Most aquaculture systems are built to grow fish and discharge water. Most fertilizer plants are built to manufacture chemistry from gas and rock. This container does neither. It grows catfish, captures their manure as a live ferment, polishes the water, and recirculates it. The fish leave by truck. The ferment leaves in drums.
Two things about this post before the detail. The composition and microbiology of the ferment are laboratory work and they are published in full on the evidence page. Almost everything about the machine that produces it is, today, a company statement rather than a reading. Both facts are stated where they belong below, rather than blended together.
The loop, in the order water moves through it
- Fish tanks. Catfish at mixed biomass classes, juveniles through harvest-ready, in sub-tanks for size sorting.
- Solids capture. Manure and uneaten feed settle into a collection layer. This is the raw material.
- Biofilter. A live microbial bed converts ammonia from fish respiration to nitrite and then nitrate. Standard nitrification, biological rather than chemical.
- Oxygenation. A small injection step holds dissolved oxygen in the band the fish need.
- Return. Polished water goes back to the tanks. The loop closes.
What does not happen matters as much. There is no discharge of nitrogen-rich water to a drain. There is no chlorination or chemical sterilization step, which would kill the community the whole system depends on. The biology that grows the fish is the biology that ends up in the ferment.
Initial fill on the 40ft unit is 30,000 litres, which is Thomas Rosso’s figure on a recorded call and is being verified. Make-up water replaces what leaves as product, and how much that is per day is being metered rather than estimated.
Why catfish
Catfish breathe air, which means they survive handling and power interruptions that would kill trout or bass. They hold high stocking density at 26 to 28 degrees C. Their waste is liquid, which is what a continuous extraction line needs. The company puts feed conversion at about 1.2 kg of feed per kilogram of growth and an all-in cost near USD 2.50 per kilogram of fish; both are company figures, being verified, and neither has a document behind it yet.
Change the feed and you change the ferment. That is a real lever and also a real constraint on batch-to-batch consistency, which is why the finished-product panel described below matters.
What comes out, measured
This is the part with reports behind it. Sample AF4_1 was drawn on 30 June 2026 and analysed by NviroTek Wynland Laboratories in Wellington, South Africa.
The chemistry, report S26/3362, page 1 of 2: total nitrogen 421.5 mg/L, of which 244 mg/L is ammonium; potassium 216.8; phosphorus 16.5; calcium 105.6; magnesium 34.6; sulphur 29.3 mg/L. pH 7.96, electrical conductivity 374 mS/m, total solids 0.12 percent. Arsenic, cadmium, cobalt, chromium, mercury and lead all below the detection limit. Molybdenum at 1,690 ug/L, which is high, with copper below detection. Eleven of the twenty-six analytes sit inside the laboratory’s schedule of accreditation and the evidence page marks each row.
The biology, report M26/9725: total plate count above 30,000,000 cfu/g, which is the method’s reporting ceiling rather than a number, with yeast at 10 and mould at 3,200 cfu/g. All three results are inside the accredited scope.
The sequencing, report M26/9726: 952 bacterial taxa, 516 resolved to species, Shannon index 4.49. Pseudomonas at 19.7 percent of reads, then Aeromonas 9.8, Comamonas 9.1, Flavobacterium 7.9, Acinetobacter 7.8.
A litre carries about 0.42 g of nitrogen against 460 g in a kilogram of urea. This is a biological input, and any comparison that treats it as a nutrient product gets the arithmetic wrong by three orders of magnitude.
What comes out, not yet measured
Output volume is the figure buyers ask about first and it is the one we can least support. The company quotes up to 2,000 litres a day at full fish load. That is a company statement, being verified, and it is not yet settled whether it describes tank liquor or diluted product. Litres drawn per day are being metered on the Abu Dhabi unit and the readings will replace the quoted figure.
Fish output planning uses 20,000 kg a year for a 40ft unit, four harvests of about five tonnes. That is a design figure from the company’s own deck, not a harvest record. Nothing has been metered against it.
Chain of custody on the analysed sample is incomplete. No document ties sample AF4_1 to a specific container, site or batch, and the evidence page prints "Not recorded" against those fields rather than filling them in. Until that chain exists, the panel above describes a sample, not a product specification. A finished-product panel on a retail batch has been ordered.
Energy
Connected load is 3 kW: a 1 kW circulation pump, a 1 kW inverter air conditioner, sensors and controls. That is a company figure and the equipment list does not itself sum to it.
Daily consumption is a different quantity from connected load and depends on duty cycle and ambient temperature. No figure for it publishes here, because no unit has been metered and the equipment list above will not settle it on its own. Consumption is being metered on the Abu Dhabi unit alongside make-up water, and both publish when the readings exist.
Every unit installed so far runs on grid power with generator backup. A solar specification publishes after we have run one that way, not before.
The instrumentation
The container reports continuously: water temperature, dissolved oxygen, pH, ammonia, nitrite, nitrate and conductivity; pump runtime, biofilter performance and oxygen injection rate; feed events, biomass estimates and mortality flags; collection volume, processing tank levels and output batch numbers; energy draw and grid status.
For an operator the value is the alerting. A slow drift in dissolved oxygen surfaces hours before anyone on a daily round would catch it. For an agronomist the value is the batch record: every drum carries a timestamped ID linked to the feed batch, the biomass and the water chemistry it was produced under. That is the mechanism by which a finished-product panel becomes meaningful rather than a one-off snapshot of one sample.
Where the technology came from
The containerized system architecture, the integration of tanks, biofilter, manure capture and recirculation inside a shipping container, came out of development work in South Africa and is licensed from the Hosha group. Containers are built in South Africa and shipped to site. Hosha Vitality operates the deployments and licenses the technology; Rosso and Scanavino Family Farms is the farm-facing brand. The full group table with registered names, jurisdictions and registration numbers is being assembled for the company page, and where a registration number has not been supplied it will say so.
What we would tell an engineer to be sceptical about
- Output volume, until the meter reads.
- Daily energy, until the meter reads.
- Shelf life. The safety data sheet claims 24 months with no stability data behind it. A viable-count series at 3, 6 and 12 months is commissioned.
- Batch consistency. One sample from one batch is what exists. The community is a fermentation mix rather than a formulated inoculant, which makes consistency a question worth asking rather than assuming.
- The startup curve. Fish biomass takes months to reach steady state and ferment output follows biomass. Year one is not the steady-state year.
Next
The laboratory reports, the accreditation marks per analyte, and the interpreting scientist’s six points of attention are on the evidence page. The application regimes and the paired-plot trial protocol are on the trials page. The system hardware is described at the system.
If you want to walk a deployment site or run a trial alongside our biology, apply and we will set up a visit and a soil test.