☰

Fresenius Kabi AG

Digital foundations for biopharma and MedTech growth

Industry
Pharmaceuticals and Medical Devices
Headquarters
Bad Homburg vor der Höhe, Germany
Public information as of
February 2026

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

Strategic priorities

Fresenius Kabi is running its #FutureFresenius programme, now in the phase it calls Rejuvenate: upgrading the core business while scaling two high-growth engines. Biopharma grew about 33 percent organically and MedTech's cell and gene therapy line about 40 percent in 2024, against a group EBIT margin target of 16.0 to 16.5 percent. The company frames the route to that margin as advanced manufacturing, standardised IT infrastructure and a move from dashboards to decision-support AI.

The portfolio spans four business units with very different data shapes: Pharma (generic IV drugs and fluids, about €3.8 billion in 2024), Nutrition (about €2.4 billion), MedTech (infusion systems and cell and gene therapy devices, about €1.6 billion) and the fast-scaling Biopharma unit (biosimilars and CDMO services via mAbxience, about €0.6 billion). Two recent acquisitions — mAbxience for biosimilar drug substance and Ivenix for smart infusion pumps — brought their own systems and tech-transfer work into the group.

Capacity is being re-shaped around a local-for-local strategy at the same time: high-capital sterile and prefilled-syringe investment at Wilson, North Carolina and Melrose Park, Illinois, an expanded R&D centre in Pune, India, and a sustainability commitment to cut absolute Scope 1 and 2 emissions by 50 percent by 2030 and reach net zero by 2050. All of these lean on the same underlying capability: process, quality and device data that can be read outside the equipment that produced it.

Challenges we see

  • Digital Integration

    Updating infusion pump software after it ships

    The Ivenix Large Volume Pump, part of the smart infusion portfolio Fresenius Kabi acquired in 2022, was the subject of recalls in 2024 addressed through remote software updates. A growing share of the device's function lives in software that can be changed after the unit is installed.

    When a device's behaviour can change after it leaves the factory, delivering, verifying and evidencing an update across an installed base becomes part of the product rather than an after-sales activity.

  • Operations Manufacturing

    Confirming package integrity on the line rather than after it

    An FDA inspection of the Canton, Massachusetts compounding operation, with a Form 483 issued in November 2025, recorded packaging observations including IV bag ports and container seals, where establishing the full set of affected lots after the fact is part of the follow-up work.

    Where seal integrity is confirmed by sampling after packing, the population under review is a whole lot. Inspecting each unit as it is packed narrows that population to the units that actually deviated.

  • Digital Integration

    Capturing microbiology results as data rather than observation

    Some microbiological results are recorded as a written turbidity observation, and several laboratory KPIs are tracked in spreadsheets held separately from live process trends. In November 2025 Fresenius Kabi issued a voluntary nationwide recall of three lots of Famotidine Injection over out-of-specification endotoxin results in reserve samples.

    Results that exist only as a written observation cannot be trended against the process that produced them, so a drift across batches is visible only once someone assembles the numbers by hand.

  • Operations Manufacturing

    Reading data off long-lived production lines

    Alongside state-of-the-art sites such as Wilson, North Carolina, the Pharma unit runs a global network of older bag-making and syringe-filling lines in Europe and India, some served by isolated industrial PCs that predate shared data layers.

    Equipment installed before shared data layers were standard still has years of service in it, so the practical question is what can be read out of it without disturbing a validated line.

  • Compliance Regulatory

    Evidencing aseptic technique across sites

    FDA 483 observations in 2024 and 2025 covering sterile operations referenced first-pass air in laminar flow hoods and the assessment of microbial growth, across sites including Canton, Massachusetts and Baddi, India. FDA drug inspections rose sharply in 2025, roughly 73 percent above the prior year by one industry count.

    Aseptic technique is trained once and performed thousands of times, so the evidence that matters is what actually happened at the bench — which is only available if the working session itself is recorded.

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. Making device software updatable across the installed base

    A growing share of the infusion portfolio ships with software that can change after installation, and the 2024 Ivenix recalls were resolved through remote updates. Delivering and evidencing an update across every deployed unit is a lifecycle question, not a one-time release.

    Moving the pump's software onto a resilient, containerised deployment model lets updates roll out with redundancy and a recorded audit trail, so a change reaches the installed base continuously rather than as an exceptional field action.

    • Fresenius Kabi Ivenix Infusion System connected to second recall in 2024, MD+DI
    • Fresenius 2024 Sustainability Statement, digital transformation section

  2. Inspecting each unit as it is packed

    Where seal and port integrity is confirmed by sampling after packing, a defect surfaces as a customer complaint and the affected population is a whole lot until an investigation narrows it. The Canton 483 in November 2025 recorded packaging observations of this kind.

    A vision inspection station on the packing line checks every unit as it is produced and records the result against the batch, so a deviation is caught at the point it happens and the affected population is the units that actually failed.

    • Fresenius Kabi Compounding LLC, Canton MA, Form 483 issued 11/05/2025, FDA

  3. Trending microbiology and QC results as data

    Microbiological results captured as written observations, and KPIs kept in spreadsheets apart from live process trends, cannot be trended automatically against the process that produced them. The November 2025 Famotidine recall followed out-of-specification endotoxin results in reserve samples.

    Capturing results at the instrument into a shared model with process context lets a drift across batches be trended as it forms, and gives the quality organisation the same live view of the process that the process scientists have.

    • Fresenius Kabi issues voluntary nationwide recall of three lots of Famotidine Injection, FDA, November 2025
    • Fresenius 2024 Annual Report, digital transformation

  4. Reading data off legacy lines without disturbing them

    Older bag-making and syringe-filling lines served by isolated industrial PCs hold years of remaining service life, but their data does not reach shared systems. Replacing them is neither necessary nor timely, so the practical question is what can be read out non-intrusively.

    Retrofitting a data-acquisition layer that reads process values from existing controllers without touching the validated control logic brings legacy lines into the same data estate as new sites, and sets up condition monitoring on assets kept in service.

    • Fresenius Kabi Oncology manufacturing facilities, company site
    • Fresenius 2024 Annual Report, opportunities and risk report

  5. Recording the working session for aseptic evidence

    Aseptic technique is trained once and performed thousands of times, and the evidence that a specific session followed protocol is only available if the session itself is observed. FDA 483 observations in 2024 and 2025 referenced first-pass air and microbial-growth assessment across sterile sites.

    External cameras and computer vision can record and flag deviations from laminar-flow and cleaning protocols at the bench in real time, turning aseptic evidence from a periodic training record into an account of what actually happened during each session.

    • USFDA 483 observations 2024, sterile drug manufacturing trends
    • FDA inspections in 2025: heightened rigor, Reed Smith

What we'd propose

  • Digital CDMO

    Resilient software deployment for connected infusion devices

    We move the device software onto a containerised, high-availability deployment model with redundancy and a recorded audit trail, so updates reach the installed base continuously and with evidence rather than as an exceptional field action.

    • Containerised deployment

      Software decoupled from hardware

      Run the device software in containers with N+1 redundancy so a node can be updated or fail over without interrupting the service, and a new version can be staged and rolled out rather than swapped in place.

    • Evidenced update pipeline

      Updates with a recorded trail

      Build the release and rollout as a pipeline that records what version reached which unit and when, so the evidence a regulator asks for after a field update is generated by the process instead of reconstructed from it.

    • Fleet health and version visibility

      One view of the installed base

      Surface the software state of the deployed fleet in one place so the scope of any update or correction is known from the record on day one rather than assembled per site.

    • A software change reaches the installed base as a controlled rollout instead of an exceptional field action.
    • The evidence behind an update is produced by the pipeline, so a post-update inquiry starts from the record.
    • Redundancy removes the single points of failure that make an update disruptive to plan.

    We've done this before

    See how we do Digital CDMO
  • Digital CDMO

    In-line vision inspection for the packing line

    A non-invasive computer-vision inspection station on the packing line that checks each unit as it is produced — seal fusion, port integrity, fill and label — and records the result against the batch, so a deviation is caught at the point it happens.

    • Per-unit defect detection

      Every unit checked, not a sample

      Use line-side cameras and image processing to check each packed unit for seal, port and fill defects in real time, so the population under review narrows from a whole lot to the units that actually deviated.

    • Result written to the batch record

      Inspection tied to the batch

      Record each inspection outcome against the specific unit and batch, so the set of affected units is known from the data rather than reconstructed during a complaint investigation.

    • Drift and trend monitoring

      Early warning across a run

      Track the rate and type of flagged defects across a run so a developing problem on the line is visible before it produces a batch of complaints.

    • A defect is caught at the point it happens rather than as a customer complaint.
    • The affected population is the units that failed, not the whole lot, because every unit carries a recorded result.
    • Line trends give early warning before a developing fault reaches a batch of shipped product.

    We've done this before

    See how we do Digital CDMO
  • Enterprise AI

    One data model for microbiology and QC results

    An ontology-based data platform that captures microbiology and QC results at the instrument into a shared model with process context, so results are trended automatically against the batches and processes that produced them instead of assembled by hand.

    • Instrument data capture

      Results captured at source

      Connect laboratory instruments through OPC UA (Open Platform Communications Unified Architecture) and vendor connectors so results arrive as data with instrument identity, method version and timestamp, rather than as a written observation.

    • Shared model with process context

      One agreed set of terms

      Define sample, result, batch, instrument and lot as explicit entities with agreed relationships, so a microbiology or endotoxin result can be trended against the process run that produced it.

    • Live trending and retrieval

      Drift visible as it forms

      Expose the model through dashboards and a retrieval layer so quality and process teams see a drift across batches as it develops, rather than after someone assembles the numbers for a report.

    • Results are trended against the process that produced them instead of assembled by hand for review.
    • Quality and process teams share one live view rather than reconciling separate spreadsheets.
    • An out-of-specification result is visible in context, so the investigation starts from the record.

    We've done this before

    See how we do Enterprise AI
  • Digital CDMO

    Data acquisition retrofit for legacy production lines

    We add a non-intrusive data-acquisition layer that reads process values from existing controllers without touching the validated control logic, bringing older bag-making and syringe-filling lines into the same data estate as newer sites.

    • Non-intrusive line readout

      Data without touching the logic

      Read process values from existing PLCs and industrial PCs through a separate acquisition layer, so a validated line keeps running unchanged while its data becomes available to shared systems.

    • Unified IT/OT data path

      Shop floor to enterprise

      Bridge shop-floor signals to enterprise reporting with segmentation and OPC UA or MQTT, so legacy lines report into the same estate as new sites and audit questions are answered from one place.

    • Condition monitoring on kept assets

      More life from installed equipment

      Use the acquired data to watch for the early signs of wear on equipment kept in service, so maintenance is planned from signals rather than from failure.

    • Legacy lines join the shared data estate without a rip-and-replace or a control-system revalidation.
    • Audit and continuity questions are answered from one data path instead of per-machine.
    • Condition data extends the useful life of equipment that still has years of service in it.

    We've done this before

    See how we do Digital CDMO
  • Agents

    AI agents for regulatory and quality document work

    Narrow, reviewable agents that take the repetitive part of document work: drafting deviation, CAPA and tech-transfer summaries from source records, checking a document against its template before review, and finding every controlled document a standards change affects. A named person approves every output.

    • Drafting from source records

      First drafts from system data

      Generate the first draft of a deviation, CAPA or tech-transfer document directly from the underlying system records, so the author edits and judges rather than assembles.

    • Template and completeness checking

      Gaps found before review

      Check a submitted document against its template and the site's checklist, returning missing or inconsistent sections before it enters the human review queue.

    • Change impact search across the document set

      Which documents a change touches

      When a standard, method or specification changes, retrieve every controlled document that references it and rank them by how directly they are affected, so the update scope is known on day one.

    • Review queues move faster because documents arrive complete.
    • The scope of a standards or method change is established by search rather than by recollection.
    • Every output is traceable to the source records it came from and signed off by a named reviewer.

    We've done this before

    See how we do Agents

Digital maturity: today and target

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

Source: A4BEE analysis of public sources
Device software lifecycle 38 → 82
A growing share of MedTech value sits in field software, and the 2024 Ivenix recalls were addressed through remote updates. Continuous, evidenced deployment across an installed base raises the bar above a per-release validation.
In-line quality inspection 42 → 84
Package and seal integrity is confirmed largely by sampling after packing, which the November 2025 Canton 483 packaging observations are consistent with. Per-unit inspection at the line is ahead of the current approach rather than behind it.
Lab and QC data 35 → 80
Some microbiology results are recorded as written observations and several KPIs live in spreadsheets apart from live trends, which favours capturing results as data with process context over manual assembly.
IT and OT connectivity 40 → 80
New sites such as Wilson are highly automated, while older European and Indian lines run on isolated industrial PCs, so the estate spans several equipment generations and a shared data path is the work ahead.
Batch and process trending 33 → 78
The November 2025 Famotidine endotoxin recall and ongoing FDA attention both favour results that can be trended against the process as they form over numbers assembled for periodic review.
Site energy and emissions 40 → 74
The 2030 target of a 50 percent cut in absolute Scope 1 and 2 emissions needs per-site and ideally per-line energy measurement to be managed through the year rather than reported after it.

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 February 2026. It reflects A4BEE's own interpretation and opinion, is not affiliated with, endorsed by, or verified with Fresenius Kabi AG, 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].