redalpine Closes $250M RAC VIII Fund for European AI-Science Startups

redalpine has closed its latest early-stage venture fund, redalpine capital VIII (RAC VIII), at $250 million, and the oversubscribed vehicle will back roughly 20 companies across Europe that combine software with scientific advances. The Zurich-based firm, marking 20 years in business, now manages more than $1.4 billion across its platform.

RAC VIII will primarily lead Seed and Series A rounds, writing initial checks of $1 million to $15 million. The first three investments from the fund — Enact Intelligence, Helical and Arlequin AI — map where redalpine sees the next phase of AI value: robots that feel physical forces, virtual laboratories for drug discovery, and systems that uncover relationships in complex datasets.

That thesis matters for Europe's AI ecosystem. While much of the market's attention sits with large language models, redalpine is placing concentrated bets on AI that moves into scientific and industrial workflows — a segment where technical depth, not just software distribution, decides which companies win.

What does the new fund actually cover?

The $250 million figure represents RAC VIII's capital commitments. The $1.4 billion-plus covers redalpine's assets under management across its broader investment platform, which is designed to support companies from Seed through later stages. The two numbers describe different things, and the firm draws that line explicitly.

redalpine says it reviews more than 5,000 companies every year. A planned portfolio of around 20 businesses leaves room for concentrated positions, and the $1 million to $15 million range describes initial checks rather than each company's eventual total financing.

Founded in Zurich in 2006, redalpine's portfolio includes Proxima Fusion, Mistral AI, N26, Basecamp Research and microagi. Half of its investment team consists of trained scientists — biologists, physicists and engineers. That mix matters when a startup's central challenge involves robotic control, biological measurement or a new computing architecture as much as it involves software.

"Twenty years ago, we started redalpine with the conviction that the most important companies would emerge where software and science meet," said Dr. Michael Sidler, founding partner at redalpine. "That belief hasn't changed. With RAC VIII, we're excited to keep backing these founders early, before the opportunity is obvious."

In short: the fund is a concentrated early-stage vehicle run by a team built to evaluate science-heavy deals.

Why teach robots to feel force?

One of RAC VIII's first investments, Enact Intelligence, is developing force intelligence for robots. The technical premise is straightforward: seeing an object and understanding an instruction provide only part of what a machine needs to manipulate the physical world.

In its October portfolio update, redalpine describes Enact as feeding torque and touch information into a foundation model alongside vision and language. Those signals give a robot information about contact and physical interaction. For dexterous tasks, that could let a system adapt its actions as resistance or contact changes, rather than relying solely on a visual estimate.

The investment illustrates a broader challenge in physical AI. Useful behavior depends on connecting perception to sensing and control, and Enact's approach seeks to make force awareness part of the learning process itself.

The takeaway: robots that can sense what they touch, not just what they see, are the bet here.

What is a virtual biology lab?

A second RAC VIII investment, Helical, is building a virtual AI laboratory for drug discovery. Its platform applies biological foundation models to research workflows, including target discovery, cell-state analysis and patient stratification.

Helical's Model Factory adapts and benchmarks models for particular biological questions, while its virtual lab supports computational experiments. The company also maintains an open-source framework for working with biological foundation models, including genomics and single-cell data.

The practical opportunity is making model-based investigation repeatable and accessible inside a research workflow. There are limits worth noting: computational findings still need experimental validation, and a virtual lab helps researchers prioritize and investigate hypotheses rather than establishing a treatment's effectiveness on its own.

In plain terms, Helical wants to compress the early, exploratory phase of drug research into software.

How do you make data analysis traceable?

The third named investment, Arlequin AI, focuses on extracting patterns and relationships from large, fragmented datasets. Its HuDex platform presents applications ranging from financial-transaction analysis and due diligence to communications forensics and supply-chain investigation.

Arlequin describes its approach in terms of unsupervised and topological AI, emphasizing relationships across data and the ability to trace findings back to supporting records. In high-stakes analysis, that traceability matters: users need to inspect the evidence behind a connection before acting on it.

The company's stated approach emphasizes understanding patterns in data, which extends the fund's AI exposure beyond generating text. redalpine is effectively hedging across different AI capabilities — sensing, biological modeling, relational analysis — rather than concentrating on any single architecture.

The takeaway: analysis you can audit is the selling point, not just analysis that produces answers.

Who is backing the fund?

RAC VIII's limited partners include Aviva Investors, Flexstone Partners and KfW, alongside entrepreneurs, executives and business owners. That mix gives the fund institutional capital plus investors with direct operating experience — useful for a portfolio that will need technical and commercial guidance in equal measure.

redalpine reports 12 portfolio-company exits since 2023. It also says companies backed by RAC VII, its previous early-stage fund that closed in 2024, have subsequently raised more than $1.1 billion in aggregate financing. Both figures describe portfolio activity and follow-on fundraising; they do not establish the new fund's investment returns.

Still, the track record is the context institutional investors looked at when committing to RAC VIII, and the oversubscription suggests demand exceeded the firm's original target.

What does this mean for European AI?

For redalpine, the new vehicle extends a strategy that connects technical expertise with early capital and support as companies grow. For Europe's AI ecosystem, the initial portfolio offers a concrete view of that strategy in action: investing in the sensing, biological models and analytical architectures that turn general advances in AI into capabilities for specific industries.

The fund's first three bets suggest the next competitive layer in AI may sit below the headline models — in the hardware-adjacent sensing, the domain-specific biological models and the traceable analysis tools that make AI usable in regulated, high-stakes fields.