3 AI Stocks to Buy (and 2 to Skip) as the AI Buildout Hits a New Gear
Listen to the audio version of this article (generated by AI).
Think back to the summer of 2023, back when nuclear and uranium stocks were dead money. Back then, everyone kept asking if Big Tech would actually choose nuclear to power its data centers.
Then the deals started landing.
Microsoft (MSFT) and Constellation (CEG). Google (GOOGL) and Kairos. And those uranium and nuclear stocks took off like rocket ships — Cameco (CCJ), Oklo (OKLO), the whole complex tripled and quadrupled.
I bring this up because I think geothermal just had its “summer 2023” moment. And that is only one of five stories from this week’s Being Exponential with Luke Lango podcast — a week that perfectly captures the strangest setup I have seen in years.
We just lived through arguably the best earnings season of all time. Blended second-quarter S&P 500 earnings growth is running at 52%, or the highest since Q2 2021, when it hit 91.6%. Dell (DELL) grew revenue 58%. Nvidia (NVDA) guided to 70% growth for fiscal 2028. And yet stocks have done basically nothing since May. The Nasdaq-100, the semis, the whole AI complex round-tripped.
Good earnings are not enough right now, because this is a liquidity problem, not a fundamentals problem. And that divergence (fundamentals ripping while prices stall) is actually a gift.
So let me walk you through the five stocks I broke down this week, from the cheap winner you buy today to the dark horse I would avoid entirely — and the one macro signal that tells you when to back up the truck:
The Cheapest Way to Own the AI Buildout
Dell is firing on all cylinders right now, with revenue up 58% year-over-year to a record $47 billion.
The AI server backlog grew to a record $60.9 billion in orders booked in a single quarter, and management hiked its AI server revenue guide expecting roughly 3X year-over-year growth. Dell boosted its full-year revenue guide by $2 billion to $192 billion — implying 68% revenue growth this year. Two years ago, this was a company growing 8%.
Yes, growth is expected to cool after this to roughly 19%, then 14%, then 13%. But gross margins are holding steady in the high teens to 20%, and economies of scale could produce actual margin expansion.
DELL stock trades at just 16.6 times forward earnings for a high-teens-to-low-20s EPS grower, and it is bouncing right off its 50-day moving average. That is dirt cheap for this sort of growth profile. Dell is one of the most attractively valued AI stocks in the entire market. I recommend it.
A Broken Chart on a Great Business
Credo Technology Group (CRDO) reported fantastic numbers. Revenues rose 115% year-over-year and 10% sequentially to a record $479 million. Adjusted net income jumped 140%. Operating margins hit 48.2%. Management boosted its Q2 revenue guide to $525 million to $535 million and maintained its forecast for 85%-plus full-year revenue growth with roughly 50% adjusted net margins. AECs remain the largest business, Credo has deep relationships with five of the six big hyperscalers, and its neocloud exposure is growing.
And the stock fell nearly 20%.
Why?
The growth mix is changing. AEC growth is slowing, and optics is becoming the primary growth engine — but optics is lower margin, second-half-weighted, and carries execution risk. Gross margins are expected to stay flat. And there is real customer concentration: as CFO Dan Fleming spelled out on the call, four customers represent 84% of revenue, at 33%, 28%, 13%, and 10%.
The chart is broken — CRDO nosed beneath its 200-day moving average. But historically this stock bottoms just below the 200-day, maybe around the $150 level. At 22.6 times forward earnings and about 19.5 times forward EBITDA — both roughly two-year-low multiples — for a company growing the top line 87%, then 55%, then 30% into 2029, the story is fully intact.
Wait for signs of bottoming, then buy the dip. Because the rebound could be big.
Geothermal’s Nuclear Moment
Fervo Energy (FRVO) just landed a 396-megawatt power purchase agreement with Google for its Cape Station project — the world’s largest enhanced geothermal PPA to date, with an option to expand offtake by roughly 600 megawatts to nearly 1 gigawatt by June 2030.
That is real, validated revenue visibility from a top-tier counterparty.
As Google’s head of advanced energy, Michael Terrell, put it, “The next chapter of advanced power generation technology is being written in Utah.” The stock had declined from above $40 to around $15 before bouncing to roughly $20.
This is the summer-2023 parallel I opened with. Everyone wondered if Big Tech would choose geothermal. Google just answered. The Google PPA could be geothermal’s version of the moment that sent uranium and nuclear stocks up like rocket ships.
It is risky, but after this deal, the bull thesis outweighs the bear thesis.
The Poor Man’s Bloom Energy
FuelCell Energy (FCEL) signed its first-ever data center power agreement — 75 megawatts in Texas — and added $2.4 billion in awarded capacity backlog through a Fit Energy expansion option.
The balance sheet is clean, so bankruptcy risk is not real. But the quarterly numbers are ugly: revenue dropped about 30% year-over-year. A huge ramp is expected — 66%, 48%, then 110% growth — but there is no growth right now.
This is a show-me story competing against a very capable Bloom Energy.
Between the two, I would go Bloom Energy all the way.
Great Growth, Scary Balance Sheet
Eos Energy Enterprises (EOSE) is a battery storage play that also landed a big Google deal.
Unlike FuelCell, Eos is growing dramatically — revenues up roughly 350% last quarter, around 170% this year, and 93% next year. But the balance sheet is troublesome: only $305 million in cash against $420 million of cash burn over the trailing 12 months, with roughly $640 million in debt.
More dilution, more raises, and more hiccups are coming. The stock sits 50% to 60% below a declining 200-day moving average.
I’ll probably become a buyer eventually… just not here.
The Macro Signal That Changes Everything
Here is why the tape is stuck: Excess liquidity has dried up.
A hawkish Fed regime shift plus higher Treasury yields — the 10-year and 30-year breaking out — means slower money supply growth, and hot inflation from oil creates a double liquidity squeeze. That is why even Nvidia’s blockbuster numbers could not restart the rally.
But the fundamentals keep strengthening. The August ISM Manufacturing PMI came in at 54.6, down from July’s near four-year high of 55.6 — a “Goldilocks bad” economy, weak enough to cap long-term yields but not weak enough to derail growth.
And underneath it all is what I call the AI Bifurcation: AI-linked manufacturing and construction are in a supply-constrained boom while the broad economy slows.
PwC’s Global Data Centre Outlook projects a central case of $31.6 trillion in cumulative AI data-center capex through 2050 — with a plausible upside near $50 trillion — a 25-year building supercycle, orders of magnitude larger than the railway and internet buildouts, with annual spend growing from roughly $800 billion in 2026 to $1.8 trillion by 2050.
As PwC put it, “Every prior infrastructure wave — railways, electrification, the internet — front-loaded construction capex and tailed off as the network matured. This wave inverts the pattern.” Broadcom and Ciena just confirmed it: demand is running materially ahead of supply, with visibility through at least 2028.
EPS trendlines are diverging from price trendlines across AI names — fundamentals up, sentiment down. That is a buying opportunity. In this environment, you do not bet on dark horses, moonshots, or lottery tickets. You hunker down in your favorites — in AI energy, that is Bloom Energy, the top dog — and wait for excess liquidity to return.
I expect these dynamics to reverse by year-end.
The Bottom Line
I break down all five of these names, including the valuations, the charts, and exactly which ones to buy and which to avoid, in this week’s “Being Exponential” episode.
Go watch it now!
Every name I just walked you through — Dell, Credo, Fervo Energy, FuelCell, Eos — exists on the same shopping list. There are trillions of dollars flowing into building, cooling, and powering the AI buildout, and these are the best AI stocks to buy in that regard.
And that shopping list keeps growing.
Chips needed servers. Servers needed cooling. Cooling needed power. Power is now the bottleneck, which is exactly why Google is cutting nine-figure checks to geothermal startups nobody had heard of two years ago.
I have spent months tracking that same dynamic inside Elon Musk’s empire specifically — Tesla (TSLA), SpaceX (SPCX), and xAI.
Musk hates bottlenecks, so he tries to build almost everything himself. But he simply can’t build everything.
And the companies that remain indispensable to him, the ones he still has to write checks to, are exactly the kind of overlooked suppliers I look for.
It is the same logic that led me to Fervo this week.
On Wednesday, September 9, at 8 p.m. Eastern, I am hosting a free workshop with my colleagues Louis Navellier and Eric Fry to show you what I believe Musk is building next, where his empire’s biggest remaining bottlenecks are forming, and which suppliers I think could get paid to solve them. I will even give you the name and ticker of one stock for free, no purchase necessary.
All you need to do is reserve your seat here.
