Chamath Warns of America's AI Cost Disadvantage
Chamath Palihapitiya pointed out that leading AI has diverged into two paths: the closed-source models in the U.S. costing $26-56 per million tokens, and the open-source weight models in China costing $0.50-1. If U.S. companies are forced to bear costs 50-100 times higher, it will lead to financial damage and a stock market crash, similar to the government forcing purchases of oil at $800 when the market price is $80. In the short term, revenues for U.S. labs may be maintained, but customer losses and overseas users turning to cheaper options will ultimately drag down revenue. U.S. AI companies and investors are under pressure from high costs, while beneficiaries of open models are expanding their global share; capital is accelerating the shift from closed-source to hybrid or open-source strategies, with the AI industry transitioning from U.S. dominance to cost-competitive restructuring.
ABAB AI Insight
Chamath has previously invested in early AI and tech projects through Social Capital and has publicly criticized the overvaluation bubble and excessive regulation in the U.S., accurately predicting adjustments in SPACs and tech stocks after 2022. In terms of capital flow, Chamath advocates for resources to shift towards low-cost open-source and hybrid solutions, motivated by the need to prevent U.S. companies from losing global competitiveness due to policies or closed strategies, prioritizing long-term shareholder returns over short-term lab revenues. Similar to the semiconductor industry, where the U.S. relied on high-priced proprietary technology but was disrupted by low-cost manufacturing in Asia, and the transition of mobile operating systems from closed to open Android, the U.S. AI sector is currently in the early stages of losing competitiveness due to cost barriers. Essentially, this is a transfer of pricing power: the high-priced closed-source strategy cannot maintain global dominance in the long term, as the mechanism of rapidly iterating open-source weight models compresses the cost curve, forcing capital to shift from U.S. labs to the Chinese supply chain and open-source ecosystem, achieving cross-border efficiency arbitrage. ABAB News · Cognitive Law 1. Cost 50 times, competitiveness zero. 2. Closed moats become financial graves. 3. In global competition, being expensive means losing.