Biomanufacturing works. Manufacturing is the gap.

One engineered block that grows biomass on its own: tanks, seed train, media and cleaning systems, utilities, instrumentation and controls, built and tested before it ships. Designed once and manufactured repeatedly rather than drawn up per project, which is what takes the cost and the calendar out.

Software that runs the equipment rather than reporting on it. Each module sets feed, gas, temperature, pH and harvest timing against the recipe by itself, with operators supervising. Because every module is identical, a yield gain learned at one site is a software update at the others.

A slab with enough floor load, and access to water, gas and electricity. Everything else arrives with the module.

Rent if you want output without an asset: a subscription holds the capacity, we operate the module, and you pay what each run costs. Buy if volume is certain and you would rather own the steel. Some members do both, proving the process in The Fleet and then taking modules of their own.

Campaigns are billed cost plus, meaning media, labor, utilities and consumables plus a defined margin to the site that runs them. Modules are quoted against the configuration you specify. Send us one and you get a price, a footprint and a schedule.

The ones the configurator carries today: plant cells, mammalian cells, insect cells, yeast, bacteria and filamentous fungi, and others on request. Each loads the tank differently, so the module is configured to the organism rather than the reverse.

Your choice. A module can arrive with DSP integrated or without it, running as an upstream block that hands broth to whatever you already have. Integrated means the standard DSP train, dewatering, drying, milling and packaging sized to the output, so what leaves the module is dry biomass rather than broth. Anything beyond that, extraction, fractionation or purification to a specific spec, comes as an add-on rather than part of the module itself, because the module stays standardized. We quote it separately for you.

Much faster than a plant, because nothing is designed from scratch and the heavy work happens off site while the pad is prepared. Utilities, connection and qualification are still real work, and there is risk in any date.

That is where most of the demand and most of the members are, and where a site and its utilities can be secured fastest. Latin America and India follow, then Europe and the Middle East. Members with volume in a region get a say in when it gets a site.

Because two different things come off a batch, and only one is yours. Machine behavior, how fast the jacket pulls heat out, how oxygen transfer falls as broth thickens, how a pH loop settles, is a property of the equipment, and it is what the control models train on. Your process, the strain, the media, the setpoints, the titre, never enters the shared model.

Your runs stay in your own space, visible only to you and the crew running your campaign. Training happens on de-identified machine signals, and what moves between modules is the trained model, not the data behind it.

It is in the contract as well as the architecture: you own your process data and any improvement specific to it, we own the general control models. A module can also run with learning switched off.

Nothing. Two companies in the same category can run in the same building without touching each other's work: campaigns are separated by module and by schedule, and confidentiality is contractual. What is shared is the equipment design and how Jarvis, our Physical AI, performs, never the product.

Two ways. License a site: you fund and host it, we license the module design, Jarvis, our Physical AI, and the operating standard, and capacity is sold through The Fleet. Or invest in a licensed site: back a site somebody else hosts, which is exposure to capacity in use rather than to one product working.