IBM CEO warns AGI infra investment could lead to losses
What's the story
IBM Chairman and CEO Arvind Krishna has raised doubts over the economic viability of current artificial intelligence (AI) infrastructure investments. Speaking on The Verge's Decoder podcast, he said that building data centers for artificial general intelligence (AGI) could require trillions of dollars in capital expenditure. "It's my view that there's no way you're going to get a return on that," Krishna said, adding an $8 trillion investment would need about $800 billion profit just to cover interest payments.
Infrastructure investments
Major tech firms are investing heavily in AI infrastructure
Krishna's comments come amid major tech firms' investments in AI infrastructure. OpenAI, for instance, has partnered with NVIDIA to deploy at least 10 gigawatts of systems for its next-generation AI infrastructure. NVIDIA plans to invest up to $100 billion as each gigawatt is deployed. Additionally, OpenAI has a multi-year agreement with AMD for six gigawatts of GPUs and a roughly $300 billion deal with Oracle for computing power over five years.
Financial outlook
Investor scrutiny over generative AI's financial model has intensified
Despite these hefty investments, investor scrutiny over generative AI's financial model has intensified. Reports suggest that OpenAI is targeting $600 billion in compute spend through 2030 and expects over $280 billion in revenue by the decade's end. The International Energy Agency has also predicted that global electricity consumption by data centers will double to about 945 terawatt-hours by 2030 due to accelerated servers driven by AI adoption.
Investment skepticism
IBM continues to advocate for AI's enterprise value
Krishna's comments position IBM cautiously in the AI investment cycle. The company continues to advocate for AI's enterprise value while questioning if AGI-specific infrastructure can yield sufficient returns. IBM's own AI strategy focuses on enterprise adoption through products such as watsonx and hybrid cloud tools. However, only 25% of AI initiatives have met expected return on investment, according to a study conducted by IBM in May 2025.
Hiring approach
IBM's experiences with high-profile AI bets
IBM has had its own experiences with high-profile AI bets. The company's Watson system won Jeopardy! in 2011, but Krishna admitted that pushing Watson into healthcare was "inappropriate." The debate over whether AI is in a bubble has also entered IBM's hiring conversations. Dave McCann, IBM's managing partner for Europe, the Middle East and Africa, asks job candidates if they believe the industry is in an AI bubble to gage their understanding of the issue.