The Reality Of AI Spending, Performance Delays Worries Are Hitting Stocks

Reality bites. In this case, the reality in question is the growing understanding that Silicon Valley's breathless claims of a virtually unlimited market for AI are running smack into the fact-based analysis of what it actually takes to build a data center, let alone what level of performance is required for corporate customers to begin buying AI products and services at sufficient scale for there even to be a dream of paying for the initial investments. 

The former realities have to do with the time required to get data center building permits, find sufficient materials at a reasonable price (tariffs...), the time required to build the components necessary to run the data centers (in some cases, years) and then construction workers to put them up (deportations...). And this does not even take into account growing local opposition to siting the centers because of their impact on pollution, water usage and electrical bills. That too many corporations are piloting AI and then finding it underwhelming only adds to the looming concerns that this new revolution may take a few years or decades longer than its cheerleaders would like the economy to believe. JL

Vicki Ge Huang and Sebastian Herrera report in the Wall Street Journal:

Investors had piled into shares of banks, industrial firms and materials producers early last week, with bets that lower interest rates reheat a sluggish economy sending the Dow Jones and S&P 500 to new highs. On Friday, that fell apart. Chip companies and cloud-computing purveyors have described the demand for their offerings as limitless. But recent weeks have brought home the reality that there are numerous bottlenecks on spinning up data centers and any delays in building them mean pushing back spending hundreds of billions (making) investors increasingly alert for any hint of AI infrastructure projects taking longer than expected.   "How patient are we going to be to get past the enthusiasm of the build-out of AI to when we start expecting a return?”

The potential delay of hundreds of billions of dollars in promised spending on artificial intelligence is dealing a new blow to the stock-market rally.

Investors had piled into shares of banks, industrial firms and materials producers earlier this week, with bets that lower interest rates will reheat a sluggish economy sending the Dow Jones Industrial Average and S&P 500 to new highs. On Friday, that fell apart.

Broadcom’s AVGO -11.43%decrease; red down pointing triangle worst day since the DeepSeek rout in January rekindled worries that investment in AI won’t yield the expected blockbuster profits any time soon. Shares dropped 11%, helping pull the S&P 500 and Nasdaq composite both down more than 1%.

Some analysts said the rapid reversal demonstrates how crucial to markets the AI trade remains, able to drag indexes lower even while investors try to widen their bets beyond the small group of tech giants that powered stock records earlier this year. 

“I think it’s going to be a big question for the market: How patient are we going to be for these companies to get past the enthusiasm of the build-out of AI to the timeframe when we start expecting a return?” said Steve Wyett, chief investment strategist at BOK Financial.

Broadcom provided the latest example of the dynamics on Friday. The chip designer late Thursday reported a record $18 billion in sales and strong profit growth. But the market’s response soured as investors zoomed in on questions about the margins of its custom AI chips, the timing of a massive commitment from OpenAI and the company’s ability to see over the horizon into 2027. The resulting share decline made Broadcom one of the worst performers in the S&P 500 and quickly spread to other AI companies.

 

Chip companies and cloud-computing purveyors have described the demand for their offerings as effectively limitless, with heads of the AI model companies bemoaning capacity constraints and tussling over chip allotments. But recent weeks have brought home the reality that there are numerous bottlenecks on spinning up data centers, from power supply to local politics, and any delays in building them mean pushing back spending worth hundreds of billions of dollars. 

As that understanding filters down, investors are increasingly on alert for any hint of AI infrastructure projects taking longer than expected. Shares in CoreWeave, which builds data centers with Nvidia chips, are down 26% since it revealed a delay last month. 

Coreweave fell a further 10% on Friday after a report that data centers Oracle is building for OpenAI will open later than expected. An Oracle spokesman disputed the report, saying, “There have been no delays to any sites required to meet our contractual commitments, and all milestones remain on track.” 

That followed earnings in which Oracle narrowly missed revenue estimates while disclosing greater-than-expected capital expenditures. The onetime AI darling lost 4.5% and closed the week down 13%. And the jitters reverberated, with Nvidia falling more than 4% for the week and AMD losing more than 3%.

“This is all compounding on itself,” said Rishi Jaluria, an analyst at RBC Capital Markets. “People are viewing Oracle as a barometer right now and saying ‘what does this mean for chips, or power?’ There is a lot of downstream impact.” 

It wasn’t just stocks. 

Bonds issued by so-called AI hyperscalers like Oracle, Microsoft and Meta traded at unusually high volumes, as traders rushed to reduce their exposure to the sector.

The yield premium, or spread, that investors demand to hold Oracle’s 5.95% bonds due 2055 instead of ultrasafe Treasurys jumped around 0.2 percentage point to 2.07 percentage points, according to MarketAxess, an even bigger increase than on Thursday, right after its earnings release.

The spread on CoreWeave’s 9.25% bonds due 2030 jumped around 0.3 percentage point to 7.6 percentage points.

Some investors interpret the current anxiety, driven by high valuations of AI stocks and massive AI infrastructure spending, as a healthy sign of caution that suggests the market still has room to run.

Robert Edwards, chief investment officer at Edwards Asset Management, said his clients are voicing a wide range of concerns, from AI’s impact on employment and inflation to a slowing labor market. 

He suggested that this broad worry is actually a contrarian indicator signaling a more positive outlook for the market and the economy, supported by factors like the $7.7 trillion in assets held in money-market funds and continued corporate stock buybacks. 

“My clients are definitely showing the wall of worry,” said Edwards. “The reality is things aren’t as bad as people are worried about, but as long as they’re worried and there’s money on the sidelines, and stock is shrinking in supply, you are going higher.”