AI’s big spenders stumble as “picks and shovels” quietly win

Trillion‑dollar AI ambitions are running into a simple test: the market now wants proof of who actually makes money.  

The AI story has shifted from “own everything” to a much sharper question of enablers versus casualties. 

According to Reuters, the blanket AI trade that began with ChatGPT’s launch in November 2022 has fractured as soaring capex, rising debt loads and doubts about profitability push investors to draw clearer lines across stocks, sectors and regions.  

Equity and debt markets had run so far on the theme that regulators and investors started warning about bubbles, even as Microsoft, Amazon, Alphabet and Meta laid out hundreds of billions of dollars in new AI spending. 

Recent volatility suggests a turning point as markets weigh the rising cost of that build‑out against uncertain payoffs. 

The most consistent winners sit in AI’s “picks and shovels” – and especially in memory. 

Reuters said this week’s selloff in software has widened the gap between hardware makers powering AI data centres and firms further down the supply chain.  

US names such as ServiceNow and Salesforce, and European data and analytics groups like RELX and London Stock Exchange Group, all sold off sharply, while semiconductor and data‑centre‑exposed shares fell far less, extending an existing performance gap. 

Charu Chanana, chief investment strategist at Saxo, wrote that the divergence “is not a vote against AI” but a sign that investors are distinguishing between companies that enable AI and those that may be disrupted by it. 

The Wall Street Journal drew the same conclusion in memory.  

It noted that DRAM and NAND flash play a crucial role in data centres and that current production cannot satisfy AI demand and still supply PCs and smartphones.  

Counterpoint Research estimates memory prices have jumped 80 percent to 90 percent so far in the first quarter, which hurts buyers but boosts memory makers and their equipment suppliers. 

Sandisk has nearly tripled in value over three months, Micron has surged about 66 percent, and Lam Research, a tools supplier heavily tied to memory, is up more than 40 percent. 

That dynamic shows up at the country level.  

Reuters said South Korea, home to some of the largest memory producers, has become a standout market, with its KOSPI index far outpacing the S&P 500 and Europe’s STOXX 600.  

Samsung Electronics and SK Hynix have both posted strong year‑to‑date gains, and Morningstar Direct data cited by Reuters show flows into US‑listed Korean equity funds rising 20 percent in January. 

Markets are also re‑rating the “Magnificent 7” as AI spending plans grow. 

The group has stopped trading as a block as investors move from rewarding big capex announcements to demanding visible returns.  

Microsoft and Meta both raised capex, yet on the same day Microsoft fell by double digits while Meta jumped. 

Alphabet’s capex surge briefly knocked its shares lower before they closed flat, and Amazon dropped after announcing more than a 50 percent increase in this year’s capex. 

Mark Hawtin, head of global equities at Liontrust, told Reuters that there will be “huge divergence” within the seven and that as a group they could underperform this year because “the market is no longer tolerating spending for spending’s sake.” 

As per The Wall Street Journal, Google, Meta, Amazon, Microsoft and Oracle together are expected to spend about US$715bn this year, a roughly 60 percent jump from last year.  

The Journal said investors are “not exactly thrilled,” noting that Meta’s initial post‑earnings pop has faded and that Google, Amazon and Microsoft have all fallen since their reports.  

Alphabet’s share price has held up better thanks to strong core business growth and a stronger cash position, while Microsoft and Amazon have faced tougher comparisons. 

Capital is not just moving within tech; it is moving out of crowded winners. 

Reuters said investors are turning to cheaper, smaller companies and reassessing how much risk they want in volatile assets after sharp swings hit some of the market’s previous leaders.  

In the same week that software names shed about US$1tn in value, the Dow Jones Industrial Average hit a record high. 

Tim Murray, capital markets strategy at T. Rowe Price, told Reuters that while the selloff in the names that carried markets higher may have paused, “we’re instead seeing a wave of aggressive buying of altogether different stocks.”  

He said investors are weighing the risk of AI hyperscalers such as Amazon, Microsoft and Alphabet, along with the downside for business models AI might disrupt, and that “now, they’re all chasing to buy cheaper companies, perhaps indiscriminately.” 

That shift has real breadth.  

Reuters reported that on a recent rally day the Russell 2000 small‑cap index outpaced both the S&P 500 and Nasdaq 100, and that some Magnificent Seven names, including Amazon, missed the bounce as markets questioned how it will earn a return on its planned US$200bn AI capex.  

Scott Chronert, US market strategist at Citigroup, told Reuters that the market now looks increasingly divided between longtime darlings and a newer crop of stocks that investors are eyeing for returns.  

He said that while the conversation has fixated on AI, “quietly, we’ve seen money move into energy stocks, materials companies, staples and industrials,” and that those economically sensitive sectors show double‑digit gains year‑to‑date versus roughly a low‑single‑digit rise for the S&P 500. 

The most vulnerable part of the AI chain, for now, is software and data services. 

CNBC reported that software and data stocks remain under pressure after Anthropic launched a legal automation tool that revived fears of AI‑driven disruption.  

A fund manager told CNBC that shorting software has effectively become the market’s newest AI trade.  

Sharon Bell, senior European equity strategist at Goldman Sachs, wrote that companies which collate, aggregate and disseminate software and data as a service are increasingly seen as vulnerable to AI tools. 

The S&P 500 Software & Services Index is down almost 20 percent year‑to‑date, and Goldman Sachs’ Digital Economy basket recently fell 10 percent in a single day, with names such as Salesforce, Thomson Reuters and LegalZoom among the hardest hit.  

Mark Dowding, chief investment officer at RBC BlueBay Asset Management, told CNBC that software’s heavy presence in private debt and bank lending means the unwind could send ripples across credit markets. 

The Wall Street Journal added that selling in software began last year as AI coding tools fed the belief that big firms might eventually build more software in‑house rather than sign large contracts with vendors like Salesforce, Workday and ServiceNow.  

Nvidia CEO Jensen Huang called the idea that AI will “kill software” “the most illogical thing in the world,” but that has not stopped investors from re‑pricing risk. 

At the same time, not everyone sees AI as a death sentence for software.  

Anish Acharya, general partner at a16z, told CNBC that “there is a lot more software to build” and argued that AI “does automate tasks — it does not automate entire jobs.”  

He said roles such as customer support still require relationship‑building that AI does not replace.