Decoder with Nilay Patel IBM CEO Arvind Krishna says there is no AI bubble after all
30 snips
Dec 1, 2025 
In a captivating discussion, Arvind Krishna, Chairman and CEO of IBM, explores the company's evolution in the face of AI advancements. He reveals IBM's shift from Watson-era technology to modern LLMs, addressing lessons learned from early adoption. As AI reshapes the job landscape, he emphasizes the importance of retraining while still committing to hiring. Krishna also dives into the potential of quantum computing and its future impact, advocating for a balanced view on AI's economic promises and challenges.
AI Snips
Chapters
Transcript
Episode notes
Watson Sparked AI, But Was Too Monolithic
- Watson put AI on the map but was packaged too monolithically for real-world adoption.
- Arvind says modern AI needs modular building blocks engineers can tune and compose.
Why LLMs Scaled AI Rapidly
- LLMs reduced labeling costs and made models broadly tunable across tasks.
- Arvind calls the LLM era a ~100x productivity unlock over prior bespoke deep learning.
Plan For Multi‑Year Cost Declines
- Expect hardware, architecture, and software optimizations to cut AI costs over years.
- Plan for incremental 5-year improvements rather than immediate cheap compute.