Water is cool. We value it, not just as a quantity, but because of what it means.

We need to build a technology that fights for the environment and the people. We need a new grid.

The watershed is the fundamental unit of terrestrial landscape on the planet’s surface. When we seek to use these landscapes, we alter or sustain their rivers and streams. In all activities, every moment of our day to day lives, we are either degrading or restoring rivers.

Housing for single tier version, 3d printed. Only does 2 dimensional groundwater flow monitoring, but lower cost & open source build complexity.

Working through different 3d printed designs for sensor mounts inside our groundwater flow probe. Minimize affect on flow while maintaining solid placement.

Which do we need more: neuromorphic computing circuits or healthy, sustainable landscapes? Which do you think we should invest our time and energy and resource in creating?

We need a strategy to monitor the sound

The construction of big buildings itself impacts water supply. Water tables are often artificially lowered, and water is needed for concrete, dust control, and cooling. These impacts are considered 'temporary', but they have real affects on communities during large scale industrial build out.

Nobody can argue against the Colorado River needing a water management plan, and that water allocations need to evolve. The question is, how much of that plan should be influenced by a new water intensive industry, the AI data centers?

Looks like necessary plans to cut water supply from the Colorado River are about to go on the books. What we're looking for are corresponding control on water intensive industries, right?

One of the best ways to know what is happening with groundwater is to drill and instrument a small monitoring well. With the right sensors, you know not only the compass direction of flow, but if the flow is heading up or down - cardinal signals for heaving pumping nearby.

Debugging...

2. calculator.ecologits.ai puts chat and report generation first on list to calc AI data center impact, which show small footprints. But scroll down to explore 720p - 4k video, see a different scale, and think about how much this is already being used.

1. EcoLogits calculator for AI data center water/carbon footprint is a useful tool, calculator.ecologits.ai. But there's a problem: these are mainly the hyped, consumer features, not the big use case. We to profile industrial use: "agents" that self-query continuously, automation, full scale coding

Our current sensor deployments use LoRaWAN, which is a multi-gateway star network configuration. We also want to explore Meshcore, so that it's easier for individuals to help build out local environmental monitoring networks.

We are in the midst of a massive extinction event. The easiest way to perceive this directly is to do more listening. We should be hearing birds, animals, insects all around us. Most of us hear almost nothing.

We gotta consider the whole infrastructure. There’s the water used by AI data center, then the water used by the power plants, the water used to manufacture AI chips, and finally the water used to build the power plants and chip factories themselves.

Chamber for demonstrating measurement of groundwater flow changes when a well is 'hard pumped,' depleting normal levels of groundwater.

Preparing demo chamber for our groundwater flow sensor. Simulated ‘well’ on the right.

We see several AI data center projects in dry areas, with plans for water piped in from the Colorado River. So no groundwater impacts right? But the Colorado is running dry, check lake Powell. What's the actual plan here??

Lake Powell will hit unprecedented lows, but that's the water supply for Tuscon data center... can groundwater recharge for Tuscon's aquifer flow if the Colorado river is running down?

We are getting ready to take a closer look at the groundwater...

OK, so some people are thinking about DIY sonic monitoring of these data centers. Who's game to get some recordings? The best way to solve a problem is to document the problem.

There are 1052 operating data centers in the US with a combined capacity of 48,527 MW, and 1415 planned projects that would add 352,100 MW of additional capacity. The first step is to start crowdsourcing before, during, and after : images, recordings, water and environmental data.

In this moment we have an extraordinary opportunity to properly monitor the build out of a new industrial infrastructure, and put modifications and limitations in place to make sure it's done safely or not done at all.

You have to interrogate the whole AI supply chain to grok the impacts. sourceability.com/post/the-sec...

Semiconductor water use is only growing | Sourceability

As the U.S. aims to expand its chip presence, semiconductor water use in manufacturing is exploding. In order to meet domestic ambitions, water is key.

sourceability.com

Every wave of industrial build out raises questions of design. So far, the world has gotten it really really wrong, every time.

Engineering-wise, this might solve one of many problems. First, all those billions need to convert all existing data centers. Next, engineers need to definitely solve water use, energy use, and thermal pollution. Only then can such a massive build out proceed. www.theverge.com/tech/954139/...

Nvidia says its AI data center design runs hotter to use a lot less water

Data centers running “hotter than a hot tub.”

theverge.com

EcoLogits has a responsible take on data center water use ecologits.ai/latest/ but it's a complex situation. They nail down impacts per query, but making the chips and model training are also huge. We're gonna need a lot more transparently to grapple this issue. ecologits.ai/latest/asset...

Do communities know enough how their own groundwater is used?