CDMO capacity reservation: what a long-term contract costs

Why did capacity run short when demand moved and plants could not?

Because the volume needed per product has grown, and plants take years to build.

Injectables for chronic disease, such as the GLP-1 products, are dosed over long periods and demand quantities of a different order of magnitude. Bioreactors filled up for this reason, and so did sterile filling slots. Facilities cannot be added quickly. From building design through qualification to regulatory inspection, planning to operation takes years.

The Fujifilm Diosynth contract shows the scale. It uses a new large facility at Holly Springs, North Carolina, equipped with single-use bioreactors, and Regeneron says the deal will nearly double its large-scale manufacturing capacity in the United States. One contract is as large as one company's U.S. capacity. The products and volumes have not been disclosed.

Demand moves in years but supply takes several years to grow, and that time gap created both the scramble for slots and the long-term contract.

Why do both sides want the same contract?

Because one contract is the buyer's insurance and the supplier's case for building.

From the buyer's side, a long-term contract is insurance. Reserving first is cheaper than waiting for a slot to open while development stalls. If you start looking for capacity only after approval, you are too late; that is the reasoning.

From the supplier's side, the same contract is the basis for investment. When it decides whether to build a new facility, the amount of confirmed orders is a key input. Samsung Biologics has named neither the customer nor the product for this contract. Production is at Songdo, South Korea. Some reports say eight years, but counting from the start and end dates gives seven and a half. Check the term by its dates.

Whether a supplier's expansion plan will actually be carried out is also important information for the buyer. That is why you should look at the other side's facility plans when you compare in-house with contract manufacturing (make or buy (in Japanese)).

The same contract is the buyer's insurance and the supplier's case for building; read its term from the dates, not the headline.

What happens if you do not use the slots you reserved?

You still pay. A real contract can be read in Vir Biotechnology's filings with the U.S. Securities and Exchange Commission (SEC).

In 2020, Vir signed an agreement with Samsung Biologics to reserve drug substance manufacturing slots for 2021 and 2022. According to the filing, Vir can neither cancel nor adjust the reserved slots, and Samsung will not offer those slots to a third party. Vir committed to pay about $362 million whether or not it uses the slots (take-or-pay), and the amount is adjusted every year by the Korean consumer price index. The portion used is billed batch by batch, and any shortfall is billed at the end of that year.

In a 2020 letter agreement with WuXi Biologics, roughly 70 to 80 percent of the batch manufacturing fees were payable whether or not the capacity was used.

Both are examples from a single company, Vir, but the way to read them applies to other contracts. Reserving capacity can be seen as the buyer taking on part of the supplier's capital-investment risk. If development stops or the demand forecast is wrong, the promise to pay can remain. That cost goes into your cost of goods (in Japanese). The longer you reserve, the higher the chance that the forecast is wrong. A product in phase 3 and a product in phase 1 call for different volumes and different terms.

Reserving capacity for years buys certainty, and the price of that certainty is charged even on the slots you do not use.

How should you read "cumulative orders"?

As a running total of contracts, not as revenue and not as one year's worth.

CDMO announcements stress cumulative orders, and the figure needs careful reading. With this contract, Samsung Biologics' cumulative orders reached about $21.9 billion. Cumulative orders are the total value of the contracts signed so far; they are not that year's sales. The company is reported to count cumulative orders at the amount of its minimum take-or-pay guarantee (MTOP). Before you are impressed by a large number, divide it by the term. $262 million over 7.5 years is roughly $35 million a year.

There is another point: the contents of a contract are often not disclosed. Whether the product is an approved commercial product, or one still in development with a contract that assumes approval, changes how likely the contract is to be fulfilled. If development stops partway, the contract stays but no production takes place.

Divide the headline value by the term and ask about the product's status before you compare two announcements.

Should you keep production at one site or split it?

Decide by what you lose when a site stops.

After reserving capacity, you still have to decide whether to use one site or several. Putting everything at one site keeps technology transfer (in Japanese) work to a minimum. But if that site is stopped by an inspection, supply stops with it. Splitting protects supply, but each site needs its own transfer and verification. Even identical equipment sits in a different building with different water and air handling, so some of the equipment qualification (in Japanese) must be redone.

Scholar Rock is an example where the split paid off. Its resubmission in March 2026 for apitegromab, a candidate for spinal muscular atrophy, listed two fill sites. At an FDA inspection in April, one of them (Catalent Indiana) was classified OAI (official action indicated). The company received that notice in August and removed the site from the application. The review continued with the other site. Had Scholar Rock not listed a second site because of the added cost, the application might have stalled.

Products with no substitute, or products where an interruption of supply directly affects patients, are worth splitting even at a cost. For a product that inventory can cover for a few months, a single site can be justified.

The test for splitting is what you lose when a site stops, not what the second site costs.

What to check in a contract or an announcement

These checks work whether you are in purchasing, business development, QA or sales.

1. Is the term written as a start date and an end date? Are you using a figure such as "8 years" or "10 years" as it appeared in a headline?
2. Can the reserved slots be canceled or adjusted? When is the bill for unused slots issued, and how is it calculated?
3. Is the guaranteed amount or unit price adjusted by an index such as consumer prices? Is the procedure for review written down?
4. Is the product approved or in development? If development stops, what happens to the promise to pay?
5. If the supplier's expansion does not start on schedule, how are the reserved slots treated?
6. If you split production across sites, have you budgeted for when the second site goes on the application, and for the cost of the transfer?

If you are in sales, ask your customer when its contract ends. It tells you when the search for its next slots begins, and when to time a proposal.

Related guides: how to choose a contractor is covered in CDMO versus CMO (in Japanese), the line between in-house and contract manufacturing in make or buy for biologics (in Japanese), and what happens when a plant itself is sold or moved in what changes hands when you divest a manufacturing site (in Japanese).