Every water utility on earth still validates its drinking water roughly the way it did in 1887: take a sample, smear it across a petri dish, and wait about 72 hours for the bacteria to grow numerous enough to count by eye. Which means the “real-time” dashboard your local utility is so proud of is, in plain terms, reporting on what died in the water last Tuesday. That three-day blind spot is one of the most expensive habits in the whole water industry, and killing it is the entire premise of real-time microbial monitoring. On episode 21 of season 13, I sat down with Lorenzo Falzarano, the founder of Orb, who builds inline sensors that count microbes in seconds instead of days. His framing is blunt: if artificial intelligence is the water sector’s shiny new brain, somebody still has to build the optic nerve. So here is why the eyes have been missing for 140 years, what running blind actually costs, and why the money is finally starting to pay attention.

Why are water testing methods so dangerously outdated?

Start with the uncomfortable fact. The reference method for deciding whether your tap water is safe, the culture plate, was invented in 1887, the same decade the motorcar started elbowing the horse off the road. You take a sample, you put it in an incubator, and you wait roughly 72 hours for colonies to grow visible enough to count by sight. By the time the answer lands on someone’s desk, that water is long gone, drunk or discharged or pushed into a river. We have spent the last few years building digital twins and bolting artificial intelligence onto everything that holds water, and the whole sophisticated apparatus is still being fed a number that describes the past. As Lorenzo puts it, we are running the global water system on a 3-day delay.

Culture method (the petri-dish test): you take a water sample, place it on a nutrient plate, and let any microbes multiply for one to three days until the colonies are big enough to see and count. Accurate, cheap, and almost 140 years old. Its flaw is time: the result describes the water as it was, not as it is.

The hidden cost of running a water plant blind

Now, running blind is not free. It is just an invisible line item, one I would call the safety tax. Because operators cannot see the biological load in real time, they default to the worst case: overdose the disinfectant, run the ultraviolet lamps at full tilt, hold extra contact time, all just in case. When I dug into the numbers behind this episode, I found a single 100,000 cubic-metre-per-day plant burning roughly $21,000 every single day in excess chemicals alone, which works out to about 6% of its total operating cost, spent purely to maintain a safety buffer against a risk nobody could measure. Run the UV lamps flat-out when the water is already clean and you waste something like 15% of that energy. And because overdosing creates extra sludge, and sludge handling can be 40% of a plant’s operating cost, you end up paying to manufacture waste you then pay again to haul away.

I once described this whole dynamic, in a newsletter, as the reason you walk slower across a dark hotel room: take away the sense you normally use to avoid risk, and you overcorrect on everything else. I wrote about why, in this gold rush, the smart move is to sell shovels.

What happens when nobody finds the source?

And the safety tax gets a lot more expensive when things actually go wrong. Lorenzo drops one statistic that sums up the entire crisis: by Orb’s reckoning, around 70% of water contamination investigations end as cold cases, where the utility never finds the root cause. Why? Because without high-resolution data, you are working a crime scene three days after the evidence washed downstream. Regulators have started pricing in that blindness directly. In 2025 the UK regulator Ofwat hit Thames Water with a £104.5 million penalty over failures across its sewage operations. And when cryptosporidium broke out in Brixham in 2024, the contamination cost the Pennon group £16.3 million and put roughly 16,000 homes under a boil-water notice for weeks. That was not the cost of fixing the damaged valve that let the parasite in. It was the price of not knowing the valve was broken until thousands of households were already drinking the consequences.

What is real-time microbial monitoring, really?

So here is the precise gap Orb is built to fill. At a treatment works today, operators mostly watch two things: turbidity and chlorine. Neither one tells you what is actually alive in the water, and many times there is no correlation between a turbidity reading and the real microbial load. Yet by Lorenzo’s account, the top five of the ten most common water-quality standard failures are all bacteriological, meaning they are about living organisms, the coliform bacteria and E. coli that flag something unsafe has slipped through, rather than the chemistry. Sit with that for a second: the single biggest reason plants fail their standards is the one thing almost nobody monitors live.

Turbidity: how cloudy the water is, measured by how much light it scatters. It is a useful, cheap proxy for “something changed,” but it is a health-blind one, a clear glass of water can still be teeming with bacteria.

How does Orb count microbes in seconds, not days?

So how do you compress three days into three seconds? Instead of growing bugs on a plate, Orb shines light at the water and reads the way microbial cells fluoresce, glowing back a little brighter than the plain water around them, and counts them as they flow past. No reagents, nothing consumed, nothing destroyed, just photons doing the work. The reading comes back in about three seconds instead of three days. Orb’s website pushes a sharper number, claiming the method is roughly 1,700 times faster than the incumbent way of counting microbial pollution, and you can take the exact multiplier as the marketing line it is, because the part that actually matters is the change of category, from days to seconds. And the sensor is only half of it: the signal feeds an intelligence layer that does not just tell an operator their bacteria count jumped, it points at the most likely reason it jumped. Making all that data legible to the people actually running a plant is its own hard problem.

Fluorescence sensing: certain molecules absorb light at one wavelength and re-emit it at another. Living microbial cells, rich in proteins and amino acids, fluoresce in a tell-tale way, so an optical sensor can count them in flowing water without any chemical reagents, in real time, instead of waiting for a culture to grow.

Lorenzo is candid that this means building hardware, which is exactly the thing most investors would rather avoid. But you cannot have a smart-water revolution without it. I have argued before that measuring water quality is, awkwardly, a hardware problem dressed up as software.

Orb is not chasing the petri dish alone

And no, reassuringly, this is not one company shouting into the void. The whole frontier is racing the same near-140-year-old plate, which is the best possible sign that the problem is real. On my own podcast I heard the identical complaint from a completely different angle when Sam Dukan walked me through Diamidex, whose MICA system delivers an E. coli result in 6 hours instead of 24, and a Legionella result in 2 days instead of the usual 10, while matching the standard method exactly. Different physics, same enemy: the clock.

There is a small ecosystem forming here, and I have hosted more of it than I expected. FREDsense, for instance, is chasing the same goal with an electrochemical biosensor.

Is real-time water monitoring actually a market yet?

This is where it gets genuinely interesting for anyone who deploys capital, and where my day job collides with my podcast. When I ran the numbers across my own Leviathan database, the entire peer set of real-time water-quality biosensing companies, the ones genuinely doing what Orb and Diamidex do, has raised only about $97 million across 54 rounds, ever. No single round in that group tops roughly $12 million. To put that in perspective, that is the whole category raising, across its entire history, less than a single mid-size Series B in almost any other deep-tech vertical. This is not a crowded gold rush. It is a frontier that has barely been funded at all, which is either a warning or an invitation depending on your temperament.

The shape of that line matters more than the total. For years the category raised single-digit millions and looked like a science project. Then 2024 and 2025 happened, and last year alone the peer set pulled in $34.4 million across 11 rounds, its busiest and best-funded year ever. The money is, finally, starting to notice the eyes are missing. I track where water-tech capital actually goes in my state-of-funding work. If you want the running version of this, I send it out in my newsletter.

Water reuse is ending water’s three-day grace period

So why now, after 140 years of the industry shrugging at the three-day lag? Because the future of water is reuse, and reuse quietly demolishes the old excuse. You cannot recycle water through faster and faster treatment cycles while waiting three days to confirm a batch is safe. By the time your culture result comes back, you are storing three-day-old water in a tank, which, as Lorenzo points out, is arguably worse than the water you started with. The moment treatment cycles get short enough, real-time validation stops being a nice-to-have and becomes the gate that the whole process has to pass through.

Frequently asked questions

How is drinking water tested for bacteria?

The standard method is the culture, or petri-dish, test: a water sample is placed on a nutrient plate and any microbes are left to multiply for one to three days until the colonies are large enough to count by eye. It is accurate and cheap, but slow, the result describes the water as it was up to 72 hours earlier.

How long does water quality testing take?

For the classic culture method, one to three days. Faster lab techniques have cut that sharply, Diamidex returns an E. coli result in about 6 hours, and real-time inline sensors like Orb’s read microbial signals in seconds. The trade-off has historically been between speed, cost, and how closely a method matches the regulatory gold standard.

What is real-time microbial water monitoring?

It is the continuous measurement of the living, microbial load in water using inline sensors, rather than sending samples to a lab. Reagent-free optical sensors count microbes as the water flows past and feed the readings to software that flags trends and, increasingly, the likely root cause of a contamination event.

Why is real-time water quality monitoring important?

Because the three-day lag is expensive in two directions. It forces operators to overdose chemicals and energy as a precaution (the “safety tax”), and when contamination does happen, roughly 70% of investigations never find the source. As water reuse spreads, instant validation of each batch becomes unavoidable.

What is coliform bacteria in water?

Coliforms are a broad family of bacteria used as an indicator of water quality: most are harmless, but their presence signals that conditions could allow harmful organisms through, and some, like E. coli, point to faecal contamination. Coliform failures are among the most common reasons a treatment works breaches its standards.

Is real-time water monitoring a good investment?

It is early and thinly funded. By my count across the Leviathan database, the real-time water-quality biosensing peer set has raised only about $97 million across 54 rounds in total, with no single round above roughly $12 million, but 2025 was its biggest year yet. Small market, fragmented field, accelerating demand: make of that what you will.

The eyes are finally being built

For 140 years the water industry got away with testing for life by waiting for it to grow. Reuse, regulation, and the simple cost of running blind are quietly ending that grace period, and the companies building water’s missing eyes are still cheap, few, and early. Whether Orb is the one that wins is an open question. That the eyes are finally being built is not. Go listen to my full conversation with Lorenzo Falzarano, it is worth your hour.

Sources

1. Ofwat, “Enforcement case in Thames Water’s management of its sewage treatment works and sewerage networks” (penalty announced 28 May 2025). ofwat.gov.uk
2. ITV News West Country / Pennon Group, “South West Water lost more than £16mn to Brixham parasite outbreak” (27 Nov 2024). itv.com
3. (don’t) Waste Water podcast, S13E21 (Orb, Lorenzo Falzarano) and S12E20 (Diamidex, Sam Dukan) : first-party interviews. Retrieved 2026-06-11.
4. Leviathan water-tech funding database (Antoine Walter) : real-time microbial-monitoring peer-set funding, verified 2026-06-11.

The post Real-Time Water Quality Monitoring vs the 1887 Petri Dish appeared first on (don't) Waste Water.