☰

CanChip

From prototype chips to clinical screening data

Industry
Biotechnology
Headquarters
Potsdam, Germany
Public information as of
January 2026

A4BEE prepared this analysis from publicly available sources. It reflects our own reading of CanChip's published strategy and is not endorsed by, or produced in cooperation with, CanChip. Company website

Strategic priorities

CanChip is a Potsdam-based startup built around tumor-on-a-chip technology for oncology drug screening. Founded in 2023 with the laboratory opening at Potsdam Science Park in June 2024, the company is working to translate microfluidic prototypes into platforms that pharmaceutical partners can run in regulated drug-development workflows. The CEO and Scientific Director lead a small team working on tracheal tumor chips and a CRC-on-a-Chip (colorectal cancer on-a-chip) model that has shown dose-dependent responses to 5-Fluorouracil (5-FU) consistent with clinical data.

The near-term work is industrialisation: moving from lab-built single chips to repeatable, automated production of chips that can serve global pharmaceutical screening programmes. That move has a regulatory face — first-in-human studies and the EMA (European Medicines Agency) and FDA (US Food and Drug Administration) submissions that follow — and a data face, since the same screening run has to be traceable from the instrument that captured the image back to the chip batch that produced it.

Recognition has followed the technical work: CanChip won the KfW Award "Gründen 2025" for social and scientific value and was nominated for the German Startup Awards 2025 "Newcomer of the Year" category. The company operates in a German biotech investment climate in which VC funding is around 0.02 percent of GDP, against roughly 0.07 percent in the United States, which makes each infrastructure decision carry weight relative to the runway it consumes.

Challenges we see

  • R&D Operations Manufacturing

    Moving microfluidic assembly from bench to repeatable production

    CanChip is moving from lab-scale prototypes such as the tracheal tumor chip to higher-volume production of chips for pharmaceutical screening. The laboratory at Potsdam Science Park opened in June 2024 and the team is now working through the steps that turn a one-off microfluidic build into a repeatable batch.

    Where microfluidic assembly has so far been configured for individual chips, scaling up for screening programmes puts process repeatability and per-chip traceability onto the critical path rather than at the end of it.

  • Information Technology Digital

    Bringing screening data out of paper and isolated spreadsheets

    High-throughput screening data is recorded on paper or held in isolated spreadsheet files, creating fragmented information streams across experiments.

    Where each instrument stores its own results in its own file, joining the readings from a screening run to the chip batch that produced them becomes a manual task, and the audit trail it generates lives outside the systems the laboratory runs.

  • Market Position Operations

    Competing against better-funded organ-on-a-chip programmes

    CanChip operates in a German market where biotech VC funding is around 0.02 percent of GDP, against roughly 0.07 percent in the United States. Direct competitors include Emulate, with around $225 million raised, and CN Bio, with around $44.8 million.

    Where each tooling decision competes for a limited runway, infrastructure choices need to be selective and well-targeted: the projects that compound over multiple experiments and the platforms that scale without duplicating effort tend to set the pace.

  • Regulatory Compliance

    Aligning research-grade microfluidic work with GMP

    CanChip is moving toward clinical applications and first-in-human studies, which means the microfluidic workflow has to satisfy Good Manufacturing Practice (GMP) standards and the data integrity expectations of EMA and FDA submissions. ALCOA+ (the data integrity principles of Attributable, Legible, Contemporaneous, Original and Accurate, plus Complete, Consistent, Enduring and Available) and ALCOA+ principles apply to any data that ends up in a clinical submission.

    Where data integrity is assembled at the end of a study, satisfying ALCOA+ becomes a documentation project. Where it is captured at the instrument and carried through the pipeline, the same property comes for free as part of the screening workflow.

  • Human Capital Operations

    Closing the digital fluency gap between research and industrial systems

    The team has strong technical expertise in microfluidics, cell biology and 3D bioprinting, but limited experience with industrial digital systems and automated workflows that the move to higher-volume screening assumes.

    Where the same operators move between manual microfluidic work and instrument-driven production, the practical question is whether each new system teaches once or whether every shift relearns the interface.

Opportunities, by urgency and business impact

Each bubble is one opportunity, numbered to match the list below. Further right means it bites sooner; higher means a bigger effect on the business. A bigger bubble means a bigger implementation effort.

Source: A4BEE analysis of public sources

  1. Building human-relevant models that cut IND failure rates

    Traditional oncology drug development relies on 2D cell cultures or animal models that miss the fluid dynamics and 3D architecture of human tumors, and Investigational New Drug (IND) programmes fail at a rate that traces back to that gap.

    Tumor-on-a-chip models that combine 3D cell culture with endothelial co-culture such as HUVECs (human umbilical vein endothelial cells) reproduce angiogenesis and drug absorption at human-relevant scale, and CanChip's CRC-on-a-Chip has shown 5-FU responses consistent with clinical data. Wider deployment of these models in partner screening programmes can shorten oncology R&D timelines by up to 2.5 years.

    • CanChip, About Us, accessed 23 January 2026
    • CanChip, News, accessed 23 January 2026

  2. Standardising cell detection and colony state classification

    Visual colony counting and manual pipetting are slow, vary between operators, and live in notebook pages rather than data systems.

    Camera-based machine-learning models for cell detection and colony state classification turn visual QC into a documented, repeatable step, with each reading carrying the image, the model version and the timestamp rather than a notebook entry.

    • A4BEE strategic analysis, CanChip internal assessments
    • A4BEE strategic analysis, CanChip R&D digitalisation overview

  3. Connecting legacy laboratory instruments to a shared data layer

    Laboratory data stays inside local controllers (RS-232, Profibus) or vendor-specific instrument software, so a screening run is split across the same number of files as instruments on the bench.

    Connecting instruments through documented protocols such as OPC UA (Open Platform Communications Unified Architecture) or MQTT (a lightweight messaging protocol widely used in industrial IoT) and bringing their readings into a shared time-series store gives the laboratory one record of a screening run, with each value traceable to the instrument and method that produced it.

    • A4BEE strategic analysis, CanChip IT/OT environment
    • A4BEE strategic analysis, Polfarmex reference architecture

  4. Capturing screening results as ALCOA+ data from the start

    Paper logbooks and manual transcription sit between the instrument and any record that ends up in a clinical submission. Each transfer is a place where the link between the value, the chip and the operator can be lost.

    Capturing results at the instrument and carrying them through an electronic pipeline with attribution, versioning and audit-trail handling built in keeps the data ALCOA+-compatible from the moment it is recorded, so the screening workflow and the eventual submission pull from the same evidence chain.

    • A4BEE strategic analysis, CanChip regulatory environment
    • A4BEE strategic analysis, Polbionica ALCOA+ reference

  5. Simulating microfluidic conditions before committing to physical batches

    Scaling microfluidic production by trial and error wastes material, consumes chips and pushes delivery dates for partner screening programmes outward.

    Computational Fluid Dynamics (CFD) models of the perfusion loops in vascularised chips predict fluid shear stress and nutrient gradients before a physical batch is built, so the design variables that drive cell viability can be tightened in software rather than rediscovered at the bench.

    • A4BEE strategic analysis, CanChip scale-up context
    • A4BEE strategic analysis, Polbionica digital twin reference

What we'd propose

  • Enterprise AI

    GxP-ready data foundation for tumor-on-a-chip screening

    An ontology-based data platform that joins screening results, chip batch records and instrument metadata in one queryable model, with audit trails and version handling that satisfy ALCOA+ from the first reading rather than at submission time.

    • Shared screening ontology

      One agreed set of terms

      Define assay, chip, batch, specimen, reading and instrument as explicit entities with agreed relationships, so a screening run can be queried once instead of reconciled from instrument-specific exports.

    • Pipelines from instruments and laboratory systems

      Loading each side the same way

      Build ingestion for microscope image metadata, plate reader output and laboratory notebooks against the same model, with schema validation at the boundary so a malformed record fails loudly instead of silently corrupting the run.

    • Audit trail and version handling

      Evidence that holds up under review

      Implement attribution, electronic signatures and immutable history to ALCOA+, so the data feeding a clinical submission is the data that ran the screening rather than a reconstruction of it.

    • The same record serves research, partner reporting and any eventual regulatory submission.
    • Screening results arrive with their lineage intact, so root-cause analysis starts from the data rather than from the notebook.
    • New assays and partner-specific data fields attach to the model rather than triggering another migration.

    We've done this before

    See how we do Enterprise AI
  • Digital CDMO

    Instrument connectivity for legacy laboratory equipment

    An instrumentation layer that brings readings from legacy controllers and vendor-locked software into a documented, vendor-neutral data path, so each instrument on the bench contributes to the same screening record.

    • Protocol bridge for legacy controllers

      RS-232 and Profibus into modern protocols

      Connect legacy controllers and bench instruments through OPC UA or MQTT so each reading leaves the equipment in a documented, vendor-neutral form rather than staying inside a closed controller.

    • Time-series store for screening runs

      One record of a run

      Stream instrument readings into a time-series store keyed to the chip batch and the screening protocol, so the laboratory sees one run rather than a folder of per-instrument files.

    • Segmentation and security baseline

      IEC 62443 zones from day one

      Define zones, conduits and remote-access rules to IEC 62443 (the international standard for cybersecurity in industrial automation), so the laboratory's network posture does not have to be re-argued as new instruments and partner integrations come online.

    • Each instrument's readings are reachable from the screening record rather than from a local export.
    • Connecting a new instrument becomes a documented step instead of a per-vendor integration project.
    • Cybersecurity is part of the laboratory's architecture from the first connected device.

    We've done this before

    See how we do Digital CDMO
  • Digital Lab

    Bioprocess digital twin for microfluidic perfusion

    A simulation environment that models fluid shear stress, nutrient gradients and perfusion dynamics in CanChip's vascularised chips, so the design variables that drive cell viability can be tightened in software before a physical batch is built.

    • Computational fluid dynamics model of perfusion loops

      Shear stress in software

      Build a CFD model of the perfusion loop in each chip design, calibrated against the bench data CanChip already holds, so the fluidic environment of a candidate design is known before the chip is fabricated.

    • Virtual batch testing

      In-silico screening of designs

      Run candidate designs through the model in parallel and rank them on predicted cell viability, so the physical batch budget goes to the designs most likely to perform.

    • Calibration against bench data

      Models that improve with each run

      Close the loop between simulation and bench, so each physical batch updates the model and the next design round starts closer to the answer.

    • Fewer wasted chips and shorter time-to-result for partner screening programmes.
    • Design knowledge compounds across runs instead of resetting each time the perfusion loop changes.
    • Partner conversations can start from a predicted performance envelope rather than a single bench result.

    We've done this before

    See how we do Digital Lab
  • Digital Lab

    Computer vision QC for cell counting and colony state

    Camera-based machine-learning models that automate cell detection and colony state classification in CanChip's screening workflow, replacing visual counting with a documented, repeatable step.

    • Cell detection model

      Counting that is consistent across operators

      Train and deploy a model that counts cells and classifies colony state from microscope images, with a model version attached to each reading so the QC step is reproducible.

    • Anomaly and contamination flags

      Issues surfaced during the run

      Run the same image through an anomaly detector so early signs of contamination or colony drift are flagged against the running batch rather than discovered after the screening completes.

    • QC reporting from the image set

      Reports generated, not assembled

      Generate the QC report from the image set and the model outputs, so the report carries the evidence chain back to the image, the model version and the timestamp rather than being assembled by hand.

    • Manual QC time drops and the result is the same across operators and shifts.
    • Anomalies are flagged during the run instead of after it, so the screening result reaches the partner with the QC story already attached.
    • Each QC decision is traceable to the image and the model version that produced it.

    We've done this before

    See how we do Digital Lab
  • Digital Lab

    AR-guided workflows for microfluidic assembly

    Assisted-reality procedures that overlay the next step of a microfluidic assembly or screening protocol directly into the operator's field of view, with remote expert support available when a step needs a second pair of hands.

    • AR-guided standard operating procedures

      The next step, in view

      Render each step of an assembly or screening protocol in the operator's field of view on smart glasses, with checks and confirmations logged against the run record.

    • Remote expert session

      A second pair of hands on demand

      Connect a remote expert into the operator's view for troubleshooting, so a complex microfluidic step does not have to wait for an on-site visit.

    • Procedure version control

      One source of truth for the step

      Keep every AR procedure tied to a versioned standard operating procedure, so the laboratory is always running the current version and a change to a step is reflected immediately in the workflow.

    • Operator onboarding time drops because each step is shown, not described.
    • Process compliance is visible during the run rather than reconstructed after it.
    • Specialist time is spent on the steps that need a specialist, not on routine questions.

    We've done this before

    See how we do Digital Lab

Digital maturity: today and target

Scored out of 100 across six dimensions. The target is what CanChip's own published ambition implies — not a perfect score.

Source: A4BEE analysis of public sources
Data Orchestration 22 → 85
Screening results are recorded on paper or held in spreadsheet files specific to each instrument. A shared model for assay, chip and batch is the change that makes partner reporting and any eventual regulatory submission read from the same record.
Lab Connectivity 38 → 82
Legacy controllers and vendor-locked instrument software hold readings locally. The move to OPC UA or MQTT for new connections sets a baseline that older equipment can be retrofitted to as it is replaced.
Process Simulation 18 → 80
Chip design iteration is largely bench-driven. Computational Fluid Dynamics on the perfusion loop, calibrated against existing bench data, is the path that compresses design cycles for partner programmes.
Automation Level 42 → 88
Microfluidic assembly and screening QC still depend on manual steps. The next gain comes from instrument integration and computer-vision QC, with assisted-reality procedures bridging the digital fluency gap during the transition.
Regulatory Readiness 32 → 88
ALCOA+ and EMA/FDA submission expectations apply as soon as work feeds a clinical package. Capturing screening data with attribution and audit-trail handling at the instrument is what keeps the submission evidence chain continuous.
Workforce Readiness 48 → 82
The team has strong scientific expertise in microfluidics and cell biology. The practical shift is in instrument-driven workflows and AR-guided procedures, which move the same scientific work into an industrial setting without losing the underlying expertise.

Check this yourself

Our Service Portal has free self-assessments and market comparisons. These are the ones that line up with what we've read above — no sales call required.

Think we've read this right?

Talk to us

Related reading

This is an independent analysis prepared by A4BEE from publicly available information as of January 2026. It reflects A4BEE's own interpretation and opinion, is not affiliated with, endorsed by, or verified with CanChip, and may be incomplete or inaccurate. All company names and trademarks are the property of their respective owners. To request a correction or removal, contact [email protected].