This Odd Lots podcast episode features an interview with Zvi Mowshowitz, an AI expert, to discuss the market impact of the newly released Chinese open-source AI model, DeepSeek. The hosts and guest analyze DeepSeek's surprisingly low training cost ($5.5 million for v3), its capabilities, and its implications for the AI industry. Mowshowitz explains that while the initial training cost was low, the overall development cost was in the hundreds of millions due to optimization efforts and limited chip access. The discussion also covers the implications of DeepSeek's open-source nature and its potential impact on companies like Nvidia, OpenAI, and Meta, highlighting the ongoing debate about the Jevons paradox in the context of AI compute demand. The interview concludes with speculation about the future of AI development and the competitive landscape.
