Group 1 - The core concern raised by Ben Thompson is the conservative capacity expansion of TSMC, which he believes is a limiting factor for global AI expansion [2][3] - Thompson predicts a significant chip shortage around 2029 due to insufficient capital expenditure growth to meet the exponential demand for computing power driven by AI [2][3] - He emphasizes that TSMC's cautious approach to capacity expansion is rational, as they prefer to avoid the risks associated with overcapacity and its impact on profit margins [2][3] Group 2 - Thompson advocates for tech giants to support companies like Intel or Samsung through prepayments or other means to mitigate future capacity bottlenecks [3] - He argues that the advertising model is the most effective monetization strategy for AI applications, countering the prevalent skepticism in Silicon Valley regarding advertising [4][5] - Thompson cites Facebook's advertising system as a successful automated agent, highlighting its effectiveness in delivering results for businesses [4][5] Group 3 - Thompson provides insights on the performance of major tech companies, labeling Meta as the strongest in execution despite concerns over its capital expenditures [5] - He describes Google as chaotic yet resilient, comparing it to a slime mold that adapts effectively despite its apparent disorder [5] - Concerns are raised about Amazon's chip strategy in the AI era, suggesting that its low-cost approach may not be sustainable in a rapidly evolving market [5] Group 4 - Thompson discusses the potential end of the SaaS business model if AI leads to a reduction in workforce, indicating a growth ceiling for per-seat pricing [6] - He posits that in a world of infinite content, live experiences will gain value, as they cannot be personalized by AI [7] - The future of AI-generated content will redefine value based on scarcity, emphasizing the importance of shared experiences [7]
Stratechery创始人深度访谈：预警2029年“芯片荒”，SaaS模式将终结，广告才是AI终极商业闭环
Hua Er Jie Jian Wen·2026-02-15 10:02