IBM CEO Arvind Krishna said he has serious doubts about rivals’ heavy capital spending aimed at achieving artificial general intelligence (AGI), arguing the economics do not support a viable return on investment.
According to ChainCatcher, Krishna estimated that building a 1-gigawatt data center currently costs about $80 billion. He said that if the industry were to deploy 100 gigawatts of computing power to pursue AGI, total capital expenditures would reach about $8 trillion.
Krishna added that such spending would require companies to generate $800 billion in profit just to cover interest payments, which he said makes the investment case commercially unrealistic.
He also said the probability of achieving AGI using currently known large language model (LLM) technology is extremely low—around 0% to 1%—unless there is a foundational breakthrough in integrating knowledge with LLMs.
Despite his skepticism about near-term AGI, Krishna said he still expects generative AI to unlock productivity value worth several trillion dollars for enterprise users.
IBM CEO Arvind Krishna Questions Massive AGI Spending and Puts LLM-Based AGI Odds at 0%–1%
2026-06-22 05:13:57
Disclaimer:
1. The information provided does not constitute investment advice. Investors should make independent decisions and bear all risks themselves.
2. The copyright of this content belongs to the original author. The views expressed herein are solely those of the author and do not represent the stance or position of this website.
Previous article：
Oil Falls, Global Stocks Diverge After US-Iran TalksNext article：
BCG Hit by Saudi Arabia’s Spending Pullback
