The AI Story Everyone Is Missing
Everyone is talking about AI. Almost no one is talking about what actually makes it run.
Here is the statistic that reframes the whole conversation: only 20–30% of the AI data centers planned through 2030 are built and functioning today. The other 70–80% are still on the drawing board.
That means the biggest build-out in modern American history — the cement, the cooling, the chips, the memory, the power — is mostly still ahead of us.
In this episode of Money On Tap, Ben Brayshaw and Dan Michelon go beyond the household names and dig into the hidden companies powering the AI revolution — the second wave of winners most investors still haven't heard of.
From the Mag Seven to the Second Wave
For the first couple of years, the AI trade was simple: buy the Magnificent Seven. The big tech giants were making the AI investments, and their stocks ran on the promise of what AI could become.
But something changed.
The names benefiting most from AI have shifted — from the companies spending the money to the companies receiving it.
Google, Microsoft, Amazon, Meta, and Apple are pouring cash into data centers — draining reserves, taking loans, doing whatever it takes to build capacity. They aren't the ones booking the profits yet.
The companies on the other side of those checks are. Names like Micron, SanDisk, Vertiv, Marvell, and Broadcom — barely on the radar in the first phase of the AI boom — have been the clear leaders of the second wave.
Phase one was speculation and hope. Phase two is follow the money.
Picks and Shovels, All Over Again
During the California gold rush, the people who reliably made money weren't the miners. They were the ones selling the picks, the shovels — and the jeans.
The same dynamic is playing out in AI. The "gold" is the finished AI product Big Tech hopes to sell us someday. The picks and shovels are everything required to build it:
- Chips — the trade that already happened, with demand so deep that top chipmakers have been effectively sold out years in advance.
- Memory — Micron's story. AI moves staggering amounts of data at staggering speeds, and all of it needs rapid memory. Micron is the number-one name in the space with remarkably little competition at scale.
- Storage — SanDisk's lane. All that data has to live somewhere.
- Cooling — Vertiv was essentially the only serious player in liquid cooling — the method data centers strongly prefer — for years, and still leads the space.
- Networking — Marvell and Broadcom make the components around the chips that route data where it needs to go.
Nobody grew up knowing who makes a data center's liquid cooling loop. That is exactly where the opportunity has been.
The King of AI: Taiwan Semiconductor
If there is one company nearly every AI player depends on, it is Taiwan Semiconductor Manufacturing (TSMC).
Nvidia, AMD, Apple, Broadcom, Qualcomm — they all need TSMC's fabrication expertise. If AI grows, TSMC benefits, almost regardless of which chip designer wins.
It is also why a small island a few miles off mainland China has become a genuine national security concern — the world's most critical manufacturing capability sits a stone's throw from a geopolitical rival.
For investors looking for general AI exposure rather than a single-company bet, TSMC keeps stepping to the plate across the board.
The Rotation Nobody Called a Rotation
Three months ago, markets were down and headlines asked whether a recession was coming. In hindsight, much of that turbulence wasn't an exit from tech at all.
It was a rotation — out of the Mag Seven and into the infrastructure layer.
The Mag Seven's expectations had been met, earnings were nearly perfect, and taking profits was rational. That capital didn't leave the AI theme; it moved to where profits are being earned today — memory, storage, cooling, networking.
And here's the full-circle possibility: once the hyperscalers finish building and start selling finished AI products, the money may rotate right back. For investors with a real time horizon, beaten-down giants like Microsoft may prove to be the opportunity hiding in plain sight.
The Layers Still to Come
The server rack is only the beginning. A data center is a building — with everything a building needs, at industrial scale:
- Electrical infrastructure — regulated voltage, surge protection, switches, breakers, backup generation. Companies like Eaton, Hubbell, and Powell Industries live in this space.
- Utilities and grid modernization — many approved data centers must create their own power, fueling interest in nuclear and small modular reactors. Constellation, Vistra, and NextEra are names in the conversation.
- Engineering and design — firms like Jacobs are designing these facilities for maximum electrical efficiency.
- Construction materials and machinery — Vulcan Materials, Martin Marietta, Nucor, Steel Dynamics on the materials side, and Caterpillar on the machines. None of this exists until someone puts a shovel in the dirt.
- Data center real estate — REITs like Digital Realty and Equinix lease these facilities, offering exposure with a core-income flavor.
Only 20–30% of the build-out is done. Every one of these layers has years of demand ahead of it.
The Risks Worth Respecting
None of this is a sure thing:
- Valuation risk. Speculation drove some AI names up three to five years ahead of their fundamentals.
- Capex pullback. If the hyperscalers turn the gas down on data center spending, orders drop fast for everyone downstream.
- Competition. Semi-monopolies attract challengers. Today's exclusivity will not last forever.
- Interest rates. The build-out runs on borrowed money. Higher rates mean fewer data centers get built, sooner.
- Tariffs. The trade story is far from settled, and infrastructure components could face additional price pressure.
One discipline matters more than any single pick: high conviction. Know why you own what you own. If you made a mistake, move on. If the fundamentals are intact, don't let a rough quarter shake you out.
Final Thoughts: The Build-Out Is the Story
AI is still in its infancy — it hallucinates, it breaks, it surprises. But the infrastructure being poured into the ground right now is very real, and it touches everything from memory chips to cement.
The first wave rewarded the visionaries. The second wave is rewarding the builders. And there are layers beyond that — power, engineering, materials, real estate — where the demand hasn't even fully arrived.
You don't have to guess where AI ends to recognize where it has to begin.
For investors who would rather not bet on a single company, ETFs and REITs in the space can package the theme with built-in diversification — worth a look with proper due diligence.
Next Steps
Want to know how the AI build-out fits into your portfolio — or whether your current holdings already give you the exposure you think they do? We also have a white paper on this topic with much more detail.
Call: 855-226-8551
Email: info@yourmoneyontap.com
Frequently Asked Question
What are the "picks and shovels" of the AI revolution?
They are the infrastructure companies that profit from the AI build-out regardless of which AI product wins — memory (Micron), storage (SanDisk), liquid cooling (Vertiv), networking (Marvell, Broadcom), and chip fabrication (Taiwan Semiconductor), plus electrical, power, construction, and data center REIT plays. With only 20–30% of planned AI data centers built, most of that demand is still ahead.
The views expressed are educational in nature and should not be construed as a recommendation to buy or sell any security mentioned. Investing involves risk, including loss of principal.
Money on Tap is your trusted resource for investing, retirement planning, and building long-term financial confidence.