A three-stage framework for investing in AI:
Phase 1 (2026–2027): Infrastructure continues to expand. Compute/power/data centers/chips/model training absorb a large amount of capital. Selling shovels is the most profitable.
Phase 2 (2027–2029): AI applications and agents move into large-scale acceptance/testing. Truly big companies will emerge, and many companies that look advanced but can’t charge fees will also die off.
Phase 3 (around 2029): Capex and cash flow collide positively. If application revenue can’t cover basic infrastructure costs, there will be a significant valuation compression.
Borrowing a line from Yu Wang: Around 2029 may be a dangerous point for the capital markets, but it could also be a golden starting point for AI application entrepreneurship.
Seeing which stage you’re in is more important than predicting the timing of the bubble.
#AI创业 #投资框架
Phase 1 (2026–2027): Infrastructure continues to expand. Compute/power/data centers/chips/model training absorb a large amount of capital. Selling shovels is the most profitable.
Phase 2 (2027–2029): AI applications and agents move into large-scale acceptance/testing. Truly big companies will emerge, and many companies that look advanced but can’t charge fees will also die off.
Phase 3 (around 2029): Capex and cash flow collide positively. If application revenue can’t cover basic infrastructure costs, there will be a significant valuation compression.
Borrowing a line from Yu Wang: Around 2029 may be a dangerous point for the capital markets, but it could also be a golden starting point for AI application entrepreneurship.
Seeing which stage you’re in is more important than predicting the timing of the bubble.
#AI创业 #投资框架