The great AI buildout shows no sign of slowing

By Akash Sriram

(Reuters) -A momentous week in the technology sector made it clear there is no sign the boom in building artificial intelligence infrastructure is slowing — despite the bubble talk.

Nvidia, whose processors are the AI revolution’s backbone, became the first company to surpass $5 trillion in market value. Microsoft and OpenAI inked a deal enhancing the ChatGPT maker’s fundraising ability and OpenAI promptly started laying groundwork for an initial public offering that could value the company at $1 trillion.

Amazon said it would cut 14,000 corporate jobs, just days before its cloud unit posted its strongest growth in nearly three years.

These developments, along with numerous earnings calls and interviews with executives, make clear that AI has cemented itself as the single biggest catalyst for global corporate investment and the engine of the market rally, even as some question the sustainability of both.

SPENDING WITHOUT ENDING

Soaring revenue at Microsoft, Alphabet and other technology giants was expected. But more than 100 non-tech global companies noted data centers on quarterly calls this week, including Honeywell, turbine maker GE Vernova, and heavy equipment maker Caterpillar.

Sales in Caterpillar’s division that supplies data centers jumped 31% in its most recent quarter. “We’re definitely really excited about the prime power opportunity with data centers,” CEO Joseph Creed said this week.

“The AI supply chain now spans power, industrials and cooling technology, and investors are looking at the entire ecosystem rather than just core tech,” said Ayako Yoshioka, portfolio manager at Wealth Enhancement Group.

Goldman Sachs estimates global AI-related infrastructure spending could reach $3 trillion to $4 trillion by 2030. Microsoft, Amazon, Meta and Alphabet are expected to spend roughly $350 billion combined this year.

AI investment is propping up global trade, with about 60% of U.S. data-center capex spent on imported IT equipment, according to Oxford Economics, much of it semiconductors from Taiwan, South Korea and Vietnam.

At least two dozen companies representing more than $21 trillion in combined market value reported quarterly earnings or spoke with Reuters about AI in recent days. Many, including Procter & Gamble and Boliden, noted that the hoped-for productivity gains, though uneven, are beginning to show.

“We strongly believe the future contribution of artificial intelligence within R&D, within developing innovation, will steadily increase,” Schindler CEO Paolo Compagna told Reuters, though he said AI’s impact is yet to be seen. The Swiss lift and escalator maker raised its annual margin forecast last week.

Year-over-year revenue growth in the U.S. tech sector is up more than 15%, outpacing all other sectors, according to LSEG data.

Apple said it was significantly increasing AI investment and Amazon projected capital spending of $125 billion in 2025.

WORRIES ABOUT OVERVALUATION

Since ChatGPT’s debut in 2022, global equity values have climbed 46%, or $46 trillion. One-third of that gain has come from AI-linked companies, according to Bespoke Investment Group.

Analysts warn of a quickening replacement cycle for servers, accelerators and chips as each new generation delivers exponential performance gains. The useful life of AI chips is shrinking to five years or less, forcing companies to “write down assets faster and replace them sooner,” said UBS semiconductor analyst Tim Arcuri.

The surge in AI-related spending has widened the gap between investment and returns, with a Reuters analysis showing that sales-to-capex ratios at major tech firms have fallen sharply as outlays on chips and data centers grow faster than revenue. Capital expenditures represent a larger chunk of cash generated by operating activities for some companies, causing some investor concern.

“If progress hasn’t been made toward monetization within three years, the market will start asking hard questions,” said Sumali Sanyal, senior portfolio manager at investment firm Xponance.

Microsoft reported a record $35 billion in capex in its most recent quarter and projected higher spending, prompting Bernstein analyst Mark Moerdler to ask whether the company was spending into a bubble. Microsoft Chief Financial Officer Amy Hood responded that AI-related demand still outpaces Microsoft’s spending. “I thought we were going to catch up. We are not,” she said.

Some companies are financing AI projects with debt. Oracle’s $18 billion bond sale last month was one of the largest ever for a tech company, and it looks set to be surpassed by an up to $30 billion bond sale from Meta Platforms. News of its largest ever bond sale knocked Meta’s shares down 11% on Thursday.

Still, many economists say the AI cycle is far from exhausted. Goldman estimates AI investment is currently less than 1% of U.S. GDP, far below peaks of 2% to 5% seen during the electricity and dot-com booms.

“We are in the early innings … and the pace of AI innovation is the fastest we have seen in decades,” said Nick Evans, portfolio manager at Polar Capital Technology Trust.

(Reporting by Akash Sriram, Sriparna Roy, Sneha SK and Puyaan Singh in Bengaluru and Jessica DiNapoli in New York, Bernadette Hogg in Gdansk, Poland; Editing by David Gaffen and Rod Nickel)