Funding rounds and IPOs raising 11-figure sums. Blockbuster bond sales spanning three continents. The casual announcement of an $85 billion equity raise.
Such is life on Wall Street at the dawn of the artificial-intelligence build-out. Tech companies are hungry for cash to invest in data centers, and investors are forking it over through all possible means, in all parts of the globe -- a flurry of fundraising that has mostly supported markets by powering technological advances, even as it tests their ability to absorb it all.
Alphabet's announcement that it would raise $85 billion of equity was just the latest example. SpaceX, Anthropic and OpenAI are poised for public listings that could make this year the biggest ever for money raised through IPOs.
Meanwhile, the so-called AI hyperscalers Alphabet, Amazon.com, Meta, Microsoft and Oracle have issued $159 billion of bonds globally this year -- up from $108 billion all of last year and just $17 billion in 2024, according to Dealogic. Data-center developers have raised billions more in the high-yield bond market, while upstart AI cloud companies are borrowing from banks and private-credit firms to fund purchases of chips.
Plenty of investors still think that the AI build-out could get messy, with companies overspending and Wall Street eventually weeding out the losers. In the meantime, increased equity issuance could pressure stocks by diluting shareholders -- one possible factor behind last week's tech selloff.
Still, demand from investors has generally defied the skeptics so far. The extra return that investors get for holding U.S. companies' bonds instead of ultrasafe Treasurys is hovering near multidecade lows. Tech stocks in the S&P 500 remain up 31% this quarter, with many citing encouraging fundamentals, such as businesses' surging spending on AI tools and the recent news that Anthropic would turn a quarterly profit ahead of expectations.
"I think there's been a few signals that have become more positive for the AI infrastructure build-out," said David Lefkowitz, head of U.S. equities at UBS Global Wealth Management. "That's helped give investors more confidence in the return prospects for the investment."
Tech shares rebounded Monday, lifting the Nasdaq composite to a 0.9% gain. The S&P 500 added 0.3%, while the Dow Jones Industrial Average slipped 0.2%, or 81 points.
Spending on data centers and other AI infrastructure by just four big tech companies this year is expected to total more than $670 billion -- a larger investment as a share of the economy than even the railroad expansion of the 1850s.
That scale has raised worries that Wall Street is inflating a bubble. Of particular concern are questions about how profitable it is to maintain the likes of OpenAI's ChatGPT, Alphabet's Gemini and Anthropic's Claude -- the products at the heart of the investment boom.
Most investors have long had faith that AI models could eventually be monetized. But that optimism has grown this year with the rise of Anthropic, which has focused on subscription-based coding tools for businesses. The company is poised to double its revenue to $10.9 billion in the second quarter, at least temporarily exceeding the hefty cost of training and running its models, The Wall Street Journal has reported.
OpenAI and Anthropic have each raised more than $100 billion in venture capital financing. Established tech giants like Alphabet and Amazon initially funded most of their AI investments with the gusher of cash generated from their legacy businesses. But the daunting expenses pushed all of the hyperscalers but Microsoft to the bond market last year.
This year, they have gone international. Alphabet has issued bonds not only in U.S. dollars, but also in Canadian dollars, Japanese yen, euros, Swiss francs and British pounds -- the currency in which it issued a rare 100-year bond. The company is also borrowing $1 billion for energy financing in the California municipal-bond market, according to a preliminary prospectus filed last week.
Amazon was poised to issue Canadian-dollar bonds on Monday after issuing bonds in U.S. dollars, euros and Swiss francs earlier in the year.
While big tech's borrowing binge has raised comparisons to previous credit booms that went bust, a key difference is that many of the companies now involved are hugely profitable.
Reflecting that strength, the extra yield, or spread, that investors demand to hold 10-year Alphabet, Amazon and Microsoft bonds over U.S. Treasurys is still below that of the average investment-grade corporate bond. The spread on Meta bonds is slightly above that average.
There are exceptions. Oracle, which has issued $43 billion of bonds since last September, is projected to burn tens of billions of dollars over the next several years as it tries to transform itself from a leading software company into a cloud-computing giant -- leasing out the vast clusters of advanced computer chips to OpenAI and others. Though rated investment-grade, its bonds trade more in line with the highest tier of speculative-grade debt.
"Overbuild risk isn't going away," and that risk is generally reflected in hyperscaler bond spreads, said Jordan Chalfin, a senior analyst at research firm CreditSights.
Still, Oracle's bonds have rallied recently along with the broad investment-grade market. Despite plans to issue at least $20 billion of new shares, its stock is also up 8.7% this year.
Investor sentiment has improved even more drastically for the speculative-grade CoreWeave, a former bitcoin miner turned prominent AI cloud-computing provider. At one point last year, the company's ability to borrow in the bond market was in question after a steep selloff in its shares and bonds following news that it faced data-center construction delays.
This year, though, CoreWeave's stock has rallied 43% and spreads on its bonds have tightened by around 4 percentage points, enabling the company to raise more than $20 billion through sales of stock and debt in 2026.
At this point, even some skeptical investors say there is little point in betting against the AI boom when it appears to be just getting started.
Eventually, it is likely that companies will build too much AI infrastructure and stock prices will fall, said Ayako Yoshioka, senior investment strategist at the wealth-management firm Wealth Enhancement.
Now, though, "there's still time to invest because this build-out, the scale of it, is just so large," she said.
Comments