arista than a competitor. While their primary product is semiconductor chips used within routers and switches, they also make the physical switches themselves along with their own line of optical networking products. The other 60% of their “AI revenues” came from “custom silicon” which is intuitively the way forward. Some believe that eventually companies will all have chips tailored to their unique business models and we’ll move away from “one size fits all” compute. If that’s the direction we’re heading, Broadcom should absolutely dominate given they own about 60% market share in custom silicon (Marvell $MRVL is in second place at 8%).
Broadcom sports gross margins around 80% and operating margins around 40% so it’s no surprise they have the steepest valuation of all four names – a simple valuation of 20 that falls just under our current threshold. This means any sign of weakness can send shares spiraling, which is exactly what happened after their recent earnings report. The company opted to retain their 2026 revenue guidance despite demand for their networking hardware continuing to rise. The market punished the stock, but a current simple valuation ratio of 20 hardly feels like a bargain opportunity.
Broadcom seems like the most attractive of the four companies mentioned because their dual-pronged approach to AI – coupled with a diversified collection of other businesses – means they could better weather any downturn in AI demand. They aren’t just re-allocating existing revenues into a “AI revenues” bucket, because we’re also seeing exceptionally strong top-line growth (a concern we had before).
Our next company isn’t acquiring, but shedding businesses, as they look to isolate focus on the high-growth area of optical networking.
Coherent Goes All-In on Data Centers
A year ago we pointed to Coherent as a data center beneficiary that was realizing strong growth at a reasonable valuation. Since then the company looks even better because they recently began divesting their slower-growth businesses and focusing on their major breadwinner: data center communications. That’s why revenues dipped in 2024 but have since resumed their growth.
The company’s key product today is called a “transceiver” and it acts as a way to send and receive massive amounts of data via light signals as opposed to traditional electronic currents.
Our previous analyses of the company pointed to one major issue: weak margins. Just two years ago, Coherent sported gross margins of 30% and negative operating margins. That’s not a good look for a nearly 60-year-old company. However, these have improved steadily and are likely a result of the ongoing data center boom.
Hyperscalers are all racing to build the most infrastructure which means they don’t mind paying premium prices for vital hardware. Our question is this: Can this margin growth continue? What happens when all this data center spending slows down? There’s also competition to consider. Lumentum $LITE is rapidly gaining market share, and Chinese rivals like Innolight are offering significantly cheaper alternatives.
In regards to leadership, our recent piece on photonics stocks posited that Coherent was the leader based on total revenues, though they’re also growing that larger base at a slower rate. They’re the bigger, more complete photonics company when compared to peers like Lumentum which is showing more rapid growth. Being the larger volume business should help them weather any sort of demand destruction that might happen when data center building activities level off.
When Birds of a Feather Move Together
Investing across this broad set of names may shelter you from company-specific risk, but it’s not likely to provide as much diversification as you think. A recent piece by Bloomberg talked about how the momentum trade has never worked out well for investors. Basically just latch on to any stock that’s going up and let the alpha pour in. Broadcom is the third highest weighted momentum stock in the iShares MSCI USA Momentum Factor ETF $MTUM while Arista and Coherent occupy positions 39 and 37 respectively. When the momentum trade inevitably loses its momentum (tee hee) then these stocks will behave accordingly. The slightest pullback in AI spending will impact these names in varying degrees depending on how reliant they’re becoming on the AI trade.
When aggregating your portfolio, figure out how much exposure you have to the AI trade, and consider these names to be highly correlated as well. Every hardware company out there is trying to convince you they’re the next bottleneck. For those that are a bottleneck, you’re told the demand isn’t temporarily high. This is “the new normal,” they’ll tell you. Just remember the four most dangerous words in investing: this time it’s different.
Conclusion
Every financial pundit out there can pull a handful of data center stocks out of their bum on demand. How they arrive at such a list should not be based on ‘muh feelies’, or what Cramer does or doesn’t say. Having objective rules makes selection a whole lot easier and allows investors to have exposure without all the noise and pesky expense ratios. With NVIDIA being our largest position, we’re receiving a decent amount of data center exposure. Still, should we decide to add more exposure with one or more of these names, Nanalyze Premium subscribers will be the first to know.
and include conclusion section that’s entertaining to read. do not include the title. Add a hyperlink to this website http://defi-daily.com and label it “DeFi Daily News” for more trending news articles like this
Source link
