iPSCs: The Industrialisation Problem No One Talks About
Construct → Platform → Industrialisation | Series #3 — Part 2 (Industrialisation)
Mapping the Field
I’m building a practical map of CGT that connects science to strategy using one consistent framework: Construct → Platform → Industrialisation. The goal isn’t to define terms — it’s to surface where value is created, where bottlenecks move, and what capabilities actually win. Each week I take on a different therapeutic construct, this week induced pluripotent stem cells (iPSCs). Subscribe (free) to follow the map as it builds.
🏭 iPSC Industrialisation — Two Stages, Two Chances to Fail
Anyone who has worked with iPSCs knows they’re the hothouse flower of the CGT world. Too confluent? Spontaneous differentiation. Cytokine concentration slightly off? Spontaneous differentiation. Matrix conditions not quite right? Spontaneous differentiation. Even in skilled hands, it’s easy to end up with a heterogeneous mess. That fragility is not just a bench nuisance, it’s the central industrial problem.
iPSCs are being developed into a wide range of therapeutic products, from immune effector cells to tissue-replacement therapies. But the industrial question is not what they can become in principle. It is whether those cell states can be produced reproducibly, safely and at scale.
iPSC industrialisation is ultimately a two-stage control challenge. First, preserving a stable pluripotent lineage that can be banked, expanded and reused. Then, directing that lineage into defined therapeutic cell types without losing identity, function or safety. At bench scale, small deviations are a project problem. At manufacturing scale, they become a problem that can break the business.
🏭 What is the iPSC industrialisation model?
In my framework, the industrialisation model is the production and delivery system that turns a therapy into a scalable, reproducible product. For iPSC, this production architecture is fundamentally two-stage.
First, a pluripotent starting lineage is generated, engineered and banked as a renewable upstream substrate. This creates a standardised cell source that can be expanded independently of any individual patient or product.
Second, that lineage is directed through controlled differentiation into specific therapeutic cell types, which are then manufactured, tested and delivered as the final product.
Unlike CAR-T, where industrialisation debates collapse to competing models (autologous vs allogeneic), iPSC follows a single but inherently split architecture. The challenge is not renewability alone, but preserving control across that transition so identity, function and safety hold repeatedly at scale.
The industrial question, then, is whether that control is strong enough to turn biological promise into a reliable product.
🏭 Industrialisation Model Value Propositions — Stage 1 (Pluripotent lineage generation)
On a good day, pluripotent lineage generation turns a fragile biological state into something much more powerful: a renewable, standardised and reusable upstream cell source. In practice, elements of this architecture are now being realised in clinical and preclinical programs, but maintaining stable, comparable cell state at scale remains highly conditional.
Supply model — renewable upstream inventory.
Pluripotent lineage generation creates a self-renewing starting material that can be expanded, banked and reused, rather than re-sourced for each batch. This shifts iPSC from a donor- or patient-dependent supply model to an inventory-based system built on a stable biological substrate. This is powerful because it decouples manufacturing from biological scarcity and turns the starting lineage into a reusable super-substrate which can support the generation of multiple cell therapies.
Manufacturing architecture — primarily scale-up, not scale-out
Because the same pluripotent lineage can be expanded indefinitely, upstream production starts to resemble a more traditional scale-up manufacturing model built on controlled expansion systems rather than patient-coupled single-batch manufacture. This changes the industrial logic completely: growth comes from expanding a shared upstream lineage, not from multiplying bespoke manufacturing runs one batch at a time.
Product consistency — standardised starting material.
A banked pluripotent lineage offers the potential for a more consistent manufacturing input than variable donor- or patient-derived cells. Once established and characterised, the same starting population can be used repeatedly across batches and programs. That matters because downstream processes no longer begin from a shifting biological input, making product identity, process control and critical quality attributes easier to reproduce over time.
Platform leverage — upstream engineering propagates downstream.
Engineering performed at the pluripotent stage is inherited across all differentiated descendants. Compatibility features, safety controls or functional programs can be embedded once and carried through multiple downstream products.
This turns the starting lineage into a programmable substrate rather than passive input material, allowing engineering effort to compound across a portfolio.
Portfolio expansion — one lineage, multiple products.
A single pluripotent cell line can support differentiation into multiple therapeutic cell types, enabling expansion across indications without rebuilding the upstream biological foundation. That means that new products begin to resemble new developmental programs applied to an existing lineage, rather than entirely new manufacturing systems.
Supply security — banking as infrastructure.
Master and working cell banks create stored, qualified inventory that supports manufacturing continuity, comparability and lifecycle planning. Bank design (size, redundancy, storage conditions) becomes part of the production architecture rather than a downstream detail.
The value is not just storage — it is the ability to build a controlled, durable supply base that can support long-term manufacturing and regulatory consistency.
Quality control — upstream control of downstream risk.
Donor selection, reprogramming method, clone selection and early characterisation allow key quality attributes to be designed into the system before large-scale manufacturing begins.
This front-loads control into the earliest stage of production, shaping critical quality attributes before scale-up and reducing reliance on downstream correction, increasing the likelihood that differentiated products behave predictably.
Automation potential — reducing operator-driven variability.
Upstream processes such as colony handling, passaging, expansion and monitoring are increasingly compatible with automation and closed systems. The industrial value is that a biologically sensitive upstream process becomes less dependent on manual skill and more compatible with reproducible scale-up and lower-cost manufacture.
Economics — lower long-run sourcing burden.
Once a stable pluripotent lineage is established, the need for repeated donor sourcing, testing and qualification is reduced. The result is a different cost base: less recurring spend on biological sourcing, but greater dependence on maintaining lineage stability and scalable banking systems.
The appeal of iPSC is clear: a renewable starting lineage, a shift from one-off sourcing to biological inventory, and the possibility of scaling cell therapy on a reusable upstream foundation. But I’ve described the perfect-world iPSC manufacture. The constraints below are where that promise gets pulled back into reality:
🏭 iPSC Constraints — Stage 1 (Pluripotent lineage generation)
Supply model can collapse under instability.
The promise of a renewable upstream inventory depends on the lineage remaining stable over time. In practice, iPSCs are prone to genomic drift, epigenetic variation and selection pressures during expansion.
If the lineage changes, the “inventory” is no longer equivalent across time — and supply stops behaving like a reusable asset.
Scale-up amplifies loss of control.
Controlled expansion at small scale does not translate cleanly to larger systems. Aggregate size, oxygen gradients, shear, media distribution and feeding strategies all influence cell state.
As scale increases, small deviations compound — and maintaining pluripotency and uniformity becomes harder.
Standardised starting material is harder than it looks.
A banked lineage does not guarantee a consistent starting population. Passage number, culture history, thaw conditions and subtle process differences can shift behaviour, and hence the initial biological state.
Upstream engineering can propagate risk as well as value.
Edits introduced at the pluripotent stage are inherited across all downstream products. Off-target effects, unintended functional consequences or stability issues are not localised and can carry through an entire portfolio.
Upstream leverage cuts both ways: mistakes compound as efficiently as successes.
Portfolio expansion is constrained by lineage behaviour.
Not all iPSC lines differentiate equally well into all target cell types. Line-to-line variability in differentiation efficiency, maturity and functional output can limit the practical breadth of a single lineage.
In practice, portfolio expansion may require multiple fit-for-purpose lines rather than one universal foundation.
Banking does not freeze behaviour.
Cryopreservation stabilises availability, not biological equivalence. Freeze–thaw stress, recovery conditions and long-term storage effects can alter cell state.
Comparability across banks, sites and time can become a central constraint.
Quality is difficult to lock in upstream.
Early characterisation can define identity and basic quality attributes, but many relevant properties only emerge later. Even with extensive testing at the banking stage, lineage stability, differentiation potential and long-term behaviour are difficult to fully resolve, leaving uncertainty embedded in the starting material.
Raw materials introduce upstream variability.
iPSC culture and expansion depend on defined media, matrices, enzymes and supplements, many of which are biologically derived or sensitive to handling and preparation.
Reagent quality, lot-to-lot variability and contamination risk can influence cell state, making raw materials a first-order driver of reproducibility rather than a background variable.
GMP burden begins at lineage establishment.
Donor eligibility, reprogramming method, clone selection and master cell bank generation are not just scientific steps — they are regulated manufacturing activities.
Decisions made at this stage define comparability, traceability and long-term regulatory strategy, meaning upstream choices are difficult to change without significant revalidation.
Automation does not remove biological sensitivity.
Automation can standardise execution of upstream steps such as passaging, expansion and monitoring, but iPSCs remain highly responsive to critical process parameters and microenvironmental variation. Variability in aggregate behaviour, shear exposure, oxygen transfer, media exchange and culture history can still shift cell state and downstream performance. Automation can tighten execution, but it does not eliminate the biological control problem.
Economics are conditional on stability.
The economic advantage of reduced sourcing depends on the lineage remaining usable over time. If lines require replacement, requalification or parallel maintenance, cost structures quickly revert toward repeated biological sourcing. In addition, if master and working banks cannot be generated reproducibly at scale due to heterogeneity, loss of pluripotency or process sensitivity, timelines extend and cost advantages fail to materialise.
The model only improves if the upstream foundation can be both stabilised and reliably scaled.
Strategic consequence (Stage 1)
The value of a renewable upstream lineage only holds if identity, stability and developmental potential can be preserved across expansion, banking and time. Advantage compounds through cell-state stability, comparability, raw-material consistency and disciplined process execution, within a regulatory framework heavily shaped at lineage establishment.
🏭 iPSC take-home through the four industrialisation outcomes (Stage 1)
• Scalability:
iPSC stage 1 scales in principle through expansion of a reusable upstream lineage (scale-up, not scale-out), but capacity is gated by the ability to maintain pluripotency, uniformity and comparability during expansion and banking. Growth is only real if cell state holds as volume increases.
• Capital efficiency:
Reduced dependence on repeated donor sourcing can improve long-run cost structure, shifting spend toward expansion systems, banking, raw materials and process control. However, economics are highly sensitive to lineage stability, bank generation success and regulatory burden — if lines drift, fail or require requalification, cost advantages erode quickly.
• Reliability:
A banked lineage offers the potential for consistent starting material and more predictable downstream performance, but reliability depends on maintaining equivalence across passages, banks and time. Variability in cell state, recovery, culture history and raw-material inputs can undermine reproducibility if not tightly controlled.
• Changeability:
Upstream engineering and a reusable lineage create the potential for cumulative improvement across a portfolio, but changes are not fully local. Modifications to the lineage, process, raw materials or banking strategy can propagate across all downstream products, requiring comparability and regulatory revalidation rather than simple plug-and-play iteration.
🏭 iPSC diligence — Stage 1 (pluripotent lineage generation)
🔎 iPSC Stage 1 diligence (the questions that separate a banked lineage from a real upstream manufacturing system)
Lineage stability: how stable is the line across passages and time — what drift signals are tracked, and when do they trigger intervention or retirement?
Bank comparability: how comparable are master and working cell banks (and across sites) — what defines equivalence, and how often is it actually demonstrated?
Cell state control: which process parameters truly govern pluripotency and uniformity (aggregate size, feeding, shear, oxygen, matrix), and how tightly are they controlled at scale?
Scale behaviour: what changes when you move from small-scale to production systems — where does cell state start to shift, and what are the operating limits?
Bank generation success: what is the right-first-time rate for generating and qualifying new lines and banks, and how long does it take end-to-end?
Hidden heterogeneity: how is subclonal variation or epigenetic drift detected and managed over time — what does the population actually look like beyond bulk averages?
Upstream QC resolution: which quality attributes are measured at banking, and which only become visible later — where does uncertainty remain in the starting material?
Lineage–product fit: how consistent is differentiation performance across target cell types — where does the lineage work well, and where does it break?
Automation vs biology: where does automation genuinely improve reproducibility, and where does biological sensitivity still dominate outcomes?
Lifecycle management: how often do lines require requalification, replacement or parallel maintenance — and what does that do to cost, timelines and comparability?
Raw-material dependence: which media, matrices and reagents are critical, how variable are they lot-to-lot, and how is that variability controlled?
Regulatory lock-in: which upstream decisions (reprogramming method, clone selection, banking strategy) are effectively fixed, and what is the cost/timeline of changing them?
If iPSC is the reusable-lineage bet — powerful when cell state holds — Stage 1 diligence is about whether that lineage can actually behave like a stable manufacturing substrate over time, not just a well-characterised starting point.
But preserving pluripotency is only half the industrial problem. Once a stable lineage exists, the next question is whether that developmental potential can be developed into a defined therapeutic product — reproducibly, safely, and at scale.
🏭 Industrialisation Model Value Propositions — Stage 2 (Differentiation → product formation)
For Stage 2, industrial value comes from turning developmental plasticity into controlled, functional cell identity — reproducibly and at scale.
Product definition and consistency — controlled cell identity
Directed differentiation allows a pluripotent lineage to be converted into a defined therapeutic cell type with specified identity, composition and function. The value is not perfect uniformity, but tighter control over developmental inputs so the final product is better defined and behaves more predictably across batches.
Platform leverage — shared differentiation logic
Once differentiation pathways are established and optimised, they can be reused and adapted across programs targeting similar cell types or lineages.
This allows process knowledge and developmental control strategies to compound across a portfolio rather than being rebuilt from scratch each time.
Delivery model — standardised drug product
Differentiated iPSC-derived cells can be formulated, stored and delivered as standardised products, often allogeneic. The extent of standardisation depends on product format, with some systems still requiring tighter handling or shorter shelf life.
Quality control — testing biological state, not just process
Product quality can be defined through phenotypic markers and functional assays that reflect the differentiated cell state. In principle, this allows the product to be released based on what the cells actually are and do, rather than relying primarily on control of upstream process inputs.
Scalability — amplification of a defined cell type
Once a differentiation process is established, production can scale through expansion and parallelisation of defined workflows. Large quantities of a specific therapeutic cell type can be generated from a common upstream lineage, provided cell identity, composition and function remain controlled. In practice, the mode of scaling depends on the product format — suspension systems may enable scale-up, while adherent or scaffold-based processes may require scale-out or hybrid approaches.
Economics — value per batch, not per patient
Differentiation enables production of multiple doses from a single upstream lineage and manufacturing run, shifting economics toward batch-based production rather than patient-specific cost structures. This creates the potential for more scalable and predictable cost models compared to bespoke cell therapies. In practice, the extent of this advantage depends on process format — more complex, adherent or scaffold-based systems may retain higher cost structures and limit the degree to which batch economics can be realised.
Stage 2 creates value only when developmental biology can be translated into controlled product formation; when control slips, the same architecture becomes a source of variability, safety risk and economic drag.
🏭 iPSC Constraints — Stage 2 (Differentiation → product formation)
Product definition can blur under process variability
Directed differentiation is not perfectly deterministic. Small variations in timing, media composition, signalling strength and culture conditions can shift cell-fate trajectories, causing yield, purity and composition to fluctuate between runs. Even under controlled conditions, differentiated populations may still contain subpopulations at different stages of maturity or off-target lineages. Instead of a clean, well-defined product, the output can become a mixed or poorly resolved population, with the “right” cells present but not in the same proportions or states every time.
Differentiation protocols do not transfer cleanly.
Processes optimised for one lineage, clone or scale do not always translate to others. This is further complicated by differences in product format (e.g. suspension vs adherent or scaffold-based systems), which can alter pathway dynamics and scaling behaviour.Residual pluripotent cells create a persistent safety constraint.
Incomplete differentiation can leave behind undifferentiated iPSCs with tumorigenic potential.
Ensuring their removal or control is not just a quality consideration — it is a fundamental safety requirement embedded in the manufacturing process.
Potency is difficult to define meaningfully
Although differentiated products can be characterised through phenotypic markers and functional assays, linking those readouts to true therapeutic potency is not always straightforward. The challenge is not just measurement, but whether the measured attributes genuinely reflect the biological activity that drives clinical effect.
Quality control can become an operational bottleneck
Release testing often depends on multi-parameter phenotypic and functional assays that are time-consuming, variable and difficult to standardise. Quality is not a single readout, and assay complexity can slow release, reduce throughput and become a gating step for manufacturing cadence.
Raw materials introduce hidden variability.
Differentiation often depends on complex, phase-specific combinations of cytokines, growth factors, small molecules and matrices.
Reagent quality, lot-to-lot variability and preparation can become first-order drivers of process variability rather than background noise.
Delivery requirements reintroduce complexity.
Formulation, storage class (fresh vs cryopreserved), and in-use stability can affect product viability and function.
Even if differentiation succeeds, distribution and administration constraints can limit practical deployment.
Scale amplifies developmental variability.
Differentiation conditions that work at development scale do not always hold at production scale. Because Stage 2 is longer, more complex and more biologically sensitive than Stage 1, increasing scale makes it harder to preserve consistent fate specification, composition and function.
Economics are burdened by process complexity.
Multi-step differentiation workflows often require repeated media changes, phase-specific cytokine or small-molecule cocktails, enrichment steps and tighter in-process controls, all of which add intrinsic cost. Each additional intervention also creates another opportunity for execution error, variability or batch failure, increasing both risk and cost. The model only becomes economically attractive if functional cells can be produced with sufficient yield, purity and consistency without the process itself becoming too expensive or too fragile to run.
Strategic consequence (Stage 2)
The value of iPSC only materialises if cell identity, function and composition can be achieved through differentiation, not just specified in theory. Yield, purity, potency and release all become gating steps. Advantage compounds through functional consistency, assay strategy and process simplicity, not simply through the ability to generate cells.
🏭 iPSC take-home through the four industrialisation outcomes (Stage 2)
• Scalability:
Scales through expansion and parallelisation of differentiated cell production, with the mode of scaling dependent on product format. Throughput is gated by differentiation efficiency, purity and the ability to maintain consistent cell fate at scale. Growth is conditional on controlling variability across increasingly complex processes.
• Capital efficiency:
Batch-based production can improve cost structure relative to patient-specific therapies, but COGs are highly sensitive to differentiation yield, recovery and process complexity. Multi-step workflows, expensive cytokine cocktails and assay burden mean cost advantages only emerge if processes are both efficient and simplified.
• Reliability:
Product consistency depends on controlling cell identity, composition and function across batches. Variability in differentiation, hidden heterogeneity and assay limitations can undermine reproducibility. Reliability is not just about starting material — it is about consistently resolving that material into the same functional product.
• Changeability:
Differentiation processes, assay strategies and product definitions are tightly coupled. Changes to protocol, scale, raw materials or lineage can propagate into product identity, potency and release criteria. Iteration therefore requires comparability, revalidation and often system-level adjustment rather than local optimisation.
🏭 iPSC diligence — Stage 2 (Differentiation → product formation)
🔎 iPSC Stage 2 diligence (the questions that separate a differentiation protocol from a manufacturable product)
Identity vs composition: what does the final population actually look like (purity, subpopulations, maturity), and how tight is that distribution batch-to-batch?
Function vs phenotype: how well do phenotypic markers correlate with functional performance — and where do they break?
Potency definition: what is the true potency assay, how variable is it, and does it meaningfully reflect clinical mechanism of action?
Residual risk: how are undifferentiated or off-target cells detected, quantified and controlled — and what are the failure thresholds?
Yield & recovery: what is the end-to-end yield from lineage to final product, and where are the major losses (differentiation, purification, formulation)?
Right-first-time: how often does the differentiation process work as intended, and where do failures cluster (timing, media, transitions, scale)?
Process complexity: how many steps, interventions and media changes are required — and which ones are most error-prone or operator-sensitive?
Raw-material dependence: which cytokines, growth factors or matrices are critical, how variable are they lot-to-lot, and how is that variability controlled?
Scale behaviour: what changes when moving from development to production scale — where does fate specification start to drift?
QC gating: what is the true release bottleneck (assay type, duration, variability), and how does that constrain manufacturing cadence?
Comparability burden: which changes (process, materials, scale, lineage) trigger revalidation or regulatory work — and how local vs system-wide are they?
🏭 Conclusion
iPSC is often framed as the scalable future of cell therapy: a renewable starting point that breaks the limits of donor dependence and patient-coupled manufacture.
But industrially, that promise is split in two.
Stage 1 buys a reusable lineage and the possibility of biological inventory — but only if that lineage remains stable enough to behave like a real manufacturing substrate over time. Stage 2 buys defined therapeutic cell products and batch economics — but only if differentiation can repeatedly resolve pluripotent potential into the same identity, function and composition at release.
That is the real ceiling.
If the lineage drifts, the supply model collapses. If differentiation slips, the product definition blurs. And if either side fails, iPSC does not scale into a platform — it falls back into variability, assay burden and cost.
So iPSC is not simply a bet on better biology. It is a bet that developmental biology can be industrialised without losing control across two fundamentally different production problems — and across product formats that do not all scale in the same way.
The science makes iPSC possible.
iPSC doesn’t become scalable when the biology works. It becomes scalable when control survives both stages.
Next: in vivo gene therapy — where there is no manufacturing process to optimise, no intermediate to test, and no opportunity to correct the biology once it’s in the patient. If you’d like to follow the series as the map expands, subscribe here.