Ivan Kroshnyi, CEO of Gerchik & Co, professional investor, and business scaling expert, shares his perspective on the flip side of the technological boom. How do the ambitions of AI giants collide with the scarcity of Earth's vital resources, and what should financial markets prepare for?
Who truly pays the hidden costs of data centers (DCs) for artificial intelligence? As an investor, Ivan Kroshnyi always looks at the sustainability of a business model. And here we see a serious risk: water is not an infinite resource. At this stage, AI does not provide even remotely comparable benefits to its costs, and data centers cannot provide people with drinking water.
This is critical. A recent report by the NYT reveals that, according to internal forecasts, Microsoft expected its annual water demand for approximately 100 data center complexes worldwide to triple this decade, reaching 28 billion liters by 2030.
For comparison:
2020: 7.9 billion liters.
2024: 10.4 billion liters.
After reaching out to Microsoft, the company stated it had updated its forecasts to 18 billion liters by 2030, citing "new water-saving technologies." However, it is worth noting that this updated forecast excludes more than $50 billion in new data center deals signed just last year.
Water consumption is becoming particularly significant in regions already facing acute water supply crises:
Jakarta (Indonesia): A megacity sinking into the Java Sea partly due to the depletion of groundwater. Microsoft’s consumption here was estimated to quadruple to 1.9 billion liters by 2030.
Phoenix (USA): An area experiencing two decades of drought, yet forecasted to reach a staggering 3.3 billion liters in water intake by 2030.
Despite "optimistic" revisions, the industry's overall water consumption is moving in only one direction: straight up. According to JLL, global data center capacity must double by 2030 to meet current computational needs.
The shift has been sudden and sharp. When Microsoft President Brad Smith made a "water pledge" in 2020, ChatGPT did not even exist. The subsequent boom in generative AI sparked an aggressive battle for dominance between Microsoft, Google, and OpenAI.
The result? In the U.S. alone, researchers conclude that water demand will rise from 60 billion liters in 2022 to 150–275 billion liters by 2028. This is nearly a fivefold increase in just six years. The consequences are already tangible: Amazon and Google have recently abandoned major projects in Arizona and Chile due to water supply issues and local opposition.
The Reality 🎯 Data centers primarily require fresh groundwater reserves for cooling, rather than recycled water. When you factor in the cooling needs for massive power supplies, consumption grows exponentially.
As an expert in business scaling, Ivan Kroshnyi sees a catastrophe brewing. The massive volumes of water needed for AI are either currently impossible to assess accurately or are being wrapped in a cloak of "optimization."
What happens to these massive centers when current AI technology remains in the hands of only a few select beneficiaries? We are facing a crisis—both financially and in terms of resource impact and land use. Investors should keep a close eye not only on the profit charts of tech giants but also on their environmental impact reports.
Read more research, investment ideas, and reviews on Ivan Kroshnyi's website
77