Baillie Gifford is an Edinburgh investment partnership founded in 1908, famous for ultra-long-horizon, high-conviction growth investing — its early stakes in Amazon, Tesla and NIO are classics. Its "actual investors" philosophy holds world-changing companies on 5-10 year views; AUM is around $120bn. The Insights column carries its managers' investment views and thematic research.
This report says AI can design protein drugs in seconds, but they still need to be physically built and tested. Twist Bioscience's DNA synthesis platform is like the factory for that, making it key infrastructure for AI-driven drug discovery. The author is bullish on Twist, arguing AI will create more demand for physical testing, not less. Key holdings: Twist Bioscience (AI-related orders jumped from near zero to ~$25 million in 2025, expected to exceed $100 million by 2027); Anthropic (its AI Claude designs proteins with 22-27% hit rate vs. 10-15% traditional, but all designs go to Twist to build); and hyperscale cloud providers (new customers paying more for full antibody characterization, raising revenue per order from ~$100 to ~$400).
One-sentence summary: The author holds a strongly [bullish] stance on Twist Bioscience’s infrastructure role in AI-driven drug discovery, arguing that its DNA synthesis platform will benefit from the explosion of the “AI design, physical testing” closed loop.
The report argues that AI can design proteins in seconds, but physical synthesis and testing are ultimately required, positioning Twist Bioscience's DNA synthesis platform as a critical infrastructure layer for AI-driven drug discovery. The author draws an analogy to the software industry: AI reduces the cost of writing code, thereby generating more software demand. Similarly, after AI lowers the cost of designing drug candidate molecules, physical synthesis and testing become the new bottleneck. The author states, "AI doesn't eliminate experimentation – it may create more of it," meaning: "AI does not eliminate experiments—it may generate more of them." The core thesis is that Twist's competitive advantage lies in its silicon-based miniaturization process, enabling low-cost, large-scale, high-speed parallel synthesis of DNA strands. Its current production capacity is far from theoretical limits and can scale with growing AI demand.
Anthropic's experiment shows that Claude-designed protein binders achieve a hit rate (22-27%) far exceeding traditional methods (10-15%), but all designs require Twist for physical synthesis and testing, forming a closed loop of "AI designs, Twist builds and tests." The author notes that Claude successfully generated novel protein binders against 14 out of 15 targets, with binding affinities comparable to high-quality natural antibodies. However, Claude cannot self-validate—over 1,200 computer-generated DNA designs were sent to Twist for manufacturing and physical characterization. The author states, "Claude designs; Twist builds and tests," meaning: "Claude handles design; Twist handles construction and testing." This relationship may become increasingly important.
Twist's AI-related orders are projected to grow from approximately $25 million in fiscal 2025 (nearly zero the prior year) to at least double in fiscal 2026, and potentially double again to over $100 million in fiscal 2027. The report argues that this growth trend indicates Twist is transitioning from a pure DNA manufacturer to an infrastructure layer for AI-driven drug discovery. The author emphasizes that AI-native companies and hyperscale cloud providers—rather than traditional pharmaceutical firms—are becoming new customers. These clients increasingly want Twist not only to manufacture DNA but also to perform protein expression and characterization, raising per-order revenue from about $100 (DNA only) to approximately $400 (fully characterized antibodies).
The report clearly favors Twist Bioscience's infrastructure role in AI-driven drug discovery, but readers should note this is a long-position perspective. The author uses the "design-build-test-learn" cycle to argue its long-term value, while only briefly mentioning competitive risks. The core investment thesis is that AI will not reduce experimental demand but will instead create more experiments, making Twist's scalable synthesis capacity a scarce resource. Key variables to monitor are whether Twist can sustain its advantages in cost, scale, speed, automation, and service breadth to fend off competitors.
| Ticker | Direction | Author's One-Sentence View | Key Data |
|---|---|---|---|
| Twist Bioscience | Hold & Watch | Bullish on its long-term value as an AI drug discovery infrastructure layer, but acknowledges competitive risks | AI-related orders expected to rise from approximately $25 million in fiscal 2025 to over $100 million by 2027; per-order revenue increasing from $100 to $400 |