Twist Bioscience surged after Leerink named it the top beneficiary of Anthropic’s biology push. On Sept. 24, Leerink raised its price target from $120 to $160, and the stock rose 16% to $184. In the rush into AI biology, Twist is the one selling the shovels.
Claude designs, Twist builds
Anthropic is funding lab tests of 5,000 AI-designed proteins, each one built from Twist’s synthetic DNA. The work runs as a loop. Claude writes amino acid sequences as text. Twist turns each design into synthetic DNA. The proteins are expressed and screened for binding. Then Claude reads the lab data, which it did in under 23 minutes when results came back. Each loop takes days, not months.
Design and analysis now run at computer speed, but the physical build in between is a step every lab still has to pay for.
Code can’t bind, DNA can
Claude’s output is just text, a string of amino acid letters. To learn whether a design works, someone has to write it into physical DNA, express the protein, and measure binding. As AI makes designs cheap, the bottleneck moves to the lab, and Twist is one of the few suppliers built for that volume.
The designs are also landing more often. In its first protein design run in August, Claude hit 14 of 15 targets, with the designs built and tested by Twist and Adaptyv. Of 1,320 Claude designs built and tested by independent labs, 354 bound. Claude’s designs showed a binding hit rate of 22% to 35%, against 10% to 15% typical in the field. More hits per design means more campaigns worth running, and every one starts with a DNA order. Twist CEO Emily Leproust summed up what AI is doing to demand: “It just creates more sequences.”
Two giants, one supplier
On Sept. 16, Twist joined Lilly TuneLab as its preferred antibody-testing partner. There, 75+ companies can order antibody testing on Twist’s preferred protocols and pricing, and the results feed back into Lilly’s shared AI models. Anthropic’s 5,000-design program runs on Twist DNA too. Being the default inside both workflows means recurring orders instead of one-off pilots.
The business was already growing. On Aug. 3, Twist reported third-quarter revenue of $118.4M, up 23% year over year and its 14th straight quarter of growth.
Wall Street can’t keep up
Twist closed at $184, while the average analyst target sits near $110, about 40% below the stock. Even Leerink’s raised target trails it by 13%. The gap keeps widening. In March, the average target sat 6% above the stock. By June it was 28% below. Since then, the price has run and analysts have been chasing it.
The bull case is that every AI design still needs real DNA, the fiscal 2026 revenue guide is $456-457M, and AI drug-discovery orders have more than doubled again. The bear case is that the company is still not profitable, the Anthropic work is already in its guidance, and cloud labs like Adaptyv could push testing prices down. The stock may be priced for a future the business hasn’t reached yet.