A plan to own a plant and a plan to bring an ingredient to market are different commitments. Contract production or shared infrastructure can let a team test its process without financing the whole facility. Ownership can become appropriate when demand, process performance and utilisation support it.
The relevant question is which production arrangement fits this process at this stage, rather than which arrangement looks most ambitious.
A facility also introduces fixed costs, construction dependencies and commissioning risk. Those commitments can be difficult to reverse if the product, process or market changes. This is a capital-allocation mechanism, not a claim that one decision explains a named company’s failure.
The pattern is structural: Food Biotech startups that build infrastructure before their process is transfer-grade are converting venture capital into fixed assets before they know what the process actually costs at scale. The ones that rent first - shared facilities, CDMOs, co-location partnerships - are not making a compromise. They are making a better capital allocation decision.
This issue gives you the framework for that decision. When renting is smarter. When ownership becomes justified. And the 10-question checklist that tells you which stage you are actually in, not the stage your pitch deck describes.
The assumption to test
The myth: Owning your own fermentation plant signals seriousness, operational control, and competitive moat.
The reality: Owning steel before your process is transfer-grade is not a moat. It’s a fixed cost around an assumption.
One-minute brief
Founders: compare renting, co-location and ownership against documented process performance, customer demand and financing. The questions below are a discussion guide. They do not produce a validated readiness score.
This comparison helps identify the risk of financing a facility before the process has been tested under an external operating arrangement.
Investors: ask what has been measured, where an external operator has run the process and how demand supports planned utilisation. Missing evidence is a reason to examine the investment case, rather than an automatic rule against every capital purchase.
The checklist in this issue converts that assessment into a structured diligence filter.
R&D teams: establish reproducibility, test an external transfer and update the cost model. Development is iterative; the order and amount of evidence depend on the process and the next commitment.
The checklist maps these directly to the experiments and documentation tasks that move you from RENT to BUILD. The sequencing matters more than the science.
What the evidence needs to show
Compare process improvement with added capacity
The strongest quantitative argument against premature infrastructure is not about discipline. It is about arithmetic.
A cost model should compare improving titre with adding capacity. Their relative benefit depends on the process, recovery, facility design and utilisation. Modelled scaling relationships from a specific scenario should not be presented as universal laws or a fixed percentage advantage for biology.
Building capacity before testing the main process assumptions can lock capital into a design that later needs to change. Compare quotations and scenarios for the actual product instead of relying on one generic facility-cost range.
GFI’s techno-economic review treats feedstocks, capital costs and process performance as important drivers. Model upstream and downstream requirements explicitly. Neither one facility’s capital estimate nor one downstream cost share is a universal input for a new process.
Industry facility counts need their original definitions and dates. An innovation centre, pilot plant and commercial production site do different jobs and should not be counted as interchangeable capacity.
A disclosed tank is not necessarily available, compatible food-grade capacity. Check the operator, process restrictions, downstream equipment, campaign schedule and commercial terms before treating a capacity listing as a production option.
The structural pattern behind premature builds
Three mechanisms drive founders toward building too early.
The facility-announcement trap is to treat a construction milestone as evidence of production performance. Construction, commissioning, regulatory permission and commercially useful throughput are separate dependencies. Each needs its own evidence and contingency.
Grant conditions can influence what must be built, where and by when. Compare those conditions with the process-development plan. NIZO announced a EUR 12.5 million grant supporting BFF-Ede’s shared scale-up facilities; that announcement does not establish a universal incentive or a currently available production slot.
Facility ownership can become an identity marker. Treat that as a question to investigate, not a diagnosis of a founder’s motives. Ask which technical or commercial requirement ownership resolves and whether a partner can resolve it sooner.
For founders
Use the ten questions in this guide to identify the missing evidence. Do not add the answers into a universal readiness score. A single material safety, financing or process constraint can change the decision.
Illustrative example: a dairy-protein team has internal data at 200 litres but has not transferred the process. Its next useful milestone may be an external trial, rather than a facility design. Confirm compatibility and availability with an operator before putting that trial in the plan.
Vivici and Liberation Labs announced a manufacturing partnership in May 2025. It is a concrete example of using a production partner. The announcement does not, by itself, disclose Vivici’s full capital allocation, prove achieved production economics or establish that renting is always preferable.
Examples of organisations to investigate for relevant European development or scale-up support:
BFF-Ede, Netherlands: NIZO announced an open-access fermentation scale-up project with capacity up to 10,000 litres. Its June 2025 announcement described a future site. Confirm the current commissioning status, process fit and availability directly.
21st.BIO, Denmark: a potential fermentation development and scale-up partner. Confirm the host system, equipment, food-grade status, confidentiality terms and available campaign capacity for your process.
CPI, UK: a research and scale-up organisation to investigate for relevant process-development support. Confirm the specific facility and equipment required; a general capability does not establish a compatible production slot.
Before the next facility-planning meeting, list the unanswered questions and compare the next external trial with the next ownership commitment.
Before you contact a CDMO, write the batch record. The act of writing it will surface more process gaps than the next six months of bench work - not because the writing reveals new science, but because it forces you to distinguish between what you know and what you have assumed.
For investors
For diligence, ask the team to show evidence for each material assumption. Link financing milestones to that evidence where appropriate. The guide is not a validated investment score and does not supply a universal cutoff for capital spending.
Three questions that surface this risk in the first meeting, before formal diligence begins:
1. ‘What is the highest scale at which this process has been operated by someone other than your founding team?’
A specific answer - a facility name, a volume, a yield comparison to internal benchmarks - means Question 2 in the checklist is answered. A vague answer - ‘we have spoken to several CDMOs’ - means it is not.
2. ‘What is your measured DSP recovery rate across multiple runs?’
An estimated recovery rate is an assumption. Show its process basis, the uncertainty and the sensitivity of the cost model. A measured recovery rate needs defined inputs, outputs and repeated runs. There is no single downstream cost share for all food proteins.
3. ‘What percentage of your planned Year 1 production volume has a signed offtake commitment?’
Signed, not discussed. An LOI under negotiation is not a commitment. A pilot agreement with a corporate partner is not a procurement commitment. The distinction matters because facility capex amortisation depends on utilisation, and utilisation depends on demand that is contracted, not projected.
The risk is committing to fixed costs before process and demand evidence support utilisation. Investigate the company’s actual dependencies; a checklist does not establish the cause of a named company’s failure.
For R&D teams
Link each unanswered question to an experiment, a document or an operator discussion. Process development, external trials and cost modelling should inform one another. There is no universal sequence that replaces the requirements of the actual project.
Compare process improvement and scale in the same model. Higher titre can reduce some costs per kilogram, but it does not automatically halve total COGS, and capacity increases do not follow one universal cost-reduction factor.
The three transfer-grade documentation standards required before the first external facility engagement:
Batch records complete enough for an external operator to follow without asking a question. Media formulation with supplier-specific specifications, CPPs with confirmed acceptable ranges (not just targets), in-process control points, expected performance bounds.
Contamination event log with root-cause analysis for each incident. Not a summary. A log, with dates, vessel identities, observed symptoms, probable cause, corrective action.
A downstream protocol with recovery data from repeated runs, including the input/output basis, variation and product specification. Choose the number of runs to fit the decision and process variability; three is not a universal validation standard.
These documents make a technical review more productive. Each operator has its own intake requirements, safety constraints and acceptance criteria.
The Build-vs-Rent Decision Framework
Use the ten questions below as a structured comparison of renting, co-location and ownership. They are included on this page; no unavailable spreadsheet or paid companion is required.
The outcome is a reasoned next step with named dependencies. It may be an external trial, a co-location discussion or further ownership engineering. It is not an automatic investment instruction.
Ten questions for the production decision
- What saleable product and specification are being transferred?
- Which process has been run by an external operator?
- What recovery has been measured across repeated runs?
- Which safety and regulatory requirements apply?
- Which equipment and operator are compatible?
- What demand supports the planned utilisation?
- What do current quotations include and exclude?
- How much cash and contingency does each option require?
- Which process or scheduling dependencies could change the comparison?
- What evidence would justify ownership instead of renting or co-location?
Sources and scope
This is an edited adaptation of the author’s full Substack essay, first published 2026-04-07. Examples and editorial frameworks are distinguished from measured evidence. Source checks and editing: 16 October 2026.
- GFI, techno-economic insights
- NIZO, BFF project announcement, January 2025
- NIZO, future-site kick-off, June 2025
Read the original Substack version. Read the FoodEdge editorial standards or suggest a correction.
Cover photograph: Photo by Daniel Vogel / Unsplash. Context image, not a photograph of a company or experiment discussed here. Photo licence.
