Just nine days after dosing the first patient in a Phase 3 trial, Generate:Biomedicines has filed for an IPO.
The Flagship Pioneering-backed startup has become a richly-funded leader of AI bio since its founding in 2018. The company has brought in $934 million, signed pharma partnerships with Amgen and Novartis, and is the first AI-focused biotech to enter Phase 3 with its lead drug program.
Generate’s IPO could rank as one of biotech’s largest in recent years. It comes during a reopening of IPO markets, which for years had remained stubbornly shut during biotech’s market downturn. But in recent weeks, a handful of biotechs have
filed
and
gone
public
, in many cases seeing significant investor demand.
Generate needs the cash a public offering would bring. In its
200-plus page SEC filing
that became public Wednesday, it said it spent $225 million on R&D in 2025, ending the year with $221.5 million in cash and equivalents — giving it about a year of operating runway at its current burn rate. The company said it expects its R&D expenses to “increase substantially for the foreseeable future” as more programs enter clinical trials.
The company plans to list on the Nasdaq under the ticker
$GENB
. It didn’t say how much it planned to raise or at what valuation, amounts typically disclosed later in the IPO process.
Much of Generate’s existing cash — and possible IPO proceeds — will likely go to its lead program, a long-acting antibody called GB-0895. The drug targets an immune protein called TSLP and is now in late-stage studies for severe asthma. The company also expects to use the capital to advance two cancer drug programs into Phase 1 studies this year.
That’s a stark contrast with
a couple of years ago
, when CEO Mike Nally described a pipeline of 17 programs with plans to add 10 more annually. If that pace were kept, Generate would likely have over 40 programs today, but the company lists only those three drugs.
A key unknown is whether Generate’s TSLP program will be able to differentiate from Amgen’s blockbuster TSLP-targeting antibody, Tezspire. Generate’s antibody is effectively a tweaked version of that molecule that binds 20 times more strongly to its target and is designed to be dosed every six months rather than monthly.
Generate’s promise is to use generative AI to design brand new proteins from scratch — a technique called
de novo
protein design. It was based on a diffusion model, which powers many AI image generators.
The company
made waves in late 2023 by publishing in
Nature
its diffusion model for protein design called Chroma. A competing model called
RFdiffusion
, published the same year by David Baker’s lab at the University of Washington, became the basis for startup Xaira Therapeutics, which
launched with $1 billion
the following year.
Since then, the AI protein design field has become increasingly crowded, with startups like Chai Discovery and Nabla Bio building their own models and vying for state-of-the-art capabilities.
But Generate’s most advanced programs, such as the TSLP antibody, focus on optimizing existing proteins rather than AI designing new ones from scratch. It was not immediately clear if the company’s two cancer programs were designed
de novo
or merely optimized with AI.
One of those programs is an antibody designed to help clean up a potent chemotherapy called MMAE, which is released from antibody-drug conjugates like Pfizer’s Padcev. The second program is a CAR-T cell therapy,
developed with
Roswell Park Comprehensive Cancer Center in New York, that targets the protein MUC16 in ovarian cancer.
Even though Generate has framed itself as a leading AI drug developer, the company is not all computers. It has invested heavily in two labs, including one 25 miles north of its Somerville, MA, headquarters that houses
four expensive and sensitive microscopes
that the company feared would be jostled by the building’s elevators and Boston’s light rail.
Those cryogenic electron microscopes map the structures of Generate’s computer-generated proteins in atomic resolution, providing key data to help validate its AI designs and improve its models. They’re a key part of the company’s “design-build-test-learn” cycle that it believes will “shift risk to the preclinical setting.”
The IPO will be particularly important for Generate’s founder and key backer Flagship. The Cambridge-based venture incubator behind Moderna and its entities own 56.6% of the company, according to the filing.
That appears somewhat unusual, even by Flagship’s standards. For Moderna’s IPO,
Flagship entities owned 19.5%
of the company. AstraZeneca and Viking Global were also listed with stakes above 5%.
It’s been four and a half years since the last Flagship-incubated biotech attempted an IPO. Omega Therapeutics debuted in a
$126 million listing
in July 2021. The epigenomics biotech
went bankrupt
last year. In 2021, another AI-focused Flagship biotech tried to get onto the public markets, but Valo Health axed its deal to combine with a special-purpose acquisition company that November.
Kyle LaHucik contributed reporting.