One To Watch
Cradle
Company Overview
A Dutch-Swiss generative-AI software company whose protein-engineering platform is being embedded into early drug discovery workflows at six of the world's top 25 pharma companies, including Lundbeck, Bayer, and Novo Nordisk. Cradle is not a drug developer — it sells AI SaaS infrastructure that accelerates the design and optimization of biologics before any molecule enters the clinic. Its platform covers antibodies, enzymes, and bio-based materials across medicines, food ingredients, detergents, and bioplastics. The thesis is simple but commercially potent: that AI language models trained on amino-acid sequences can compress the most expensive part of biologics R&D — the iterative engineering of protein properties — from years to months.
Headquarters and Global Presence
Cradle is headquartered in Amsterdam, the Netherlands, with a second technical office in Zurich, Switzerland. It operates its own wet lab in Amsterdam, used to generate experimental training data and validate platform outputs — a deliberate choice that makes the feedback loop proprietary rather than purely computational.
Founding and History
Cradle was founded in 2021 by Stef van Grieken, Jelle Prins, Elise de Reus, Eli Bixby, and Harmen van Rossum; van Grieken and Prins both came from Google, including stints at Google AI and Google X. The company raised a EUR 5.5 million seed round from Index Ventures in November 2022, a $24 million Series A led by Index Ventures in November 2023, and a $73 million Series B led by IVP in November 2024, bringing total funding past $100 million inside three years of operation. That fundraising pace reflects genuine customer traction rather than speculative positioning.
Therapy Areas and Focus
On the pharma side, Cradle's most prominent programs sit in CNS and cardiometabolic disease, reflecting the biases of its largest partners. Lundbeck is using the platform specifically for brain disorder biotherapeutics, while Novo Nordisk has applied it to next-generation GLP-1 and cardiometabolic candidates — potentially the successors to Ozempic and Wegovy. The platform is agnostic to indication, however; what it optimizes are protein properties rather than disease targets, which means any biologics program with an antibody or enzyme at its center is addressable.
Technology Platforms and Modalities
Cradle's core platform applies large language model architecture to protein sequences, treating amino acids as a vocabulary and learning the rules of protein function from vast training datasets. Scientists input desired property improvements — higher binding affinity, better thermal stability, reduced immunogenicity, easier manufacturability — and the platform proposes sequence modifications ranked by predicted efficacy. The critical differentiator is the lab-in-the-loop model: experimental results from physical testing feed back into the model, making each iteration smarter than the last. Cradle's Amsterdam wet lab exists precisely to generate high-quality proprietary data for that retraining cycle.
Key Pipeline and Programs
Cradle has no proprietary drug pipeline; its value is the infrastructure running inside partners' programs. The Lundbeck collaboration, announced 3 June 2026, is the most strategically telling: it covers two antibody programs for CNS diseases, with one already in active optimization at announcement. The platform is optimizing affinity, stability, immunogenicity, and manufacturability simultaneously — the multi-parameter challenge that makes antibody engineering slow and expensive by conventional methods. The Bayer three-year collaboration, signed January 2026, gives Bayer's antibody scientists full platform access and includes a joint machine-learning research project; Bayer retains all resulting IP, a model that is likely to become Cradle's commercial template. Novo Nordisk's engagement targets next-generation GLP-1 variants, a commercially critical program given the scale of Ozempic and Wegovy — even marginal improvements in half-life or manufacturability at that volume represent enormous value. Across all customers, Cradle reports more than 50 active projects running simultaneously.
Recent Developments
The Lundbeck partnership announced on 3 June 2026 is the most recent and most strategically significant milestone, establishing Cradle's first foothold in CNS biotherapeutics and marking Lundbeck's first end-to-end AI-guided protein engineering workflow. Before that, the Bayer three-year collaboration launched in January 2026 extended Cradle's reach deeper into large-cap pharma antibody discovery. The $73 million Series B, closed November 2024 and led by IVP, funded both platform development and wet-lab expansion. Customer metrics published alongside recent deals — up to 12x acceleration in early R&D cycles, up to 90% cost reduction — are self-reported but consistent with the kind of claims that sustain enterprise SaaS renewals.
Key Personnel
Stef van Grieken serves as CEO and co-founder, having previously worked at Google AI and Google X before founding Cradle in 2021. Co-founder Jelle Prins shares a Google background and brings applied machine-learning expertise to the platform's architectural foundations. Co-founders Elise de Reus, Eli Bixby, and Harmen van Rossum round out the founding team, combining scientific and engineering depth that distinguishes Cradle from pure-software AI plays with no wet-lab grounding.
Strategic Partnerships
Cradle's commercial model is partnerships with pharma and biotech customers who pay for platform access and run their own experimental validation. Named partners include Johnson & Johnson, AbbVie, Novo Nordisk, Bayer (three-year collaboration from January 2026), Lundbeck (June 2026), and Ginkgo Bioworks. The Bayer deal structure — full platform access, joint ML research, customer retains IP — is the clearest public template for how Cradle commercializes without co-owning drug assets. Six of the top 25 global pharma companies are now customers, which at this stage of the company's life is a more meaningful signal than any single deal value.
FAQ Section
The bottleneck is not compute but data — specifically high-quality, experimentally validated protein sequence-function pairs, which take years and significant wet-lab investment to generate. Cradle's Amsterdam wet lab and lab-in-the-loop retraining cycle produce proprietary training data that a pharma company starting from scratch would struggle to replicate quickly. Bayer's decision to run a joint ML research project alongside full platform access, rather than simply licensing a static tool, suggests the ongoing data flywheel is the real asset.
Protein sequences, like natural language, have deep statistical regularities: certain amino-acid patterns confer stability, others drive binding. Large language models trained on millions of known protein sequences can learn these rules implicitly and propose variants that preserve function while improving target properties. The limitation is that the model predicts, but does not guarantee — experimental validation remains essential, which is why Cradle's lab-in-the-loop architecture matters more than the model architecture alone. Programs in highly novel sequence space, far from training data, remain harder to navigate.
Cradle is a pure-software infrastructure play — it does not own pipeline assets, take milestone payments on drug approvals, or share in downstream royalties. Recursion and Insilico both have proprietary drug pipelines and take discovery risk; Cradle sells SaaS access and lets partners retain all IP, as the Bayer deal explicitly confirms. That model means lower upside per program but faster, lower-risk revenue scaling — and no exposure to clinical trial failure. It is a closer analog to Dotmatics or Schrödinger than to Recursion.
Under the June 2026 partnership, Cradle's platform is being used to optimize two Lundbeck antibody programs targeting brain disorders, with the focus on improving affinity, stability, immunogenicity, and manufacturability simultaneously. One program was already in active optimization at the time of announcement; neither program has been named publicly, and both are at pre-clinical or early discovery stage. The deal represents Lundbeck's first fully AI-guided protein engineering workflow, meaning Cradle is embedded from the design stage rather than brought in for remediation.
The pharma-facing programs span CNS (Lundbeck antibodies), cardiometabolic and GLP-1 biology (Novo Nordisk), and general antibody discovery and optimization (Bayer, J&J, AbbVie). Beyond pharma, the platform is applied to enzyme engineering for food ingredients, detergents, and bioplastics — a breadth that reflects the platform's modality-agnostic design. The common thread is any application where a protein's functional properties need to be improved iteratively, regardless of end market.
Cradle is a commercial-stage software company: it has paying enterprise customers, 50-plus active projects running in parallel, and over $100 million raised across seed, Series A, and Series B rounds closed between 2022 and 2024. The Bayer and Lundbeck deals in 2026 extend a run of major partnership announcements and confirm that the platform has moved past proof-of-concept. The next inflection point is whether any partner program progresses to clinical development, which would validate the platform's real-world impact on drug timelines rather than just internal R&D metrics.
Cradle's near-term story is one of enterprise adoption widening and deepening — but several specific watchpoints will determine whether the platform translates from R&D acceleration tool to durable pharma infrastructure.
Latest Cradle News
Today's issue
| Headless Content Management with Blaze