Things we were asked/told by a panel:
1. This will dry out cities & there won’t be water vapor left
2. ACs are condensing water while running so there will be less water for us
3. NASA is not a credible body to cite while explaining that vapor replenishes naturally in 8 days
….
We can build highly efficient and water positive data centers by using on-site atmospheric water generation powered by waste-heat from servers! @uravulabs
The human brain: 2% body mass, but consumes 20% of its energy. Cortical neurons fire 0.16 times per second. BUT they are capable of firing at 40 or more. A 250-fold gap. If more than a few percent of neurons fired at high rates simultaneously, the brain would literally overheat. So less than 1% fire at any given moment. Frontier AI models have the same two constraints: sparse activation and thermal limits. Mixtral activated 27.6% of its parameters per token. DeepSeek-V2 activated 8.9%. DeepSeek-V3 has 671 billion parameters and activates 37 billion of them. That's 5.5%. NVIDIA hit the same wall. The GB200 generates 120 kilowatts per rack. Air couldn't cool it. They switched to liquid and unlocked 30% more compute. Now, what would happen if we could cool our brains? Neurons that fire faster produce measurably higher IQ scores, but three things stop us: heat dissipation, oxygen delivery, and ion channel reset time. There's already a device that achieved a 3°C brain temperature drop in 30 minutes by running chilled saline through the nasal cavity. So the first human IQ-overclock device might look less like Neuralink and more like a beer helmet with tubes running up your nose.
We won $10K at MIT and had to figure out how to start a chemical company with it. So I made a spreadsheet. Every single application of hydrogen peroxide, sorted by dollar per gallon. Industrial use? Low margin. Semiconductors? Long sales cycle. Sensory deprivation tanks? Float spa owners pay premium for clean peroxide and there were hundreds of them. I cold-called every one I could find and sent them our MIT pitch video. Within weeks we had recurring orders. Our PVC reactor held together with twist ties was cash flowing $10K a month selling "bioperoxide" to float spas. That's how a multibillion dollar chemicals company got its first customers.
@ShanuMathew93 Bang on! Only that district heating is limited to colder weather and requirement is not all year round. More interesting applications - water from air like @uravulabs is doing, or even carbon capture will turn out to be more scalable and less geography + infra dependent.
Liquid cooling will enable access to 35 C to 45 C heat and thus open up interesting opportunities for solutions like @uravulabs which makes water from air while cutting cooling costs!
Pictured: 3rd-generation Cooling Distribution Units, providing liquid cooling to an Ironwood superpod.
We’ve deployed liquid cooling across 4 generations of TPUs. Google first used liquid cooling in TPU v3 in 2018. Today, it’s used in Ironwood → https://t.co/IPDRS7OD7Z
Thank you @bbgoriginals for featuring @uravulabs on The Optimists Guide to the Planet - I spoke about our water from air solution, scaling up to millions of liters per day, and our inspiration @starwars
Water is our most precious resource, but it's increasingly becoming scarce. @uravulabs, founded by @swapnil_ides, Venkatesh RY, and Balaji GB, is changing that by creating a sustainable, energy-efficient solution for water generation directly from the air. Using cutting-edge Water From Air technology, Uravu Labs harnesses atmospheric moisture and solar-powered systems to provide reliable, decentralized water sources for industries, households, and agriculture. With this breakthrough, Uravu Labs is not only addressing water scarcity but also creating a future where clean water is accessible anywhere.
This innovative solution is a critical step in making clean, sustainable water a reality for the millions facing water stress worldwide.
Visit the Emerge 50 website to discover Uravu Labs and other startups making a real difference: https://t.co/6e0g6VdNr7
@nasscom@nasscomai@NasscomR@nasscomCoEIoT@DSCI_Connect@sangeetagupta29@ShaliniDewan@mayankkumar_17
@ShanuMathew93 I think they won’t do it till they develop their own ASICs - TPUs have shown it works and when you can serve your own compute cost effectively, making it available for other players is definitely good business
@AlexEpstein The embedded footprint of water in the energy mix >>>> direct on-site use for cooling.
An efficient cooling solution is a better bet - @uravulabs we are building an integrated cooling + water from air tech that’s more efficient than water cooled chiller & makes its own water!
@deanwball The embedded footprint of water in the energy mix >>>> direct on-site use for cooling.
An efficient cooling solution is a better bet - @uravulabs we are building an integrated cooling + water from air tech that’s more efficient than water cooled chiller & makes its own water!
@ArmandDoma The embedded footprint of water in the energy mix >>>> direct on-site use for cooling.
An efficient cooling solution is a better bet - @uravulabs we are building an integrated cooling + water from air tech that’s more efficient than water cooled chiller & makes its own water!
@firstadopter The embedded footprint of water in the energy mix >>>> direct on-site use for cooling.
An efficient cooling solution is a better bet - @uravulabs we are building an integrated cooling + water from air tech that’s more efficient than water cooled chiller & makes its own water!
@AndyMasley The narrative is more emotional, as it’s always the case with water! Also, the embedded footprint of water in the energy mix is much higher than direct on-site use for cooling.
A highly efficient cooling solution is likely a better bet - @uravulabs is building exactly that!
@Noahpinion Actually 50-90% of water footprint for a data center in the US is because of gas & coal. On-site water use is actually minuscule.
But doesn’t mean it’s not a problem. At @uravulabs we build cooling + water from air tech to turn data centers water +ve.
Community water is also a factor for the pushback on data centers.
https://t.co/LgkalNv88A
@uravulabs is solving this by building an integrated cooling + water from air solution to turn DCs water positive while reducing cooling load by upto 80%
Google was about to lose a vote by the Indianapolis City council and withdrew a proposal for a $1B data center.
Microsoft pulled a similar project in Wisconsin based on community pushback.
Are these isolated incidents or the beginning of a trend?
If the data below is accurate, then we should expect that more local communities will pushback to stop data center buildouts in their region if it spikes electricity prices so much.
A solution, however, is for these hyperscalers to use their robust free cashflow to go into local areas and pay for solar+storage or pay for the electricity for every resident.
Either way, the hyperscalers should take the electricity costs of local residents to zero and start buying goodwill. Otherwise, I expect more local communities to push back on these data centers which will complicate the AI buildout that needs to happen.
@ryanweather@chamath@uravulabs is building an integrated cooling + water from air solution to turn data centers water positive while reducing cooling load by upto 80%
At the heart of our system is a proprietary liquid desiccant salt which rapidly absorbs water vapor in the air.
DC waste heat (30-60°C) from servers—heat that is typically just vented—is channelled to this "wet" desiccant. This heat provides the energy to desorb (release) the captured water molecules.
We then condense this pure water vapor into high-quality, demineralized water—perfect for cooling towers or direct liquid cooling loops.
The process of pulling water from the air also has a powerful side effect: it directly reduces the cooling load by upto 80%, returning facility water at 22-32 °C
@asianometry has a great video - The Big Data Center Water Problem.
Optimising PUE, WUE, and heat reuse for data centers has to be a top priority to turn DCs into a water positive and low-carbon infrastructure.
@uravulabs is building an integrated cooling + water from air solution to solve the Big Data Center Water Problem