$46.9 million is the ceiling ARPA-H's own award record puts on ESCALATOR, the agency's bet that messenger RNA can be manufactured on a benchtop instead of in a campus-sized plant. Waterfall Scientific's announcement, published eight days after the award start date, says $54.5 million. Both figures carry the words "up to."
The gap is 16%. For a company built on university research, federal grants and a seed round, that is the kind of difference an investor notices before pricing the science.
What the money buys is a single controlled line: cell-free enzymatic DNA production, continuous-flow RNA synthesis, lipid nanoparticle formulation, fill-finish, and automated lot-release testing. ARPA-H's stated target is an mRNA therapeutic ready to dose within a week of diagnosis. The same sequence takes months today, and the agency puts the cost of a customized genetic therapy in the hundreds of thousands of dollars.
The technical ambition is real, and the commercial constraint is equally concrete. Shrinking the factory reaches the part of the cost stack that automation can compress. Licensing contracts control the larger share, and automation leaves it untouched.
Six organizations, one auditable system
Waterfall is prime contractor. The consortium spans the United States and Canada, and each partner owns a specific unit operation. The architecture depends on the handoffs between them.
| Organization | Role on ESCALATOR | Base |
|---|---|---|
| Waterfall Scientific | Prime contractor; DuoTether continuous-flow RNA synthesis | Worcester, MA |
| Ginkgo Bioworks | Automation hardware, systems integration, control software | Boston / Emeryville |
| Ansa Biotechnologies | Enzymatic, cell-free DNA template production | Emeryville, CA |
| Polymorphic Biosciences | Lipid nanoparticle encapsulation | Vancouver, BC |
| Tensentric | Fill-finish, sterility and device engineering | Boulder, CO |
| Volta Labs | Automation and miniaturization for DNA template production | Boston, MA |
Consortium roles as disclosed by Ginkgo Bioworks, 10 September 2026.
ARPA-H's award page lists the start date as 25 August 2026 and names Inna Shcherbakova as principal investigator. The program announcement on 1 September committed up to $125 million across five teams. GenomeWeb put the split at HDT Bio up to $39 million, Waterfall up to $46.9 million, and the University of Utah up to $9.8 million; HDT Bio's own release on 15 September cited up to $39.5 million. Ginkgo's subcontract is up to $17.5 million.
ARPA-H's federal award record caps ESCALATOR at up to $46.9 million. Waterfall's company release and UMass coverage cite up to $54.5 million. Neither party has published a reconciliation. For diligence, the federal record is the conservative figure.
How the project got here
The lineage starts in 2018, when Craig Martin's RNA chemistry became a case study in Sarah Perry's microfluidics course at UMass Amherst. Martin had spent more than 35 years on RNA synthesis; Perry builds microscale devices that move fluids through channels the width of a human hair. Their collaboration produced the co-tethered synthesis approach Waterfall now calls DuoTether.
Waterfall was founded in 2022, incubated through the UMass Institute for Applied Life Sciences, which was set up in 2014 with more than $150 million from the state and the university. The company completed The Engine's Fall 2025 cohort and set up lab space at the Massachusetts Biomedical Initiatives incubator in Worcester.
The award itself moved quickly. ARPA-H opened the ESCALATOR award on 25 August 2026. It announced the five GIVE teams on 1 September. Waterfall went public with the $54.5 million figure on 2 September. Ginkgo disclosed its $17.5 million subcontract on 10 September and said work would begin with a technical kickoff later that month.
By 13 October, ARPA-H program manager John Schiel was describing GIVE to Fierce Pharma as a "first-shoring" program: building the capability before demand arrives. He described a phased plan, with roughly six months of pen-to-paper design, a two-year Phase 2 demonstration of a non-GMP end-to-end process, and a Phase 3 enclosed system intended for real-world use.
One chronology trap for anyone reading the company's website: Waterfall's flash talk at the International mRNA Health Conference in Berlin took place on 11 November 2025, before the award, and the announcement on Waterfall's site carries no year. The 2026 edition of the conference ran 17–19 November in Boston; the Martin lab's news page listed it among upcoming presentations on 4 November.
The part automation cannot reach
Manufacturing cost and selling price are different quantities, and for mRNA the cleanroom accounts for only a small share of manufacturing cost. A 2026 medRxiv analysis of mRNA-based cell and gene therapy manufacturing estimates licensing and royalty fees at roughly 83% of base-case cost of goods. The base case is $16.8 million per production run at 15 L batch scale across 43 batches, or $56.09 per dose, with a scenario range from $3.68 to $383.22.
Lipid nanoparticle licensing alone accounts for $6.0 million of $6.996 million in license fees in that model, about 86%. Those fees are structured as a percentage of sales, so batch count has little effect on them. The authors conclude that cost-per-dose reduction through volume scale-up "provides limited relief."
The cost study is a medRxiv preprint, scenario-based and not peer reviewed. Treat the 83% share as an order-of-magnitude signal about where the money goes; it is too scenario-dependent to serve as a valuation input.
Continuous-flow manufacturing attacks the other 17%. It can cut footprint, cycle time, labor and the capital tied up in cleanrooms. The licensing structure stays where it is. An investor model that assumes a benchtop line resets the cost of a personalized mRNA therapy by an order of magnitude is assuming away the largest line item.
What is actually being underwritten
Three risks, in order. First, integration: instruments from six organizations must behave as one controlled, auditable system that a trained technician can operate. Ginkgo chief executive Jason Kelly framed the problem plainly, saying distributed manufacturing "has stayed a promise rather than a practice" because that control layer did not exist. Second, regulation: ARPA-H says it will engage the FDA throughout the program, and Schiel has published the phase structure, but no public calendar puts dates on GMP demonstration, regulatory submission or field deployment. Third, licensing: if the platform works, a meaningful share of the economics may accrue to the holders of the LNP and RNA chemistry patents rather than to the integrator.
There is a strategic upside the announcement undersells. As we wrote in November, the FDA's feature-space approach to individualized gene editing already showed how a regulatory template can cut the cost of variant products. If distributed mRNA production becomes the default for personalized oncology and rare disease, value sits in the operating network, the quality data and the regulator-ready control stack. That layer looks like a software-and-services business built on top of hardware.
Waterfall's execution bench is stronger than the average first-time grantee. Martin is president and chief scientific officer. Perry chairs the technical advisory board. Melissa Moore, who ran Moderna's platform research from 2016 to 2023, chairs the board. Shcherbakova previously worked on Moderna's early programs and led RNA platform work at ReNAgade Therapeutics.
The milestone to watch is the first evidence that the four modules run together and produce a lot that passes release testing. Until then, ESCALATOR remains a development contract, and the federal record's $46.9 million is the ceiling with documentary support.
Distributed manufacturing can shorten the wait for a personalized therapy. Licensing and royalty fees will still set the floor under what that therapy costs to make.