IBM CEO Arvind Krishna has raised serious concerns about the skyrocketing investments being poured into artificial intelligence infrastructure by industry giants such as Microsoft, Google, and Amazon. While AI continues to transform industries and attract massive capital, Krishna argues that the current pace and scale of spending on data centers and specialized AI hardware may be economically unsustainable.
Krishna estimates that building and equipping a 1-gigawatt AI data center now costs around $80 billion. With major tech companies collectively signaling interest in constructing approximately 100 GW worth of AI capacity, the total investment could reach $8 trillion.
According to Krishna, there is “no way” this type of investment can generate a viable return under current economic conditions. An infrastructure project of this magnitude would require around $800 billion in annual profit just to cover the cost of capital a target he suggests is highly unrealistic.
Another issue Krishna highlights is the rapid turnover cycle of AI accelerators like GPUs. These chips are typically depreciated over five years, but technological improvements are so fast that companies cannot simply extend or reuse hardware they must completely replace it every cycle. This creates what Krishna calls a “compounding financial burden”, raising questions about how long companies can sustain the current arms race.
Despite the global push toward Artificial General Intelligence (AGI), Krishna remains skeptical that today’s large language models or current AI frameworks will lead to it. He gives the probability of achieving AGI with existing technology at just 1%.
However, he remains bullish on near-term AI benefits, calling current models “incredibly useful for enterprise” and predicting they will unlock trillions of dollars in productivity across industries such as finance, cybersecurity, logistics, and healthcare.
Krishna also addressed the recent wave of tech layoffs. He rejects the idea that AI is replacing workers, instead attributing the cuts to companies over-hiring during the pandemic in some cases by 40% to 100% growth. While he acknowledges AI could eventually affect up to 10% of U.S. jobs, he believes overall employment will increase as new roles emerge.
Krishna’s warning serves as a reality check: while AI presents transformative potential, the industry may be entering a phase where financial sustainability and strategic investment matter just as much as innovation.
