What is Uravu Labs known for?
Uravu Labs was founded in Bangalore in 2019 by engineers whose campus had run dry in 2016; co-founder and CEO Swapnil Shrivastav was one of five finalists in the $1.75 million Water Abundance XPRIZE, the only Indian team of the five, picked from 98 entrants (WaterWorld covered the 2018 finalist round). That prize asked for 2,000 liters a day at 2 cents a liter on renewable energy, a bar Uravu would chase for the next seven years.
The XPRIZE glory did what glory does: the first customers never came. They tried everything the “water for all” playbook prescribes, CSR budgets, NGOs, community devices, and nobody would carry a new technology’s risk. So they shifted gears, and the reset became the company, in Swapnil’s words: “today we sell the water… we don’t sell the technology”, to customers who already pay bottled-water prices.
We actually tried everything, right? From the day we won XPRIZE, we were like, we’ll solve water for all. We tried to work with CSR departments, NGOs. We, you know, try to make a community-based device, but the people who are not ready to pay for new innovations and the risk and a lot of this money there. Then we started shifting gears.
An atmospheric water generator, to define it before judging it, pulls the moisture sitting in ambient air and turns it into drinking water, by cooling air below its dew point like a dehumidifier or by soaking moisture into a sorbent and releasing it with heat.
How much does atmospheric water generation cost?
Swapnil Shrivastav’s levelized cost curve for Uravu Labs, stated on my microphone in August 2025, runs from 10 to 12 cents a liter five years ago to 5 cents two years ago and 2 cents today, computed with grid electricity at 12 cents per kilowatt-hour, and 1 cent where the heat comes free. Levelized simply means the machine’s whole life is in the number, capital included.
Christopher Gasson, the publisher of Global Water Intelligence, famously called atmospheric water generation an “egregious waste of money”, and the numbers backed him, with desalination at 30 to 40% of its thermodynamic optimum and AWG near 5%. Swapnil doesn’t argue with the ceiling (water from air, he says, will not beat desalination for 10 or 15 years), so the question worth asking becomes “cheap enough for whom”.
And the curve explains the strategy: the drop from 12 cents to 2 came from engineering, while the drop from 2 to 1 comes from somebody else’s heat. Once you see that, you stop asking where water is expensive and start asking where heat is free.
The strategy: stop looking at water as just water
Uravu Labs runs two tracks at once, and the split is the strategy. Track one keeps the lights on: hospitality and beverage customers in Bangalore, an Italian-style microbrewery among the first believers, roughly $1 million in annual recurring revenue from a segment Swapnil honestly caps at about $10 million, served by a factory that pays itself back in under 24 months. He calls it the Tesla strategy (premium niches first while the technology matures), and over 80% of every liter made in two years was actually sold.
Track two is the bet, and he framed it with the line that gave this article its spine:
Stop looking at water as just water. Probably look at some of the allied applications along with water, which actually makes the whole solution better, right? Because see, water again doesn’t have to be thought of in silo.
In practice that means hunting two or three markets whose own economics carry the water along, data centers and climate-controlled agriculture are the two he names, with a 2-meter cube that fits a standard 20-foot shipping container as the deployable unit. I covered where that discipline comes from in my newsletter on seed-strapping: too much cash, in his words, makes you solve problems by throwing money.
Do data centers really need water from air?
Data centers are the flagship application Uravu Labs is hunting, and the sizing goes like this: in 2025, Global Water Intelligence estimated the water CAPEX of data centers at $569 million for the year, cooling taking about 80% of it, while the International Energy Agency’s 2025 Energy and AI report projects data centers consuming around 945 terawatt-hours of electricity by 2030. Nearly all of that leaves the building as heat, and against that torrent, the projection I walked through in the episode editorial puts reused data-center heat at about 3.5 terawatt-hours a year by 2030. In the editorial I called it what it looks like: “not a gap, it’s a canyon”.
The 1 MW arithmetic
The expensive step of desiccant-based water-from-air is the heat that drives the water back out of the salts, and a data center manufactures exactly that all day long. By Uravu’s projections, feed the system a 45°C stream and you drop it to 30°C for each megawatt of waste heat used, collecting about 20 cubic meters of water a day. I fact-checked that claim in the episode: it works out to 0.83 liters per kilowatt-hour of heat, within the envelope of peer-reviewed efficiency figures, on the efficient end. And the output is distilled, which happens to be exactly what a cooling loop wants to drink (scaling and fouling, the two words every cooling engineer mutters in their sleep). I dissected the full thesis in my newsletter edition “VCs vs. Physics: Who Wins?”, and if you enjoy this kind of rabbit hole, that’s the one to subscribe to.
Where could the strategy break?
Swapnil Shrivastav supplies his own counterweight, and he does it better than the skeptics: asked when a data center actually runs on this, he answered that any integration at scale is realistically two years away, because data centers are critical infrastructure and nobody lets a Bangalore startup plug an experiment into a live cooling loop. In the companion newsletter I added two structural caveats of my own: operators “are here to play with Data, AI and cute pictures of babies stored on drives”, not to optimize exotic water tech, and “if heat networks take off, selling heat will always be a better monetary deal than selling water”.
The $497 million counterfactual
And picking the wrong application is not a hypothetical risk. I mapped the whole category in my Leviathan database: 32 companies carry the atmospheric-water-generation tag, with disclosed funding of $497 million. So let’s do the maths together: SOURCE Global, which sold hydropanels as a “water for all” play, raised $270 million by ESG Today’s 2022 tally ($223.8 million of it in rounds my database can see), and was declared dead in early 2025, four months before Uravu closed its fifth round. That one bet absorbed 45% of every disclosed AWG dollar, and with the other casualties added, 45.4% of the category’s lifetime capital now sits in dead or distressed companies. Colin Goddard, then SOURCE’s VP North America, had told me on this same microphone the goal was “the cheapest water on Earth”, and thermodynamics had other plans.
Is the smart money buying the strategy?
The data-center thesis reached Uravu Labs through its own cap table: about 18 months before our conversation, investor Peter Yolles of Echo River Capital kept sending Swapnil stories about data centers drinking whole towns’ worth of water; the founder went hunting, and the investor later cut a follow-on cheque. By late 2025 I counted nine AWG companies raising within the year, Echo River’s follow-on among them, in my “Four Questions About Source” edition.
Now the wallet-level evidence, from my Leviathan database again: Uravu’s syndicate has widened every full round, 5 investors in the 2021 pre-seed, 8 in 2022, 9 in 2023, 10 in the July 2025 round that funds the pilots, with $3.5 million disclosed (Inc42 pegged the total near $4 million in April 2023, so read that figure as a floor). And one stat I couldn’t resist computing: not one of SOURCE Global’s 17 backers, Gates, Bezos, Jack Ma, Masayoshi Son, Branson, BlackRock, appears on Uravu’s cap table. For the wider context, my State of Water Tech Funding 2025 has the full picture.
FAQ: the questions people actually ask
Do data centers use water?
Yes, mostly for cooling: Global Water Intelligence sized data-center water CAPEX at $569 million in 2025, cooling taking about 80%. Uravu’s angle is to flip the same problem around: use the heat side to produce water instead of only consuming it.
What is an atmospheric water generator?
A machine that harvests the moisture in air and turns it into liquid water, by chilling air below its dew point (the dehumidifier route) or by absorbing moisture into a desiccant and driving it out with heat. Uravu abandoned the first route after 1.5 years: below 50 to 60% relative humidity, it stops cooperating.
How do you make water from air without electricity?
Strictly speaking you don’t, but you can make it without dedicated energy. By Uravu’s fact-checked projection, a 45°C waste-heat stream dropped to 30°C drives the desiccant cycle at about 0.83 liters per kilowatt-hour of heat, landing near 1 cent a liter versus 2 cents on grid power.
Do atmospheric water generators really work?
They work; the honest question is where they make sense. The category runs near 5% of its thermodynamic optimum against desalination’s 30 to 40%, so pure drinking-water plays lose to wells, tankers and bottles almost everywhere. They earn their keep where quality or an allied application, cooling above all, carries the economics.
What is the revenue of Uravu Labs?
About $1 million in annual recurring revenue from the Bangalore consumer and hospitality business, a segment Swapnil himself caps at around $10 million, which is precisely why the company is hunting bigger applications.
Is Uravu Labs water safe to drink?
What leaves the machine is distilled water at near-laboratory grade (pH around 7, near-zero dissolved solids, in Swapnil’s words almost pharma-grade ASTM Type 3), which the company then supplies to hospitality and beverage customers.
What is the future of Uravu Labs?
Ten to fifteen replications of the Bangalore factory, a project deployed in Abu Dhabi, and data-center pilots targeted within 6 to 12 months, with integration at scale a couple of years out by the founder’s own estimate.
What a liter of water costs in Uravu’s home market (tap to expand)
| Source | Price as stated | Roughly, per liter | Who said it |
|---|---|---|---|
| Water tanker, Bangalore-style (3,000 L) | $40 to $60 per tanker | 1.3 to 2 US cents | Navkaran Singh Bagga, AKVO, on the podcast |
| 20-liter jar water | 4 to 5 rupees per liter | about 5 to 6 US cents* | Navkaran Singh Bagga, AKVO, on the podcast |
| Bottled water, retail | about 1,000x the price of tap water per liter | tens of US cents and up | Jacob Bossaer, BOSAQ, on the podcast |
| Uravu Labs, production cost (levelized) | 2 cents grid / 1 cent waste heat | 1 to 2 US cents | Swapnil Shrivastav, this episode |
*at roughly 85 rupees per US dollar (2025). Production cost and retail price are different animals: the ladder shows the price environment Uravu sells into, and why hospitality-grade water in India can carry a 2-cent production cost. Source: (don’t) Waste Water podcast transcripts, verified in my Leviathan database, July 2026.
The two-year test
So, does the strategy hold? My honest read: the physics argument for waste-heat AWG is sound, the fact-check landed on the efficient-but-plausible end, and the funding pattern (small cheques, a widening syndicate, an investor who seeded the thesis) looks like conviction rather than fashion. But the strategy has written its own exam: within roughly two years, either a data center lets Uravu bolt onto its cooling loop, or the flagship application stays a well-argued pitch while hotels and breweries keep the company alive.
If there are people in the data center space hearing this, I would really like to tell them that cooling and water can be looked at through a single lens, and a lot of opportunities to really make a lot more efficient, affordable, and sustainable solution.
Swapnil, who at 29 calls himself probably the oldest person in his company, is spending the new round to force the answer, and he closed our conversation with an open call to data-center and climate-controlled-agriculture operators who want to co-create a pilot. And if you want to judge the bet yourself, listen to the full episode below, because the best due diligence in water tech is still hearing a founder argue against his own pitch.
Sources
- Swapnil Shrivastav on (don’t) Waste Water S13E5 (2025-08-13) – all quoted material verbatim from the transcript corpus, timestamps per YouTube HMLrixueZ9U.
- SOURCE Global $270M total raised – ESG Today, “Sustainable Drinking Water Tech Provider SOURCE Global Raises Over $130 Million” (2022-07-19), https://www.esgtoday.com/sustainable-drinking-water-tech-provider-source-global-raises-over-130-million/ (retrieved 2026-07-12; page states “SOURCE has now raised a total of $270 million”); death: dww.show/source-global-failure (2026-06-11); DB status Dead 2025-03-01.
- Water Abundance XPRIZE finalist ($1.75M purse, 5 finalists, $50k grant, 2,000 L/day at 2 cents/L spec) – WaterWorld, “Five teams advance to final round of $1.75M Water Abundance XPRIZE” (2018), https://www.waterworld.com/drinking-water/potable-water-quality/article/16225255/five-teams-advance-to-final-round-of-175m-water-abundance-xprize (retrieved 2026-07-12, 200).
- IEA ~945 TWh data-centre electricity by 2030 – IEA, “Energy and AI” (2025), https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai (retrieved 2026-07-12).
- GWI $569M 2025 data-center water CAPEX, cooling ~80% – Global Water Intelligence WaterData platform (2025; subscription service, no public URL), cited in-episode (editorial, start_ms 310580) and in the newsletter edition “VCs vs. Physics: Who Wins?” (2025-08-13), https://www.linkedin.com/pulse/vcs-vs-physics-who-wins-antoine-walter-ovboe/ (retrieved 2026-07-12).
- Gasson “egregious waste of money” – quoted in the same newsletter edition + episode editorial; original GWI editorial [locate exact issue at assemble].
- Peter Yolles / Echo River Capital investment in Uravu – Crunchbase person profile + Echo River updates (retrieved 2026-07-12); the nine-2025-AWG-raises count + Echo River follow-on: newsletter “Four Questions About Source” (2025-10-21), https://www.linkedin.com/pulse/four-questions-source-my-follow-up-antoine-walter-djkye/ (retrieved 2026-07-12).
- Uravu ~$4M raised as of April 2023 – Inc42, “Here’s How Uravu Labs Is Conjuring Drinking Water Out Of Thin Air” (2023-04-19), https://inc42.com/startups/heres-how-uravu-labs-is-conjuring-drinking-water-out-of-thin-air/ (retrieved 2026-07-12).
- Operator-priorities and heat-networks caveats – newsletter “VCs vs. Physics: Who Wins?” (2025-08-13), quoted verbatim, https://www.linkedin.com/pulse/vcs-vs-physics-who-wins-antoine-walter-ovboe/ (retrieved 2026-07-12; see evidence-pack “Newsletter verbatim additions”).
- Leviathan database computations (as of 2026-07-12): AWG category 32 companies / $497.3M disclosed / SOURCE 45.0% / dead-distressed 45.4%; Uravu 5 rounds / $3.5M disclosed / 22 investors / syndicate 5-8-9-10; zero cap-table overlap with SOURCE’s 17 investors.