China's strategy of releasing powerful, free open-source AI models is not just about technological competition. It's an economic play to commoditize and deflate the value of the US service sector, where AI's impact is largest, giving China a strategic advantage.
Newer AI models may have low per-token prices but are often "token hungry," requiring more tokens to complete a task. This can make them more expensive overall. The true measure of economic viability is the final cost-per-task, not the misleading per-token price.
The market for AI models is bifurcating. Users either pay a premium for top-tier frontier models for high-stakes tasks like cybersecurity or use extremely cheap, small models for high-volume, simple tasks. Mid-tier models struggle to find a viable use case, getting squeezed from both ends.
As more of the internet and code repositories are generated by leading AI models, any new model trained on this public data inadvertently "distills" the knowledge and quirks of those proprietary systems. This blurs the line between original training and outright copying.
Soaring demand for high-bandwidth memory (HBM) for AI has created a supply bottleneck, allowing chip makers like Micron to quadruple prices. This has led to an enormous transfer of wealth, making them the primary financial beneficiaries of the AI boom while model providers bear the costs.
Mark Zuckerberg initially agreed to sell Facebook to Yahoo for $1 billion. However, after Yahoo's stock dipped, CEO Terry Semmel cut the offer to $800 million. This act of renegotiation destroyed Zuckerberg's trust in Yahoo as a partner, causing him to reject the deal even after the original $1B offer was restored.
The performance gap between frontier closed-source AI and open-source models provides a crucial window for cybersecurity. "White hat" hackers use the most advanced models to find vulnerabilities before "black hat" hackers can exploit them with widely available open-source tools.