Season 1 · Episode 107
The $5.5 Million Breakthrough: DeepSeek’s AI Disruption
Discover how DeepSeek-V3 is disrupting the AI market with massive cost savings and technical innovations like Multi-Head Latent Attention.
My Weird Prompts · Daniel Rosehill
December 26, 202517m 41s
Audio is streamed directly from the publisher (dts.podtrac.com) as published in their RSS feed. Play Podcasts does not host this file. Rights-holders can request removal through the copyright & takedown page.
Show Notes
In this episode of My Weird Prompts, Herman and Corn dive deep into the seismic shift occurring in the artificial intelligence landscape as Eastern models like DeepSeek and Z.ai challenge the status quo. While Western giants like OpenAI and Anthropic spend hundreds of millions on training, DeepSeek has managed to produce world-class performance for a mere $5.5 million. The duo explores the technical "wizardry" behind this efficiency, including Multi-Head Latent Attention (MLA) and FP8 mixed precision training, which allow these models to run on less expensive hardware without sacrificing power. They also tackle the strategic implications of open-sourcing these models under MIT licenses, the impact of hardware export bans on innovation, and how Western developers are increasingly turning to these cost-effective alternatives to build the next generation of apps. Is AI intelligence becoming a cheap commodity like electricity? Join Herman and Corn as they unpack the economic and technical forces turning the AI world upside down.