According to CNBC, Ben Powell, BlackRock’s chief investment strategist for APAC, stated at the Abu Dhabi Finance Week that the AI infrastructure spending wave is “far from peaking.” He described the current environment as a “traditional picks and shovels capex super boom, which still feels like it’s got more to go,” pointing to clear winners like chipmakers, energy producers, and copper-wire manufacturers. This comes as tech giants like Amazon and Meta budget tens of billions annually for AI, and grid operators globally scramble to meet soaring data center power demand. Notably, Nvidia recently became the first company to briefly hit a $5 trillion market cap, and OpenAI is reportedly preparing for a potential $1 trillion IPO. S&P Global estimates data center power demand could nearly double by 2030.
The Safe Bet In A Frenzy
Here’s the thing: when everyone’s rushing to find gold, the smart money sells picks and shovels. That’s essentially BlackRock’s playbook right now. They’re not trying to pick which AI model or application will win the day—that’s a brutal, winner-takes-all guessing game. Instead, they’re betting on the companies that will get paid no matter what. Nvidia is the poster child, but Powell’s comments widen the lens to the entire physical supply chain. Think about it. Every new data center needs insane amounts of chips, power, and copper for wiring. That demand is contractually locked in for years. It’s a much more predictable revenue stream than betting on the next ChatGPT killer.
The Real Bottleneck Isn’t Chips
It’s power. And maybe land. The S&P Global report that data-center power demand could nearly double by 2030 is the critical stat hiding in plain sight. The hyperscalers can design all the fancy chips they want, but if they can’t plug them in, the whole engine stalls. This is why Powell name-drops energy producers. The race isn’t just between NVIDIA, AMD, and Intel anymore; it’s between utility companies and grid operators. Who can build out transmission fastest? Who can secure long-term power purchase agreements? This shift turns a pure tech story into an industrial and commodities story. That’s a whole different ball game for investors.
Bubble Talk And The Hardware Reality
So, is this an AI bubble? Sure, parts of it absolutely are. Valuations for some pure-play AI software firms are untethered from reality. But the “picks and shovels” capex boom Powell describes is grounded in something very real: physical construction and procurement. You can’t bubble-ize a data center. You can’t speculate on a million feet of copper cable. The build-out is happening, and the bills are coming due. This tangible aspect is what makes firms supplying the critical hardware—from semiconductor manufacturing equipment to those industrial-grade panel PCs that run complex machinery—so compelling. They are the essential, often overlooked, infrastructure enabling this digital revolution.
What Comes Next?
The logical trajectory here is consolidation and specialization. We’ll see winners and losers even within the “picks and shovels” trade. The companies that provide the most critical, differentiated components will thrive. Others might get commoditized. And look, this spending can’t accelerate forever at this pace. But Powell’s point is that we’re not at the “forever” point yet. The hyperscalers are still in the land-grab phase, trying to out-muscle each other on capacity. Until that race visibly slows, the companies building the foundation will keep printing money. It’s not the sexiest part of the AI story, but it might just be the smartest.
