IBM CEO Warns AI Infrastructure Costs
IBM CEO Arvind Krishna warned on The Decoder podcast that the industry's push to build AI infrastructure may be financially unsustainable, estimating about $80 billion per one‑gigawatt data center and roughly $1.5 trillion to equip 20–30 gigawatts. He said a 100‑gigawatt AGI pursuit implies about $8 trillion in capital and $800 billion a year to service interest. Krishna noted chips depreciate every five years and rated current LLMs' AGI chance at 0–1% without breakthroughs.
Key Points
- 1Estimates $80 billion per 1GW and about $8 trillion for 100GW global AI infrastructure.
- 2Highlights unsustainable capital and interest burden, requiring roughly $800 billion annually to service costs.
- 3Implies practitioners must factor five‑year chip depreciation into long‑term capital and deployment planning.
Scoring Rationale
Official CEO cost estimates signal industry-wide financial risk; limited novelty because it's commentary, not a technical breakthrough.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
