Google prioritizing AI research over cloud customers, Sundar Pichai reveals
What's the story
Google CEO Sundar Pichai has revealed that the tech giant is prioritizing artificial intelligence (AI) research over its cloud customers. The decision comes as a result of the increasing demand for computing power to further AI development. Speaking on the Big Technology Podcast with Alex Kantrowitz, Pichai said that Google's DeepMind requires access to computing power as a non-negotiable requirement.
Resource allocation
Tech companies' race for computing power
Pichai's comments come as tech companies are racing for chips, servers, and data center space to train and run AI models.
He warned that accommodating new customers looking for computing power in 2027 could be "incredibly difficult."
The Google CEO explained that infrastructure planning takes time, with today's available computing capacity being a result of investment decisions made in 2024.
Business expansion
Google Cloud's growth amid challenges
Despite the challenges, Google Cloud continues to grow.
The business posted revenue of $24.8 billion in Q2, an 82% year-on-year increase. Its contracted revenue backlog was about $514 billion at the end of June, showing strong demand for its services.
However, Pichai stressed that maintaining enough computing power for Google DeepMind's research is crucial, and other businesses must plan around this requirement.
Tech evolution
Recursive self-improvement and its significance
Pichai emphasized that advanced computing resources are critical for Google DeepMind to develop frontier AI models and discover capabilities that could be integrated into products.
He anticipates that increasingly capable AI systems will help researchers create the next generation of models, a process known as recursive self-improvement.
This involves an AI system assisting with research, coding, or testing required to build a more capable model.
Demand surge
Pressure from Google's AI products and future capital expenditure
Google's own AI products are also putting pressure on available resources.
Pichai revealed that Gemini 4 Argon processed nearly 25 trillion tokens in its first 72 hours, three times more than what Gemini 3 handled in the same time frame.
Despite these challenges, Alphabet expects capital expenditure to hit $195 billion to $205 billion in 2026, with a large chunk going toward AI and Cloud infrastructure.