A Misfit CDMO Once Meant a Difficult Technology Transfer. Now It Can Cost the Production Window.

Supplier selection frameworks tend to outlive the market conditions that produced them. Most of the criteria still in circulation — capacity, quality systems, geography, price — were assembled when peptide manufacturing had slack in it. Choosing the wrong partner in that market was expensive and slow, but recoverable: the programme absorbed a difficult technology transfer and moved on.

That is no longer reliably the shape of the risk. Each of the six pitfalls below is a question that does get asked during diligence. It simply gets asked at a level of generality that a capable supplier can answer honestly without the answer being informative.

A capable CDMO can answer every generic question honestly and still be the wrong partner for the programme.

What the Data Shows

• Commercial-scale expansion remains concentrated. The largest publicly announced expansions tracked by CDMO Hub sit with a small number of parent groups, and the 2026 transaction wave has reduced the independent field further.
• Capacity does not transfer between manufacturing routes. SPPS, LPPS, hybrid convergent and recombinant-plus-synthetic architectures are not interchangeable, so an announced tonnage figure is uninformative until the route behind it is stated.
• Purification and isolation may set the actual output ceiling. For many complex or high-volume peptides, column inventory, pass count and lyophilisation throughput govern deliverable volume rather than synthesis capacity.
• Availability has to be assessed across API, fill-finish and device assembly. For injectable programmes the second and third of those can be harder to secure than the first, and they carry longer qualification paths.

Why the Cost of a Misfit CDMO Has Changed

As Issue #2 of this series set out, the number of organisations capable of supplying qualified commercial-scale peptide API is small, and 2026 has narrowed it further. CordenPharma's acquisition of AmbioPharm, announced in May 2026, completed on 3 August 2026. PolyPeptide is the subject of a CHF 1.46 billion public tender offer from Samsung Biologics, expected to complete toward the end of 2026; the transaction has not yet closed. GLP-1-class demand has taken up much of the available slack at top-tier sites, and the March 2026 expiry of semaglutide compound patents in markets including India and China has brought a growing field of generic manufacturers toward the same limited pool of qualified peptide suppliers. Protection continues into the early 2030s in markets including the United States and much of Europe, so the effect is regional rather than global.

Where capacity is contested, a misfit CDMO does not simply mean a hard technology transfer. It can mean waiting for the next available production window — and discovering that the alternative supplier who could have absorbed the volume has since booked it to somebody else. The six pitfalls that follow are not diligence failures in the ordinary sense. They are questions asked at the wrong resolution.

Watch Point 01 · Capacity Headlines Describe Volume, Not Route

Announced kilogram or tonne figures rarely specify the manufacturing route behind them, and capacity does not transfer across routes. Four architectures are industrially relevant: solid-phase synthesis (SPPS), liquid-phase synthesis (LPPS), hybrid convergent approaches, and recombinant expression finished by synthetic steps. A reactor train configured for one is not available to another. A fermentation suite is of no use to a molecule containing non-natural residues; solution-phase volume does not substitute for resin-bed capacity.

The two dominant incretin molecules illustrate the point. Published semaglutide process descriptions include a hybrid route combining a recombinantly expressed peptide backbone with synthetically produced fragments containing non-proteinogenic residues, the terminal extension requiring synthesis because it contains Aib. Tirzepatide has a published kilogram-scale GMP process combining SPPS and LPPS. Same therapeutic class, materially different manufacturing architectures, and no interchangeability of the assets that make them. Announced expansions increasingly reflect this: Neuland's Hyderabad peptide block, for instance, is described in terms of separate SPPS and LPPS volumes with different scale ceilings, which is the more informative way to publish a number.

What settles it

Ask what the announced figure refers to: which route, which reactor configuration, and whether the number is nameplate design capacity or capacity already qualified and in GMP use. Then ask what proportion of it is contractually committed, and until when.

Watch Point 02 · Synthesis Capacity Is Not Purification Capacity

For many complex or high-volume peptides, assembly is not the binding constraint — purification is. Crude peptide leaving SPPS carries deletion and truncation sequences, D-isomers, oxidation and deamidation variants, and separating the target from that mixture to pharmaceutical purity is where cycle time and cost accumulate. Preparative reversed-phase chromatography is typically the governing unit operation, and its throughput is set by column inventory and diameter, the number of passes a given sequence requires, acetonitrile supply and recovery, and — at the tail end — lyophilisation capacity, among the slowest and most energy-intensive steps in the sequence.

This is the constraint that Issue #3 examined from the solvent side, and it produces a specific and under-asked question. Purification assets are frequently campaign-scheduled and, for cross-contamination control, often dedicated by molecule or product family. A site can therefore be entirely capable of building your molecule and unable to purify it at the volume you need, in the window you need it — without either statement being untrue.

What settles it

Ask for prep-HPLC column count and diameter, expected passes and cycle time per batch for a sequence of your length and complexity, lyophiliser capacity and campaign scheduling, and whether purification trains are dedicated or shared. Ask what happens to the schedule if first-pass purity comes in low.

Watch Point 03 · The Audit Stops One Tier Too Early

Sponsors audit the CDMO thoroughly and its input suppliers hardly at all. Protected amino acids, resins and coupling reagents carry their own concentration risk, and it is not the same concentration risk as the CDMO tier. Sector analyses place a substantial share of global Fmoc-protected amino acid production with a small number of Chinese producers. Specialised building blocks — non-natural residues, fatty-acid side-chain intermediates, custom linkers — are frequently single-source by construction, because the volumes do not support a second qualified maker.

The consequence is that a fully qualified CDMO in a preferred jurisdiction can still leave a programme exposed to a single upstream supplier, and to the policy environment surrounding that supplier. Reported lead-time extension on critical building blocks during the GLP-1 ramp is the visible symptom; the structural point is that resilience assessed at the CDMO tier says very little about resilience one tier up.

What settles it

Ask for the bill of materials for your process with named suppliers and single-source items flagged, the dual-qualification status of each critical building block, safety-stock policy and current coverage in weeks, and which raw material changes would trigger a regulatory variation on your filing.

Watch Point 04 · Drug Product and Device Capacity Gets Secured Too Late

API supply and finished product supply are separately contested and separately booked, and for injectable programmes the second can be harder to secure than the first. Sterile fill-finish capacity is widely described as a practical ceiling on injectable supply: new sterile lines take years to design, build and qualify, and filling equipment itself carries long order lead times. Device capacity compounds it — pens and autoinjectors are regulated combination products whose assembly lines require their own approvals and quality oversight, with reported equipment lead times for final assembly of eighteen to twenty-four months and fill-finish lead times exceeding a year at peak demand.

The merchant market has also thinned. Novo Holdings' acquisition of Catalent, and the onward transfer of three fill-finish sites to Novo Nordisk, moved a substantial block of independent capacity out of general availability. Where large providers prioritise high-volume programmes, smaller clinical, orphan and rare-disease runs are the ones that struggle to place. Securing drug substance secures one of three sequential dependencies; the other two commonly have longer lead times and fewer alternatives.

What settles it

Run drug substance, fill-finish and device assembly as three parallel conversations from the outset, with a single timeline that shows the qualification path for each. Ask each provider for earliest available slot rather than nominal capacity — availability, not capability, is the constraint.

Watch Point 05 · Regulatory Strength Is Assessed Generically

Inspection history, certificate counts and audit outcomes are table stakes. They establish that a site operates a functioning quality system; they say nothing about whether it can defend a peptide filing. The relevant capability is molecule-specific, and in one important pathway the regulatory framework is unusually explicit about what it requires.

The FDA's 2021 guidance applies to abbreviated applications for certain highly purified synthetic peptide products that reference specified listed drugs of rDNA origin — not to peptide filings generally. For the products it covers, peptide-related impurities present at 0.10% or greater should be identified and characterised; new impurities between 0.10% and 0.5% require appropriate characterisation and justification, including assessment of comparative immunogenicity risk; and new impurities above 0.5% of the drug substance are not acceptable for that ANDA pathway.

These expectations sit alongside the ordinary ICH thresholds rather than replacing them. In practice that means deletion and insertion sequences, D-amino acid isomers, oxidation and deamidation variants each require demonstrated analytical capability and a justified specification, with orthogonal methods to guard against co-elution. European expectations under the pharmacopoeial framework are related but not identically harmonised, which matters where target markets differ.

What settles it

Ask for review history rather than certificate counts: named programmes where this site's peptide impurity control strategy has been reviewed and accepted, in which markets, and what deficiency questions were raised and how they were answered. Ask to see the analytical package — chiral methods for D-isomer control, orthogonal purity methods, and the basis on which each specified impurity limit was justified.

Watch Point 06 · Quotes Are Modelled at Pilot Scale

Price per gram at pilot scale is close to uninformative about commercial cost of goods. Landed cost is determined by step yield and overall yield at scale, purification recovery, cycle time, batches achievable per year, and solvent consumption — none of which are reliably predicted by a development-scale run. Raw materials are commonly cited as the dominant cost component in peptide manufacture, so a small change in overall yield propagates directly into the input bill rather than being absorbed by overhead.

The quoted figure also understates the relationship. Neuland Laboratories estimates that an RFP price may represent only 60–70% of the true cost of an engagement once technology transfer fees, analytical development, change orders and stability storage are included. Both effects run the same direction, and both are visible during selection if the commercial case is modelled then rather than after transfer.

What settles it

Require a commercial-scale cost model as part of the bid: assumed step and overall yield, purification recovery, cycle time per batch, batches per year, and the solvent consumption behind them. Then ask for the same model at half and at double your forecast volume. The sensitivity is more revealing than the headline number.

The Six Watch Points at a Glance

The common structure

Each pitfall is the same error in a different place: a question answered at the level of the supplier's capability when the answer that matters is at the level of your molecule, your volume and your window.

Why This Matters for Peptide Buyers

1. Announced capacity is not bookable capacity. A tonnage figure describes design intent. What determines whether a programme can be placed is the route the capacity is configured for, how much of it is qualified rather than nameplate, and how much is already committed and until when.
2. The constraint often sits downstream of synthesis. A site can be entirely capable of assembling a molecule and unable to purify and isolate it at the volume and in the window required. Purification assets are campaign-scheduled and frequently dedicated, and none of that is visible in a capacity headline.
3. Drug substance is one of three sequential dependencies. Sterile fill-finish and device assembly commonly carry longer lead times and fewer alternatives than API, and the merchant fill-finish market has thinned through consolidation. Securing API alone secures the shortest of the three paths.
4. Selection has become a reservation decision as much as a technical one. Technical fit determines whether a supplier can make the molecule. Timing determines whether they will be able to make it when it is needed — and in a constrained market, the second question is the one that decides programmes.

The CDMO Hub View: The Correction Is Timing, Not Diligence

The instinct on reading a list like this is to lengthen the questionnaire. That is the wrong correction. The six questions above are not difficult to ask, and most sourcing teams would recognise all of them. What has changed is that the market no longer reliably offers a quick second attempt, so the questions have to be asked earlier — before a shortlist exists, and before the need is visible.

Requalification timing is where this becomes concrete, and it deserves precision rather than a single headline figure. Qualifying a new peptide API supplier — assessment, site audit, technical evaluation, quality agreement and initial technology transfer — is typically quoted by CDMOs at six to twelve months. That is the qualification step alone. For a commercial product, the full second-source path adds process validation batches, stability data and a regulatory variation requiring approval in each target market, which is how the commonly cited eighteen-to-twenty-four month figure is reached. Either number leads to the same conclusion: the work has to begin well before the need appears in a forecast.

What to Request in Writing Before You Shortlist

• Route-resolved capacity. Qualified GMP capacity by route (SPPS / LPPS / hybrid / recombinant), separated from nameplate, with committed proportion and horizon.
• Purification profile for your sequence. Column inventory and diameter, expected passes, cycle time, lyophiliser capacity, dedicated or shared.
• Bill of materials with single-source flags. Named suppliers, dual-qualification status, safety-stock coverage in weeks, and which changes trigger a variation.
• Earliest available slot — for drug substance, fill-finish and device assembly separately, with the qualification path for each.
• Peptide-specific review history. Programmes where the impurity control strategy was accepted, in which markets, and the deficiency questions raised.
• Commercial-scale cost model with stated yield, cycle time and solvent assumptions — plus the same model at half and double forecast volume.
• Second-source timeline decomposed into qualification, validation, stability and regulatory approval, market by market.

A Note on Scope

This briefing concerns the selection and qualification of contract manufacturers for peptide APIs and finished peptide products. It is a framework issue rather than a supplier assessment: companies named are cited as illustrations of a manufacturing or market structure point, and no ranking, endorsement or criticism of any named organisation is intended or implied.

Where figures originate from a single commercial estimate, a technology vendor or a company's own published guidance rather than an independent or primary source, this is indicated at first use. Lead times, capacity positions and availability vary considerably by molecule, scale and region, and should be verified directly with suppliers for any specific programme. Transaction status is stated as at the date of publication.

CDMO Hub Insight

The six pitfalls share a structure: each is a question answered at the level of the supplier's capability when the answer that matters is at the level of the molecule, the volume and the window. That is why a longer questionnaire does not fix them.

The correction is not a more thorough process but an earlier one. Diligence that is technically correct but chronologically late can produce the same outcome as diligence that was never done: by the time the requirement is visible in a forecast, the capacity may already be allocated.

Looking for peptide manufacturing partners? CDMO Hub helps pharmaceutical and biotech companies identify and evaluate CDMOs, CROs and specialist manufacturing partners by route, scale, capabilities, facilities and relevant regulatory history. Explore verified suppliers and peptide manufacturing capabilities.

Inside Peptide Manufacturing is a CDMO Hub series on the industrial side of the peptide category — capacity, chemistry, cost of goods, and the decisions that sit behind them.

Notes and Sources

1. Concentration of commercial-scale expansion: CDMO Hub Peptide Facility Tracker (2024–2026); Inside Peptide Manufacturing Issues #2 and #3.
2. CordenPharma–AmbioPharm: CordenPharma announcement of completion, 3 August 2026 (transaction announced May 2026).
3. Samsung Biologics–PolyPeptide: Samsung Biologics announcement of a public tender offer for PolyPeptide Group, valued at approximately CHF 1.46 billion, and Q2 2026 update. Completion expected toward the end of 2026; not closed as at publication.
4. Semaglutide patent expiry by jurisdiction: IQVIA overview of semaglutide compound patent expiry, including expiry in markets such as India and China in March 2026 and continued protection into the early 2030s in the United States and much of Europe.
5. Semaglutide route: published process descriptions and techno-economic modelling describing a hybrid recombinant-plus-synthetic architecture. Published descriptions demonstrate a feasible route; they do not establish the process used across current commercial supply.
6. Tirzepatide route: “Kilogram-Scale GMP Manufacture of Tirzepatide Using a Hybrid SPPS/LPPS Approach with Continuous Manufacturing,” Org. Process Res. Dev. 2021, 25, 1628–1636.
7. Route-resolved capacity disclosure: Neuland Laboratories disclosures describing separate SPPS and LPPS capacity at Hyderabad.
8. Fmoc-protected amino acid supply: sector analyses of supply concentration and reported lead times during the GLP-1 ramp.
9. Fill-finish and device lead times: industry reporting on sterile fill-finish and autoinjector capacity, including reported final-assembly equipment lead times of 18–24 months.
10. Merchant fill-finish capacity: Novo Holdings' acquisition of Catalent and the onward transfer of three fill-finish sites to Novo Nordisk.
11. Peptide impurity expectations: US FDA, ANDAs for Certain Highly Purified Synthetic Peptide Drug Products That Refer to Listed Drugs of rDNA Origin (final guidance, 2021; docket FDA-2017-D-5767), read with ICH Q3A/Q3B. The guidance applies to certain synthetic peptide products referencing specified rDNA-origin listed drugs.
12. Cost of the engagement: Neuland Laboratories estimate that an RFP price may represent 60–70% of total engagement cost. Single company estimate.
13. Supplier qualification and second-source timelines: CDMO published guidance on API supplier qualification timelines, and industry commentary on validation, stability and regulatory variation requirements for a commercial second source.

Figures and characterisations are current as at August 2026. Where a claim rests on a single commercial estimate it is attributed as such at first use. Corrections are welcomed and will be carried in the following issue.