Andy Plump

Head of R&D at Takeda for twelve years. Presented at Session II on learning from failure.

Key Arguments

• Drug discovery’s funnel slide is an average — biology lacks calculus
• Failure is biology’s tax, and the only edge is failing differently each time
• The industry is too inefficient, too costly, and too failure-prone to keep doing things the old way

Two Formative Failures

TAK-994 (orexin 2 receptor agonist)

• Pioneering OX2R program for narcolepsy type 1
• Died in phase 2b (2021) from severe liver toxicity; stock dropped 15%
• Backup molecule — designed from first principles two years behind — lacks the liability and is about to be registered

Generalized myasthenia gravis biologic

• Strong biology, broken biophysics: aggregation, instability, low yield
• After years of conventional engineering, handed to Nabla Bio: returned an in-silico-designed sequence that no traditional method would have produced — it worked
• Flipped Takeda’s stance on AI: new Cambridge wet lab will be AI-driven by default

Context

• Under Plump’s tenure: ~15K → ~50K employees; ~$10B\to$30B revenue; Japan from ~50% to ~10% of both
• Founded 1781 — eleven of first thirteen CEOs were named Takeda

“You’re going to fail in this business. The question is, how are you going to fail differently going forward?”