Home / Business and Economy / IBM CEO: AI Data Center Costs Are Unsustainable
IBM CEO: AI Data Center Costs Are Unsustainable
2 Dec
Summary
- IBM CEO Arvind Krishna estimates $8 trillion in data center capex.
- Krishna believes current costs make profit impossible for AI firms.
- He estimates a 0-1% chance of current tech reaching AGI.
IBM CEO Arvind Krishna has voiced significant concerns regarding the financial viability of the massive investments being made in AI data centers. He calculated that global commitments for computing power in the pursuit of Artificial General Intelligence (AGI) could reach as high as $8 trillion. Krishna's analysis suggests that the current cost structure for building and maintaining these facilities makes profitability virtually unattainable.
Krishna elaborated on his "napkin math," explaining that the interest alone on $8 trillion in capital expenditure would require roughly $800 billion in profit annually. Furthermore, he pointed to the rapid depreciation of AI chips, which may need replacement within five years, adding to the substantial ongoing costs. These financial hurdles cast doubt on whether the current technological trajectory can yield a positive return on investment.
Adding to the skepticism, Krishna estimated the likelihood of current technology achieving AGI to be between 0% and 1%. This contrasts with the aggressive spending and capacity expansion announcements from major tech companies, underscoring a potential disconnect between ambition and economic reality in the race for advanced AI.
