April surprised a lot of people. The S&P 500 climbed roughly 10 percent while half of financial Twitter was convinced we were heading into a correction. We weren’t. But now that the relief rally has run its course, the market is getting more selective about what it actually rewards.
Here’s what I’m seeing: we’re past the phase where slapping “AI” on a press release was enough to move a stock. What’s working now is owning the companies that everyone building AI has to pay. The picks-and-shovels trade. While the hyperscalers argue over model benchmarks, these are the businesses quietly collecting checks from all of them.
One of the reasons I’ve been so focused on this infrastructure angle lately is that it keeps showing up across the advisory services I follow closely. Dr. Mark Skousen built his entire Skousen Report thesis around it, including an early call on the SpaceX IPO direction before the SEC filing was even confirmed. That kind of macro-to-stock bridge is exactly the lens I’m using to pick the names below.
For income investors who want dividend exposure running alongside these growth plays, Marc Lichtenfeld’s Oxford Income Letter is worth a look as a counterbalance. And if you want a quantitative screen on top of a fundamental thesis, the Power Gauge Report has been one of the better momentum tools I’ve seen for catching these moves early.
Here are the seven public plays I’m watching closely heading into May.
1. Arista Networks (ANET): The Backbone of the Data Center
Nobody talks about ethernet switches at dinner parties. But those switches are exactly why NVIDIA’s GPUs can actually communicate inside a data center. Without Arista’s networking gear, you don’t have an AI cluster — you just have a very expensive room full of hot silicon.
Their Q1 numbers were hard to argue with: $2.71 billion in revenue, up 35 percent year over year. CEO Jayshree Ullal said demand is the strongest she’s seen in her two decades running the company.
Wall Street vs. Reality: Analysts at Morgan Stanley got grumpy because the full-year guidance raise wasn’t aggressive enough for their models, knocking the stock down 9 percent pre-market. Meanwhile, billings growth came in at 54 percent. That gap between what’s shipping and what’s already been sold tells you the backlog is real. This is one of those pullbacks you look back on six months later and wish you had leaned into harder.
My Position: I’ve held ANET since mid-2025 and used the post-earnings dip to add. When a company is generating $1.69 billion in free cash flow and the stock sells off because guidance wasn’t optimistic enough, that’s usually a gift.
2. NVIDIA (NVDA): Still the Toll Booth on the AI Highway
Everyone already knows about NVIDIA. But I keep coming back to it because the story keeps getting bigger, not smaller. They just reported $44 billion in quarterly revenue, up 69 percent from a year ago. Their data center segment alone did $39 billion. That’s not a hot streak — that’s a structural shift in where compute dollars are flowing for the next decade.
The bearish argument is always valuation. And sure, at roughly 25x forward earnings it’s not cheap. But when you’re the only company whose chips can actually train frontier AI models at scale, “expensive” is relative.
Wall Street vs. Reality: The main concern is export restrictions limiting China sales. Fair point. But their non-China pipeline — hyperscalers, sovereign AI buildouts, enterprise inference — is so strong that the China headwind looks manageable. Blackwell demand is still outrunning supply by a wide margin.
My Position: NVDA is my largest single holding. I trimmed a little in Q1 to rebalance, but I’m not going anywhere. This is the one stock I’d be most uncomfortable not owning in this environment. Louis Navellier has been pounding the table on AI infrastructure names like this for a while now — his Growth Investor service flagged the Blackwell cycle early, which lines up with everything I’m seeing in the order data.
3. GE Vernova (GEV): Powering Everything That Needs Power
Here’s the part most people miss about the AI boom: it doesn’t just need chips. It needs electricity. A lot of it. The average AI data center uses roughly 20 to 50 times more power than a traditional data center, and the U.S. grid was simply not built for this.
GE Vernova makes the turbines, grid equipment, and power technology that utilities need to keep up. They spun out of GE in early 2024 and have been on a tear since. Revenue is growing, the order backlog is up over 30 percent, and they sit at the exact intersection of AI infrastructure and the energy transition.
Wall Street vs. Reality: Some analysts are cautious because GEV is still in early innings as an independent company and margins are lumpy quarter to quarter. But the orders are there and execution is improving every quarter. The macro case for grid infrastructure spending is about as solid as it gets right now.
My Take: I don’t have a full position yet — I’ve been watching since the spin-off and I’m getting close. The risk/reward at current levels looks interesting for anyone with a 2 to 3 year time horizon.
4. Vistra Energy (VST): Nuclear Isn’t Boring Anymore
If someone told you two years ago that a power company would be one of the best performing stocks in the S&P 500, you’d have laughed. And yet here we are. Vistra operates nuclear, natural gas, and solar assets across the country, and right now nuclear is the golden ticket because it’s the only clean, always-on power source that can actually serve a hyperscale data center under a long-term contract.
They’ve inked power purchase agreements directly with major tech companies and are expanding nuclear capacity. Q1 free cash flow was strong and management raised full-year guidance.
Wall Street vs. Reality: The knock on VST is that it’s “just a utility.” But utilities that sell power directly to the Microsofts and Amazons of the world at negotiated rates under 10-year contracts aren’t your grandpa’s utility stocks. This is infrastructure with a real moat.
My Position: I’ve had a small position in VST since late 2024. Not flashy. Just consistently right. If you’re looking at the macro case for energy from a different angle, Jim Rickards has written a lot about grid security and energy as a strategic asset in Strategic Intelligence — worth reading alongside any energy position thesis.
5. Albemarle (ALB): The Lithium King is Back
Lithium had a rough couple of years. Prices collapsed, sentiment turned, and a lot of investors gave up on the space entirely. I didn’t, and Q1 just validated that patience.
Albemarle reported EPS of $2.95 against analyst expectations of $1.31. That’s a 125 percent beat. The stock jumped over 11 percent in a single session. When a company misses consensus by that much in the right direction, the market is telling you the consensus was simply wrong.
Wall Street vs. Reality: The bears point to EV market softness and commodity price swings, and they’re not wrong that EV demand has been choppier than expected. But the story has evolved: AI data centers need massive battery energy storage systems just as much as EVs do. Battery storage deployment grew 66 percent recently, and ALB is the lowest-cost lithium producer on the planet. They just paid down $1.3 billion in debt and the balance sheet is in much better shape than most people realize.
My Position: This is still my anchor commodity play. I held through the 2024 lows because the balance sheet never actually broke, and watching them stay free cash flow positive through the trough told me the thesis was intact. For anyone who wants a forensic accounting read on commodity balance sheets, Joel Litman’s Hidden Alpha service does exactly this kind of deep balance sheet work — it’s how I think about the difference between a broken company and a bottoming one.
6. Palantir Technologies (PLTR): Where AI Actually Gets Used
There’s a difference between companies that sell AI products and companies actually deploying AI inside real operations at scale. Palantir is firmly in the second group. Their AIP platform is running inside supply chains, military logistics, and hospital systems — not sitting in a demo environment waiting for a proof of concept to convert.
The numbers back it up: revenue up 85 percent, GAAP net income of $871 million in a single quarter. Their “Rule of 40” score sits at 145 percent, which is nearly unheard of in enterprise software. For context, most software companies celebrate hitting 40.
Wall Street vs. Reality: The consistent complaint is valuation — “too expensive at 20x revenue.” But U.S. commercial revenue is growing at 133 percent and their net dollar retention sits at 150 percent. When customers get on this platform, they don’t leave. They expand. That’s the definition of a sticky business, and the market is starting to price it accordingly.
My Position: I’m a long-term bull here and I don’t trade the quarterly swings. The U.S. commercial momentum is just starting to hit its stride, and it’s one of the few software plays I’m comfortable holding alongside hard infrastructure names like ANET and GEV. James Altucher flagged Palantir early in his Investment Network as part of his AI portfolio — and while his style is different from mine, the underlying thesis was sound.
7. Constellation Energy (CEG): The Nuclear Landlord
If Vistra is the scrappy nuclear play, Constellation is the institutional-grade version. They operate the largest fleet of nuclear plants in the U.S. and have been signing long-term power agreements with some of the biggest names in tech. The Microsoft deal alone — a 20-year agreement to restart Three Mile Island — was a clear signal that serious capital is committed to nuclear as baseload power for AI.
Q1 earnings were solid across the board, and the regulatory tailwinds for nuclear are the strongest they’ve been in 30 years. Both sides of the aisle in Washington are aligned on keeping the grid reliable, which is genuinely rare.
Wall Street vs. Reality: Some analysts worry about regulatory complexity and the cost of operating aging nuclear plants. Those concerns aren’t unfounded. But when Microsoft is signing 20-year contracts with you, the risk/reward calculus shifts considerably in your favor.
My Take: I’ve been building a position in CEG slowly over the past several months. Between this and VST, I think you want some nuclear exposure in any serious infrastructure-themed portfolio right now. The thesis isn’t complicated: AI needs power, power needs a reliable baseload, and nuclear is the only option that checks all three boxes cleanly.
The Bottom Line for May:
The common thread across all seven of these picks is simple: they are the infrastructure of the next decade. Chips, networking, power, batteries, software that actually runs things. You don’t have to bet on which AI model wins. You just have to own the stuff that all of them depend on.
If you want to go deeper on the research side, I’ve found the Stansberry Investment Advisory useful for thinking through position sizing and risk management on names like these — their portfolio has 34 open positions with 10 above 100 percent gains, and the stop-loss discipline is something I’ve borrowed more than once. The Project Apex write-up is also worth reading if you want to understand how the Navellier team is thinking about AI infrastructure as a broader investment theme.
I’ll be back with a June update to see how these played out.
Affiliate Disclosure: This article contains affiliate links. If you click through and make a purchase, I may receive a commission at no additional cost to you. All stock picks and opinions are my own based on personal research and experience. Nothing here should be taken as financial advice. Always do your own due diligence before investing.