Alphabet Inc. (NASDAQ:GOOGL) CEO Sundar Pichai says Google is giving up potential Cloud revenue to prioritize AI research, even as customers struggle to secure computing capacity.
Google Cloud revenue surged 82% to $24.8 billion in the second quarter, while its contracted backlog reached nearly $514 billion.
“Our cloud business is at any given moment in time forgoing short-term revenues because we are making compute trade-offs to support frontier AI development,” Pichai told Alex Kantrowitz on the Big Technology Podcast Tuesday.
AI Researchers Get First Claim on Computing Power
Pichai said ensuring Google DeepMind has enough computing capacity to develop cutting-edge AI models is “non-negotiable,” with the company’s other businesses planning around that commitment.
He argued that the most advanced AI models could help develop their successors, making research a strategic priority even when commercial demand is booming.
That demand is already stretching into next year.
Pichai said accommodating new customers requesting computing capacity in 2027 would be “incredibly difficult to do.”
“Everyone wishes you had 2x of what you had planned for,” he said, describing shortages across the industry.
SpaceX Cashes in on the Compute Crunch
That shortage is helping fuel extraordinary growth at SpaceX (NASDAQ:SPCX), which reported $14.6 billion in quarterly revenue Tuesday, beating Wall Street expectations.
Its Super Intelligence division generated $7.6 billion, up 830% year over year, partly driven by customers renting computing power.
SpaceX CFO Bret Johnsen said on Tuesday’s earnings call that new computing infrastructure is paying for itself in less than a year at current prices.
Google itself has agreed to pay SpaceX $920 million monthly for computing capacity starting in October.
Even $205 Billion Cannot Quickly End the Shortage
Alphabet expects to spend between $195 billion and $205 billion on capital expenditures this year, much of it driven by its expanding AI infrastructure.
“The compute we have today is a function of decisions I made in 2024,” Pichai said, highlighting the years required to build new data centers.
Pichai revealed that Gemini 4 Argon processed nearly 25 trillion tokens in its first 72 hours, roughly three times Gemini 3’s initial usage. Google is already capacity-constrained serving the new model.
Google’s pursuit of AI leadership is also playing out on Polymarket, where traders give the company roughly an 82% chance of having the highest-ranked AI model at the end of October, compared with 17% for Anthropic.
The market has attracted more than $3 million in trading volume and resolves using the Arena text-model leaderboard.
With Alphabet reporting earnings on Oct. 28, investors will be weighing the returns on its surging AI spending.
Pichai did not disclose how much Cloud revenue Google is forgoing, leaving the financial cost of prioritizing AI research unclear.
Image: Shutterstock
Recent Comments