Kerna Labs is building foundation models of RNA to design therapeutically optimized mRNA sequences — longer half-life, higher protein output, better tissue targeting — extending mRNA beyond vaccines into protein-replacement and other genetic-medicine categories. The wedge is a two-step training model (the founders call it “Taika”) trained on public datasets like GTEx and gnomAD, coupled with an in-house data flywheel: sequences the platform designs get validated, feeding a proprietary dataset that competitors cannot buy. Long-term moat is the dataset plus patentable UTRs (Wilson Sonsini engaged on IP).
Validation is credentialing-heavy at this stage. CEO Julia Peng emailed on 2024-07-26 that Gradient Ventures (Google’s AI fund) and Susa’s deeptech vehicle Humba had put in a combined $2.5M as the anchor tranche; the 2024-07-29 deal call confirmed the round was $5M target with $2.5M signed and wired on a $22.5M post SAFE and $1.5M more offered. That cap re-priced upward as demand grew: the round closed at $6.1M with StoryHouse’s $150K coming in on the final $30M-post tranche alongside DVC and Pioneer Fund. In January 2025 the company launched publicly with Melissa J. Moore (ex-CSO Moderna) named as co-founder, adding a top-tier scientific credential to Amit Deshwar (ex-SVP Platform, Deep Genomics), Peng (ex-Confluent GTM), and founding scientist Michael Swift (Stanford systems-biology PhD). Seed is closed and the position is held in StoryHouse Fund I.
Two independent analyst reports place the mRNA therapeutics market between roughly $17B in 2025 and $59–83B by 2034–35, growing at a 14–17% CAGR WebWeb. The interesting slice for Kerna is the “beyond vaccines” expansion — protein replacement, oncology neoantigens, rare genetic disease — where the incumbent constraint is exactly what Kerna targets: half-life, protein output, tissue specificity, and manufacturability Internal. Michael Swift framed the “why now” on 2024-04-18 as Moderna having proved out mRNA as a modality while sequence-level optimization remains the open frontier.
| Player | Positioning | Funding / Stage | Edge vs. them |
|---|---|---|---|
| Kerna Labs | Foundation models of RNA for therapeutic sequence design (half-life, protein output, tissue targeting) | Seed · $6.1M · $30M cap | — |
| Deep Genomics | AI-driven RNA therapeutics platform focused on target ID and candidate discovery; Toronto-based, founded 2014 Web | Established private, multiple rounds | Kerna’s Amit Deshwar was SVP Platform at Deep Genomics — direct playbook knowledge; Kerna focuses on sequence-level optimization rather than target ID |
| Moderna | Incumbent mRNA developer; internal design work at scale | Public | Melissa Moore ran the platform team at Moderna; Kerna designs mRNA sequences with claimed 5–10x half-life extension Internal |
| Generate Biomedicines / Xaira / EvolutionaryScale | AI-for-biologics platforms (proteins, antibodies, some RNA) Web | Well-funded (hundreds of millions) | Adjacent modalities; Kerna is RNA-native rather than protein-first |
| In-house pharma teams | Every major mRNA player runs internal sequence design | Global pharma | Kerna’s CDMO / partnership model positions it as an enabling layer rather than a competitor |
Moat: the compounding data flywheel — every sequence designed and validated (in-house or via CDMO partnerships) enriches a proprietary training corpus that outside teams cannot replicate Internal — layered on patentable UTRs from the first program (Wilson Sonsini engaged) Internal.
Most likely path is a platform acquisition by a mRNA incumbent (Moderna, BioNTech, Pfizer, GSK) or a gene-medicine specialist (Alnylam, Intellia) looking to buy sequence-design capability rather than build it. Biopharma M&A rebounded in 2025 — cumulative YTD deal value hit $49B, already surpassing all of 2024, including multi-billion-dollar biotech acquisitions Web. On Alnylam-style platform comps, a successful Kerna outcome would produce meaningful return on a $30M-post entry; realistic downside is a talent-and-tech acqui-hire in the low nine figures if the pipeline fails to advance.
Full deal call with Julia Peng. The $5M seed was structured with Gradient Ventures and Susa Ventures leading; $2.5M signed and wired on a $22.5M post SAFE, with $1.5M more offered by other funds. Note reveals Melissa Moore (ex-CSO Moderna, Emerita) has joined as co-founder. Miles-approved next step recorded as pursuing an offer for $100K (final check ended at $150K per Airtable).
DEAL STATUS: In Conversation. NEXT STEPS: Discuss with Miles but likely pursue and give offer for $100K.
One-Liner: better mRNAs built with AI foundation models for therapeutics.
Company & Insight — Blurb: Kerna aims to develop better genetic medicines using AI to unlock full potential of mRNA as a toolkit for genetic medicine. With tailored sequences, they have half life, protein efficiency, and better tissue targeting, unlocking more diseases to cure with mRNA. Problem: mRNA gets bottle-necked by limited half-life, tissue specificity, lack of stability, manufacturability. It’s unstable, protein output is insufficient for therapeutics, and some sequences are not able to be manufactured — these are best addressed with sequence design. Insight / Secret Sauce: they can tailor-make these mRNA sequences with machine learning — usually 5–10x half life to un-bottleneck the market of protein-deficiency diseases, allowing them to tackle more diseases.
Their Model: Input datasets like GTEx, gnomAD, etc — 1B+ Mamba SSM Model — 10M probe networks — output is mRNA with better half-life and translatability. They design mRNA with sufficient protein output and duration to be therapeutically effective.
Business Model: External partnerships — CDMOs, gene and epigenetic editing — will help them build up their dataset. In-house pipeline: protein-replacement therapies. Long-term moat will be the dataset they build.
What’s the moat? Capital alone doesn’t solve the problem. Need deep expertise in machine learning for mRNA. Amit spent a decade building machine learning models, and Melissa (who recently joined as co-founder, and who led development of the platform behind the COVID-19 vaccine — Emerita @ Moderna) brings the bio side. They can patent UTRs. The moat can be their dataset down the road — loop of generating great sequences, validating them, getting more data, etc. Alnylam playbook but for mRNA and with much larger options. Working to get IP for their first one (WSGR).
Market & Competitors — Why Now: there is an explosion in R&D dollars going towards mRNA as a modality given potential: medicine companies (Moderna), gene editing (Intellia, Mammoth, Editas), in vivo CAR-T, etc.
Round Dynamics & Financials: Raising a $5M Seed. Gradient Ventures and Susa Ventures are leading. $2.5M was signed and wired on $22.5M Post SAFE. $1.5M has been offered by other funds.
Follow-up email from Julia Peng after the 2024-07-25 call. Shared the deck (docsend link) and confirmed the anchor tranche: Gradient Ventures (Google’s AI fund) and Susa’s deeptech fund Humba put in a combined $2.5M; they were raising the second half of the round. Proposed a follow-up meeting for the coming Monday.
Email from Julia Peng. “Hi Josh, it was great to meet today! Here’s the link to our deck. How is Monday 10:30am or 1pm PT?”
“Our current round dynamics: Gradient Ventures (Google AI fund) and Susa Ventures (specifically their deeptech fund Humba) put in a combined $2.5M and we are raising the second half of our round.”
Signed: Julia, Co-Founder, Kerna Labs.
Second call with the Kerna team (four months after the initial April call with Michael). Positioning was “computational pharma, foundation models for biology.” Round was $5M with half filled and closing in the next week or two; two anchors had put in $1M+ each (identified in the follow-up email as Gradient and Humba). Team backgrounds recapped in detail. Note author flagged this as over-his-head technically and asked to loop in Matthew and possibly Jane / Shenda.
Intro and intel source: LinkedIn Sales Nav and cold outreach.
One-liner: Computational Pharma. Foundational models for biology. They are developing machine-learning models to create better genetic medicines. Use foundation models of RNA to find and optimize sequences with increased protein output and prevent off-target expression. Second call with them; first was four months back just with Michael after they got into the incubator. Raising a $5M round with half filled and closing in the next week or two.
Team & Founders. Michael Swift: CMC Biochemistry 2016 grad, PhD systems biology Stanford 2023; Technical Consultant for Longitude Capital for four years; bioinformatics data scientist last year at Amplicon Biocomputing. Quantitative biologist focused on modeling the immune system, single-cell genomics and DNA sequencing on B cells. Founding scientist. Amit Deshwar: Google platforms engineer, nine years at Deep Genomics as Head of Platform and VP of Predictive Systems, scientist and engineer. Julia Peng: Wharton Computational Neuroscience. Founder of Spectral Materials (materials for chemical manufacturing). Three years at Confluent doing GTM and product. Her focus is GTM. Julia and Amit are engaged and met at Canaan; met Michael in a biology community. Amit left Deep Genomics to start this last year.
Company & Insight. Combining Moderna + AI — reprogramming cells with mRNA as a platform. They’ve figured out how to design mRNA sequences with better half-lives to try to unlock more cures. Part of accelerator HF0. Problem: bottlenecks are half life / unstable and protein output is inefficient. Insufficient protein output is the main bottleneck. Insight: optimizing for ease of manufacturing, half-life, etc. Two-step training model — Taika Foundation Model. Data moat will come from mRNA they design using the model and updating the model in an active-learning loop.
GTM. In-house pipeline: try to have three mRNAs for genetic disease by end of the year. External partnerships: CDMOs to continue getting unique dataset (trojan-horse for data), plus gene and cell therapy and antibodies. Setting up a validation study with a biopharma company — giving them two disease targets. Long-term focus will all be in-house; external partnerships are just for data play.
Round Dynamics & Financials: Raising $5M Seed. Half wired by two anchors with $1M+ checks each; they didn’t share the names. Described as a deeptech fund and an AI fund.
They asked for introductions to in-vivo CAR-T / large pharma / drug-discovery companies.
First call with founding scientist Michael Swift after cold outreach through LinkedIn Sales Nav. Kerna was two weeks into the HF0 accelerator (repeat-founder live-in program). At the time they had raised about $1M in SAFEs — a $500K uncapped SAFE from HF0 plus $500K from angels — and said they might take more in the initial SAFEs. Michael flagged that he is the founding scientist, not the fundraising lead, and that Kerna had a co-founder (Amit) who had been Head of Platform at Deep Genomics.
Intro and intel source: LinkedIn Sales Nav and cold outreach. Deal status: In Conversation. Next steps: try to get in touch with other founders and see a deck. They may take more in these initial SAFEs.
Team & Founders. Michael Swift: CMC Biochemistry 2016 grad; PhD in systems biology at Stanford, 2023; Technical Consultant for Longitude Capital for last four years; bioinformatics data scientist at Amplicon Biocomputing. Quantitative biologist and PhD focused on modeling the immune system. Did bench science at a biotech company between CMC and Stanford. Focused on single-cell genomics and DNA sequencing to look at B cells in human immune systems. Most studies that look at B cells draw from blood cells, but the most impactful signal is in bone marrow. Michael is not the fundraising lead. After grad wanted to be first hire or founder at a biotech company. Founding scientist at Kerna. One of the founders was Head of Platform at Deep Genomics (Amit).
Company & Insight. Blurb: Moderna + AI — huge space in reprogramming cells and mRNA as a platform. In practice these mRNAs are immunogenic and don’t last long, but using new generative models they can learn non-obvious things about mRNA sequence design. At MBC Bio Labs using lab space; part of accelerator HF0. They have a model to design mRNA sequence designs with much better half-lives to unlock cures for more diseases.
Insight: right now Moderna doses people once a week because half-life is roughly 10 hours; if half-life increases, they could do more.
HF0 Accelerator: exclusively for repeat founders, largely focused on AI in science, robotics, and gaming. Fractional executive coaches. Everybody lives in the house for three months. Just started two weeks ago.
Round Dynamics & Financials: raised about $1M in SAFEs. HF0 SAFE was $500K uncapped; and another $500K from HNW angels. Open to take on more in the SAFEs.