【起業家必読】累計26万部『起業の科学』がアップデート｜今、起業家が知るべき新フレームワーク #スタートアップ #起業の科学

Tadokoro Masayuki unveils Startup Science Ver.2, arguing AI-native startups have killed the J-curve and must now compete on durable moats, not product alone.

• Startup Science Ver.2 released March 24, 2026 by Diamond Publishing; cumulative sales of the series now 260,000 copies, up from 230,000 for the original.
• The original 2017 edition had 230,000 characters; Ver.2 expanded to 270,000 characters, with ~90% of case studies replaced because examples like Mercari had not yet IPO’d and Sam Altman was still running Y Combinator.
• Tadokoro replaces the J-curve model with a Mu-curve: generative AI (e.g., Claude Code) compresses prototype costs from months and millions of yen to weeks and tens of thousands of yen, making early-stage burn far shallower.
• The new framework distinguishes Sustainability (unit economics) from Scalability and treats both as required after PMF—scaling a leaky bucket is explicitly called out as a fatal pattern.
• SmartHR is used as a flywheel case study: labor-management data becomes a system of record so sticky that customers layer on evaluation, talent simulation, and career tools, building compounding lock-in across HR.
• A new three-layer time-horizon framework separates Mega-trend (decade-scale), Trend (3-year strategy), and Cycle (seasonal campaign)—tapioca shops are cited as a cycle-only bet that predictably collapsed.
• Tesla’s sequencing (Roadster → Model S → Model 3) is analyzed as backcasting from a mega-vision with disciplined phase strategy, not chaos—Tadokoro says this was under-explained in Version 1.
• Ver.2 dedicates roughly half its content to context engineering for AI-native product development, including prompt and spec-writing techniques embedded inside the PSF (Product-Solution Fit) chapter.

2026-03-25 · Watch on YouTube