Archive · 14 Dec 2025
IBM CEO Warns on AI Spending
IBM's CEO, Arvind Krishna, recently talked about how much money companies are spending on AI.
He said that building big AI data centers is very expensive, costing up to $1.5 trillion for large projects.
He also mentioned that the special computer chips used for AI need to be replaced often, which adds to the cost.
Krishna thinks that spending so much money on AI might not be worth it because the chance of creating something called AGI (a super-smart AI that can do anything a human can) is very low, around 0-1%.
He believes companies should focus on using AI to make things better now, rather than spending too much on future projects that might not work.
Big companies like Alphabet and Amazon are still spending a lot on AI, but Krishna is worried about the costs and the real benefits.
IBM CEO Arvind Krishna warns about the high costs and potential unsustainability of AI infrastructure spending.
Filling a 1-gigawatt data center costs around $80 billion, with larger projects potentially costing up to $1.5 trillion.
AI hardware has a short lifespan, requiring frequent replacement and compounding long-term costs.
Krishna estimates that an $8 trillion investment in AGI would need $800 billion in profit just to cover interest payments.
Krishna believes current AI technologies have a 0-1% chance of achieving true artificial general intelligence (AGI).
- Who
- IBM CEO Arvind Krishna
- What
- Warning about the financial risks of AI infrastructure spending and the low chance of achieving AGI with current technology
- When
- During an interview on The Decoder podcast
- Where
- Not specified
- Why
- To highlight the extraordinary expenses and potential economic unsustainability of pursuing AGI
- Estimated cost to fill a 1-gigawatt data center
- $80 billion
- Estimated cost for 20-30 gigawatts of infrastructure
- Up to $1.5 trillion
- Estimated investment in AGI
- $8 trillion
- Estimated profit needed to service AGI interest
- $800 billion
- Estimated AI infrastructure investment for 2024
- $380 billion
- IBM CEO's estimated chance of achieving AGI with current tech
- 0-1%
- Alphabet's 2025 capital expenditure outlook
- $91-93 billion
- Amazon's 2025 capital expenditure estimate
- $125 billion