AI Infrastructure Spending Is Accelerating, and Tech Stocks Are Pricing It In

The specific figure Wade cited is worth noting: the current pace of AI infrastructure spending is exceeding what we saw over the past two years. She didn't disclose exact dollar figures in the clips available, so it's hard to put a precise number on the acceleration, but the directional claim is consistent with what major cloud providers have been signaling in their own earnings calls. Microsoft, Google, and Amazon have each revised their capex guidance upward multiple times over the past 18 months. Meta committed to spending between $60 billion and $65 billion on capex in 2025, a number that raised eyebrows when it was announced. The trajectory Wade is describing isn't coming out of nowhere.

Still, it's worth being clear about what we don't know here. Wade is a portfolio manager, not an industry analyst with proprietary supply chain data. Her view is informed by market signals and company disclosures, which means it reflects what companies are saying they'll spend, not necessarily what they've already deployed. There's a distinction between announced capex and actual installed infrastructure, and that gap can be significant, especially when supply chains for advanced semiconductors and specialized cooling equipment remain constrained. It's too early to say whether the acceleration she's describing will translate cleanly into the hardware deployment numbers that industrial suppliers care about.

From my time in hardware, one thing I learned is that announced spending and actual purchase orders are very different animals. A hyperscaler can announce $10 billion in data center investment and then spend 18 months in permitting, procurement, and engineering before a single rack gets installed. The financial markets tend to price in the announcement. The physical world runs on a different schedule.

That said, Wade's broader point about the tech rally having more runway is sort of supported by the structure of the current spending cycle. AI infrastructure isn't a one-time purchase. It's a continuous buildout that requires ongoing refreshes as model architectures evolve, as inference demands scale, and as energy efficiency targets force hardware generations to turn over faster than traditional enterprise IT cycles. The companies supplying into that cycle, whether they're making chips, power distribution units, or the industrial robots that handle logistics inside those facilities, are looking at a demand environment that doesn't reset every quarter.

This raises questions about... well, multiple things. One is whether the equity markets are pricing in the spending acceleration accurately, or whether they're front-running a cycle that could stall if AI monetization doesn't keep pace with infrastructure investment. Wade appears to believe the runway is real, and Columbia Threadneedle is a significant enough institutional player that her view isn't noise. But the gap between infrastructure investment and revenue generation from AI products remains a live debate. Some analysts argue the capex cycle is self-sustaining because competitive pressure forces every major player to keep spending regardless of near-term returns. Others counter that at some point, boards will demand to see the revenue justify the outlay, and that correction could come faster than the market expects.

For the industrial automation sector specifically, the more relevant question is whether the data center buildout is pulling manufacturing capacity in ways that affect other verticals. Semiconductor fab capacity, precision machining, and advanced materials are all inputs that AI infrastructure competes for with automotive, aerospace, and traditional industrial customers. If Wade is right that spending is accelerating beyond the pace of the past two years, that competition for capacity is going to intensify. Look, that's actually a more interesting story for industrial reporters than the stock market angle, because it has direct operational implications for manufacturers who aren't in the AI supply chain but are sourcing from the same suppliers.

The Bloomberg reporting doesn't get into that level of supply chain granularity, which is understandable given the audience and format. But it's the layer underneath the financial narrative that determines whether the acceleration Wade describes creates winners and losers beyond the obvious tech sector beneficiaries.

What's clear from the June 22 reporting is that at least one significant institutional voice is not calling a top on AI infrastructure investment. Wade's framing of "at least another couple of quarters" is conservative enough to be credible without being a reckless bull call. She's not predicting a decade of uninterrupted growth. She's saying the current momentum has more near-term runway than the skeptics are pricing in.

Whether she's right will depend on a set of variables that are genuinely hard to model: regulatory developments around AI deployment, the pace of enterprise AI adoption, energy grid constraints on data center expansion, and the competitive dynamics between US and non-US hyperscalers. None of those are small uncertainties. But the underlying demand signal, the actual orders flowing into hardware supply chains, appears to support her directional read, at least based on publicly available information.

For anyone covering the industrial side of this story, the key number to watch isn't the stock price of Nvidia or the S&P tech sector weighting. It's the capex line items in hyperscaler earnings reports, and whether the gap between announced investment and actual deployed infrastructure is narrowing or widening. If it's narrowing, Wade's call looks good. If the gap keeps growing, that's a signal that the acceleration is more financial narrative than physical reality.

The real test, as always, is production volume.

James Chen · 25 Jun · 5 min