
Bloomberg Businessweek Bloomberg Businessweek Weekend: November 20th, 2026
Nov 21, 2026
Sebastian Dusterhoff, a Lightspeed Venture Partners investor specializing in AI and enterprise software, discusses why the firm backed Anthropic early, AI safety, open versus closed models, and the economics of computing. The program also explores China’s EV rise, food rescue efforts, and how younger consumers are reshaping drinking culture.
AI Snips
Chapters
Books
Transcript
Episode notes
Why AI Safety Can Support Commercial Growth
- Anthropic treats safety and commercial success as mutually reinforcing because enterprise customers need AI systems they can trust.
- Dusterhoff compares AI’s dual-use nature with earlier technologies such as fertilizers and dynamite.
Anthropic Turns Compute Constraints Into Strategy
- Anthropic’s compute strategy emerged from capital constraints, forcing the company to allocate one pool across training, inference, internal use, and safety.
- Compute remains costly, but distributing trained models across many customers creates a meaningful scale effect.
AI’s Long-Term Impact Will Not Be A Straight Line
- Dusterhoff expects AI to transform every industry and job category, but warns that investment returns will not follow a straight line.
- Periods of overvaluation and companies misjudging costs are likely even if the technology’s long-term impact proves substantial.



