Manufacturing Strategies for GLP-1s, Peptides, and High-Demand Biologics

Introduction: 

Capacity Is Not a Number

Few product classes have humbled operational forecasting the way incretin therapies have. Demand curves that boards approved as aggressive were overtaken within quarters. Sites built for one modality found themselves supplying another. And through it all, the same question was asked in the same unhelpful form: how many kilograms can we make?

The kilogram is the wrong unit, because the kilogram is not the product. A filled, assembled, serialised and distributed presentation is the product, and each step in getting there is a separate industrial asset with its own lead time, its own capital profile and its own resale value if demand disappoints. Treating them as a single pooled number called capacity is how companies end up with an active ingredient surplus they cannot fill and a filling line they cannot supply.

For American manufacturers the stakes have risen sharply in the past year. An oral GLP-1 for chronic weight management reached the US market in January 2026, converting what had been an injectable supply problem into something structurally different. Meanwhile core patents on the leading peptide began falling in several large economies while US protection runs on into the next decade. Both developments reward manufacturing strategies built for optionality, and punish those built on a single point forecast.

Four Scarcities, Not One: Decomposing GLP-1 and Peptide Capacity

Exhibit 1 separates the chain into its four real constraints. What matters to a capital committee is not that these steps exist, but that their lead times differ by a factor of four and their fungibility differs even more.

Exhibit 1.  Lead time and asset fungibility across the four stages of peptide supply.

Active ingredient synthesis carries the longest clock and the least flexibility. A dedicated peptide plant, with its solvent handling, chromatography suites and waste infrastructure, is not a facility that can be turned to another purpose in a downturn. Sterile filling sits in the middle: genuinely scarce, but at least transferable across products with qualification work. Device and final assembly is the quiet problem. Injector components, cartridges and automated assembly lines are product-specific, tooled to a single design, and among the least recoverable investments in the chain.

The strategic implication is uncomfortable for anyone who likes a single capacity number. The binding constraint is whichever stage is shortest at that moment, and it moves. Several manufacturers spent 2023 and 2024 constrained by fill and assembly rather than by chemistry, then found the constraint migrating upstream as filling capacity came online. Capital allocated against last year's bottleneck buys very little.

The kilogram is not the product. The last mile of the chain is the least fungible asset in it, and the easiest to strand.

The Oral Dose Changes the Arithmetic of Peptide Demand

The single most consequential fact in peptide manufacturing planning right now is a matter of pharmacokinetics rather than economics. Oral peptide delivery is inefficient by design: only a small fraction of an ingested dose is absorbed, so the delivered dose must be very much larger than its injectable equivalent to achieve comparable exposure.

Exhibit 2.  Active ingredient consumed per patient-year across injectable peptide, antibody and oral peptide presentations.

The arithmetic is stark. A weekly injectable maintained at 2.4 mg consumes roughly 0.13 grams of active ingredient per patient-year. A once-daily 25 mg tablet consumes about 9.1 grams. That is a ratio of roughly seventy-three to one, before accounting for titration or the yield losses inherent in synthesis and purification. One million patients on the oral presentation therefore require something on the order of nine metric tonnes of peptide a year — more active ingredient mass than a comparable antibody franchise of the same patient count.

For a chief operating officer this reframes an oral launch entirely. It is not a line extension serviced from existing chemistry. Depending on conversion rates, it is potentially a demand event larger than the entire injectable franchise that preceded it, landing on the stage of the chain with the longest lead time and the least flexibility. Offsetting this, the oral presentation removes the sterile fill and device assembly constraints altogether and substitutes solid oral dose capacity, which is abundant, cheap and highly fungible. The bottleneck does not disappear; it relocates decisively upstream.

Choosing a Peptide Manufacturing Route That Survives Commercial Scale

Route selection for peptides is usually treated as a chemistry decision made early and revisited reluctantly. At the volumes now in play it is a capital structure decision. Solid-phase synthesis is fast to develop, well understood and flexible across sequences, which is why it dominates clinical supply. Its weakness is that its cost and material consumption scale close to linearly with volume; there is limited economy of scale to grow into.

Hybrid approaches, coupling solid-phase fragments in solution, and recombinant expression of the peptide backbone followed by chemical modification, both invert that profile. They carry higher development cost, longer timelines and greater process risk, but materially better unit economics at scale. The decision therefore turns on a forecast: at what annual volume does the crossover occur, and how confident is the organisation in reaching it? Companies that lock in solid-phase chemistry for commercial supply because it was convenient in Phase II are, in effect, taking a bet that volumes will stay modest.

The Solvent Bill: Process Mass Intensity as a US Siting Question

Peptide synthesis is extraordinarily material-hungry, and the industry has quantified how much. Process mass intensity — kilograms of total material consumed per kilogram of active ingredient — sits in the range of 168 to 308 for a typical small molecule and averages around 8,300 for an antibody. Solid-phase peptide synthesis averages roughly 13,000, and published assessments put GLP-1 class peptides higher still.

Exhibit 3.  Process mass intensity benchmarks by modality, on a logarithmic scale.

Solvents account for the overwhelming majority of that mass, and the workhorse solvents of peptide chemistry are precisely the ones under regulatory pressure, having been designated substances of very high concern in Europe with parallel scrutiny in the United States. Combine the intensity figure with the oral demand arithmetic and the consequence becomes concrete: a franchise consuming nine tonnes of peptide annually implies well over a hundred thousand tonnes of process mass moving through a plant each year.

That is no longer an environmental reporting line. It is a siting constraint. Solvent recovery capacity, wastewater permitting, air permits and local utility supply now sit on the critical path of a US peptide expansion alongside cleanroom construction, and in several states they are the longer of the two. Greener solvent systems, in-situ deprotection strategies and chemo-enzymatic coupling are usually presented as sustainability initiatives. They are more accurately understood as permitting strategy and as unit cost reduction.

Case in Point: Staging Peptide Capacity Against an Unforecastable Demand Curve

The following is an illustrative composite reflecting how several US programmes have approached this problem; the figures are indexed and directional rather than drawn from any single company's disclosures.

A US specialty manufacturer with a late-stage peptide faced a demand range its own commercial team could not narrow: an upside scenario roughly triple the downside by year six, with an oral formulation in development that would multiply active ingredient requirements if it succeeded. The instinct was to build to the base case. Instead, the capacity plan was split into three tranches, only the first of which was unconditional.

Exhibit 4.  Staged capacity commitment against upside, base and downside demand scenarios.

Tranche one covered committed volume with modest headroom. Tranche two was pre-engineered, permitted and long-lead-item ordered, but construction was released only on a defined commercial trigger — in this case an oral formulation filing. Tranche three was never built at all: it was purchased as a reservation right on external capacity, at a fee that bought priority access without capital exposure. Separately, device assembly was dual-sourced from the outset despite the cost penalty, on the reasoning that it was the least fungible asset and the one most likely to strand.

The programme did not optimise for the base case, and in the base case it looks marginally expensive. What it did was cap the downside loss and preserve the upside, which is the correct objective when the forecast error is larger than the margin. Permitting ahead of the construction decision is the underappreciated move here: it converts a three-year response into a one-year response for the cost of engineering work that has value in every scenario.

The Forty-Amino-Acid Line: Why GLP-1 Franchises Erode Like Small Molecules

There is a regulatory boundary that shapes peptide manufacturing economics and that many strategy discussions get wrong. The FDA defines a protein as an amino acid polymer of more than forty residues. Above that line a product is a biologic, licensed under the Public Health Service Act, and copies must come through the biosimilar pathway. At or below it, the molecule is a drug, approved under a new drug application, and copies arrive through the abbreviated pathway used for generics.

Semaglutide has 31 residues. Tirzepatide has 39. Both sit below the line. The commercial consequence is that when exclusivity lapses these franchises will not face the slow, expensive, clinically burdensome entry that biologics leadership has grown accustomed to. They will face generic entry: cheaper to develop by orders of magnitude, faster to approve, and substitutable at the pharmacy counter. That divergence is already visible internationally, where core patents began expiring across several large markets during 2026 while US protection extends into the next decade.

For manufacturing strategy the implication is direct. Peptide assets should be depreciated and staged against a small-molecule erosion curve, not a biologics one. Cost position at scale, not capacity alone, is what determines whether a plant is still economic on the far side of exclusivity — which returns the argument to route selection and process mass intensity, and makes both of them board-level questions rather than technical ones.

Conclusion: Manufacturing Strategy Is Commercial Strategy

The GLP-1 era has made a point that the industry has long acknowledged rhetorically and rarely funded seriously: for high-demand peptides and biologics, the manufacturing plan is the commercial plan. The ability to supply an upside scenario is worth more than the margin saved by building precisely to a forecast, and the cost of stranded capacity in a downside is concentrated in exactly the assets — dedicated chemistry, tooled assembly — that are hardest to sell.

Three questions separate the plans that survive contact with reality from those that do not. Which of the four scarcities is currently binding, and which will be binding in three years? What happens to active ingredient demand if the oral presentation succeeds beyond expectation? And is the cost position at scale good enough to keep the asset running once generic entry arrives on a small-molecule timetable rather than a biologics one?

None of those is answerable by a capacity number. All of them are answerable by a staged plan with explicit triggers, permitting done in advance, and a clear view of which assets can be recovered and which cannot. That is a harder document to write than a capital request. It is also the one that determines whether the franchise is still profitable in 2032.