a16z Podcast: When (and How) Biology Becomes Engineering

Episode Summary

Executive Summary: The conversation argues that biology is shifting from a stochastic, bespoke science model toward an engineering model built on modular parts, reproducibility, and iterative design. This shift—enabled by tools like CRISPR, machine learning, and engineered biological platforms—could change drug discovery, academia, go-to-market motion, and the role of biotech companies from asset-focused to data- and platform-driven.

Main Topics: Science vs. engineering in biology (Priority: 5/5): The speakers contrast hypothesis-driven, high-risk scientific discovery with design-build-test-refine engineering, arguing biology is increasingly moving toward the latter. Biology as modular 'Legos' (Priority: 5/5): They frame biological systems as a hierarchy of parts—atoms to ecosystems—and suggest that identifying reusable components enables more predictable construction of biological solutions. Engineering disciplines entering biotech (Priority: 4/5): Mechanical, electrical, materials, and computer science principles are increasingly applied to biological systems, especially in areas like tissues, bone, muscle, and systems design. Genetic engineering and CRISPR (Priority: 5/5): Genetic engineering is presented as historically more exploratory than truly engineered, but newer tools like CRISPR and DNA design platforms are making it more deterministic and design-based. Machine learning and data generation (Priority: 4/5): ML is described as a way to turn biology into a more engineered process by creating reusable learning from data, reducing bespoke discovery and improving iteration speed. Go-to-market and proof of concept (Priority: 5/5): For buyers and partners, the key activation energy is demonstrating reproducible performance in multiple contexts; this can unlock pilots, small deals, and eventually land-and-expand dynamics. Platforms, compounding, and future biotech companies (Priority: 5/5): Platforms are becoming more valuable when they reliably create multiple assets, making later assets more valuable than earlier ones because each iteration compounds learning and capability.

Key Arguments: Biology becomes more engineering-like when its parts can be identified, standardized, and recombined predictably. CAR T is cited as an example of combining biological 'parts' to create a new therapeutic system. The industry is moving from molecule-centric drugs to cell- and system-based drugs, which increases the importance of modularity. High-throughput biology can support engineering, but the bigger shift is using experimental output to define reusable parts rather than endless screening. CRISPR and companies like Asimov are helping turn DNA into a design medium rather than a purely exploratory scientific substrate. Machine learning can make discovery less bespoke by creating a reusable process that improves as new data is added. In an engineering paradigm, failure data is valuable because both false positives and true positives inform the model and improve future design. Pharma may evolve into a data-generating and data-science-heavy industry, with larger dry labs and more drug designers than traditional bench-only roles. Platform value increases when each new program learns from the last, making the second and third assets more valuable than the first. For business development, evidence of reproducibility and predictable performance matters more than a single publication or one-off result. Land-and-expand dynamics become possible in biotech when a platform can prove value in one use case and scale into adjacent ones. Engineering progress in biology may be exponential via compounding improvements, rather than linear, if each iteration improves the platform systematically.

Data Points: Department of Bioengineering establishment: created in the last ~10 years - Used to illustrate how academia is reorganizing around the engineering-biology convergence. Synthetic biology historical hit rate: 10,000 things tried to get 1 thing to work - Describes the traditional high-stochasticity model contrasted with engineering-like predictability. Technology improvement rate: ~30% better every year - Referenced as an example of compounding progress that can double performance roughly every two years. Doubling time: every 2 years - Used to explain how sustained ~30% annual improvement compounds rapidly. Rice-grain example growth: 2 to the 32nd power (~4 million grains) - Illustrates exponential growth and how small iterative gains can become massive over time.

Pivotal Quotes: "What does it mean for biology to move from the high-risk, painstaking realm of the laboratory bench to the lower-risk, get-it-done world of engineering?" — Host narration: Sets up the core thesis of the episode: biology becoming more engineering-like. "If you understand where the screws need to go, then ... you'll eventually get to the moon." — Jorge Conde: Explains how modular understanding and stepwise engineering can turn impossible goals into achievable programs. "The second drug is more valuable than the first if you're using engineering principles because what you learned from example one sort of imbues value to example two." — Jorge Conde: Describes how knowledge compounds across assets in an engineering-driven platform model.

Implications: Biotech winners may increasingly look like engineering/data platforms: reproducible, iterative, and scalable. Expect more dry-lab talent, stronger platform valuation, faster partner adoption, and a shift from one-off discoveries to compounding biological product systems.