How an American AI Company Uses Chinese Open-Weight Models
Minerva co-founder Daniel Saeedi joins the hosts to discuss how applied enterprise AI startups navigate the choice between expensive closed frontier models and open-weight Chinese models like Qwen and Kimi. Saeedi explains the economics of fine-tuning, the security threat posed by uncensored abliterated models, the debate over model distillation, and why long-term compute demand remains structurally undersupplied.
- Accusations that Chinese AI labs win purely by scraping Western frontier models ignore architecture: distillation alone cannot bridge major capability gaps. Read →
- Rental rates on older chips continue climbing: the Silicon Data H100 rental tracking index recently hit some of its highest levels yet rather than dropping. Read →
- Data brokers are paying between $250,000 and $500,000 for complete operational datasets from companies, alongside $50,000 to $100,000 finder fees just for introductions. Read →
- Stripping the refusal weights from open-source models like DeepSeek requires changing only a few parameters, removing the built-in guardrails that stop dangerous queries. Read →