ARTIFICIAL INTELLIGENCE DATA CENTERS

But first, about that war with Iran.  Revenge pledges for the assassination of its Supreme Leader are emanating during this six-day funeral period, while causing more mischief in the Strait of Hormuz.  Also too, President Donald Trump keeps changing his mind about what to do in retaliation.  All those acts resulted in a drop in the stock market today and a 5% jump in the price of petroleum.

Trump has also been conversing a lot with Vladimir Putin on the phone, and showed unusual warmth towards Volodymyr Zelenskyy at the NATO summit in Turkey.  He also called Putin a difficult character, and the same about Zelenskyy.  Watch this video showing Zelenskyy and Trump in Ankara.

Today will be the first day of rest for the FIFA World Cup.  There will be other pauses until the championship game on Sunday, July 19 at 3PM EDT in MetLife Stadium, New Jersey.  

• Tomorrow, France against Morocco in Foxborough, Massachusetts, with lowest ticket price at $1081.  France is favored, -180, O/U 2.5.
• On Friday, Spain plays Belgium in Los Angeles, and the lowest available ticket costs $1321.  Spain favored, -160, O/U 2.5.
• Saturday:
• Norway vs England, Miami, Florida, lowest ticket price $1935, England favored, -115, O/U 2.5.
• Argentina vs Switzerland, Kansas City, Missouri, lowest ticket price $1501, Argentina favored, -145, O/U 2.5.

What has rapidly become so odious to the masses as coal power plants and nuclear waste depositories is something called Artificial Intelligence Data Centers.  Why?  Time magazine reports that:

• 
  Northern Virginia itself has 600 operational data centers, where 67% of all internet of the world's internet traffic passes.
• Video on the 10 largest AI data centers today.

• 

  800 mega data centers are being built worldwide.
• The largest is a 5 GW (5000 MW) Meta campus in Louisiana called Hyperion, expected to become operational in 2030.
• Three noteworthy others are:  
• Stargate (Open AI/Oracle/SoftBank) in Texas, which also could go up to 5 GW, for use as soon as 2028.
• Project Rainier (Amazon) in Indiana, 2.2 GW by 2028.
• Azure AI for Microsoft in Wisconsin, 2 GW.

• 
  Tech companies are set to spend $7 trillion on these facilities by 2030.  
• Microsoft and OpenAI are together building one in northern Norway for two simple reasons.
• Electricity costs 3-4 cents/kilowatt-hour.  European rates are close to 30 cents/kWh.
• Computer chips run hot, so cold climate is a bonus.
• Maybe Antarctica?

A history of data centers.

• 1946:  ENIAC (Electronic Numerical Integrator and Computer).
• Essentially a computer at the University Pennsylvania to study the feasibility of a thermonuclear weapon, leading to design of Hydrogen Bombs.

• 
  Called a Giant Brain by the press.
• Had a speed about 1000 times faster than electro-mechanical machines.
• Consumed 150 kW of electricity, or 0.15 MW.

• 
  Used no water. Cooled by forced-air blowers.  The data centers of today must use water-cooling for efficiency.
• Cost $487,000, equivalent today of more than $7 million.
• But was less powerful than a pocket calculator.

• 
  1964:  Mainframes, IBM System/360.
• 1971:  Microprocesssors, Intel 4004.
• 1982:  Personal computer revolution, like the IBM PC.

• 
  1989:  The World Wide Web conceived by Tim Berners-Lee at CERN, the European Organization for Nuclear Research.
• 1998:  Virtual servers.
• Late 1990s:  Colocation providers, like Equinix.
• 2000:  Dot-Com boom and bust.
• 2006:  Amazon Web Services led to cloud era.

• 
  2010:  Rise of hyperscale data centers for Google, Facebook and Microsoft.
• 2013:  Shift to cloud-native computing.
• 2017:  Edge computing gains traction.

• 
  2020:  The COVID Pandemic and remote work led to cloud services.
• 2022:  Generative AI, like OpenAI's ChatGPT went mainstream.

• 
  2024:  First AI data center.
• 2025:  Neoclouds.

The electricity and water problem.

• Large data centers require around 5 million gallons of water/day, about as much as a city of 50,000 people.  

• 
  Meta's Hyperion will need twice the power New Orleans.  Another Meta center in Wyoming will use as much as all the homes in the state.
• Promised to use renewable energy.
• But the sun and winds cannot provide baseload power, so Meta will also add nuclear power.

• 
  In 2023, U.S. data centers used as much electricity as the entire nation of Ireland.  This use could triple by 2028.
• In 2025, Texas data centers will use 40 billion gallons of water, to go up to 400 billion gallons in 2030, the equivalent of drawing down the largest reservoir in the U.S., the 157,000-acre Lake Mead, by 16 feet/year.

• 
  U.S. AI data centers used only 0.3 trillion gallons of water annually.  Much of this flow is evaporated and later return as rain.  But the rest is recycled in some way.
• On the other hand, we use in the U.S. 0.5 trillion gallons just to irrigate golf courses, plus 2 trillion gallons for our lawns.
• What are our priorities?

But data centers are an economic boon for the local government.

• According to Google AI:

AI data centers generate significant local economic value through multi-billion dollar capital investments, dependable property and personal property tax revenues, and the stimulation of secondary local businesses. However, because they are largely automated, they create relatively few permanent jobs, and concerns about strained utility grids and water resources often drive community pushback. [1, 2, 3, 4, 5]

• 
  Loudon County, VA collected $900 million, about the county's operating budget.
• Somewhat reminiscent of local approval to build a nuclear power plant, for public backlash leads to project delays and cancellations.

So in grand summary, Time magazine published an article this past week on this subject.

• In the public discourse, data centers have largely been cast as the villains of the AI boom—voracious consumers of natural resources that power our chatbots but threaten our climate progress. But in fact, there are clear, actionable ways we can design, build, and operate data centers that align with both our climate goals and societal values.
• Currently, data centers are highly concentrated in a handful of regions: Virginia, Ireland, Texas, Singapore. This geographic clustering puts severe strain on already fragile energy grids and aquifers. 
• 
  Interestingly enough, those locations with geothermal energy resources could benefit, for this electricity is baseload, unlike wind power and photovoltaic electricity.  From Google AI:

The most promising locations globally for geothermal-powered AI data centers are regions with high tectonic activity (like the Pacific Ring of Fire) and climates that provide natural cooling. Key geographic hubs include the southwestern United States (Nevada, Utah, New Mexico), Iceland, East Africa, and parts of Southeast Asia. [1, 2, 3, 4, 5]

These regions are uniquely suited to meet the \(24/7\) continuous power demands of AI computing: [1, 2]

1. Southwestern United States

The western US is rapidly becoming the epicenter for geothermal-powered AI infrastructure due to naturally high underground temperatures and supportive state policies. [1, 2, 3]

• Nevada: Home to operational Fervo Energy projects supporting Google's Data Centers, as well as Zanskar geothermal sites backing Amazon's Reno expansion. [1, 2, 3]
• Utah: Promoted as an "Enhanced Geothermal System (EGS) Corridor", this region features the DOE's Utah FORGE research facility alongside growing hyperscaler developments. [1, 2]
• New Mexico: Meta has partnered with providers like Sage Geosystems and XGS Energy to develop hundreds of megawatts of carbon-free next-gen geothermal capacity. [1, 2, 3]

2. Iceland

Known globally as a leader in 100% renewable energy, Iceland offers ideal conditions for sustainable AI data centers. [1, 2, 3, 4]

• Abundant Baseload Power: The country's location on the Mid-Atlantic Ridge provides cheap, abundant geothermal and hydroelectric resources. [1, 2]
• Natural Cooling: Consistently cool ambient temperatures significantly reduce the energy typically required for data center chilling. [1, 2, 3]
• Key Entities: Major providers like at North (operating multiple facilities near Reykjavík and Keflavík) and Borealis Data Center dominate the Icelandic market. [1, 2, 3]

3. East Africa (The Great Rift Valley)

The East African Rift system is considered one of the most promising emerging markets for clean-energy computing.

• Kenya: Already an established data hub for East Africa, Oracle recently announced plans for a cloud region there. The nation gets roughly half of its electricity from the Geothermal Rising resources of the Rift Valley. [1, 2]
• Ethiopia: Ranks among the highest in Africa for geothermal enthalpy and is actively shaping its data center market to leverage this clean power. [1]

4. Southeast Asia

Countries straddling the Pacific Ring of Fire are actively unlocking their geothermal potential for the data center sector: [1, 2, 3]

• Indonesia and the Philippines: Both are slated to see significant growth in geothermal capacity for tech hubs in the coming years.
• Japan: Sits on the Pacific Ring of Fire with enormous geothermal potential, making it a primary target for Asia-Pacific hyperscaler deployments. [1, 2, 3]

Perhaps the resource of greatest long-term promise is super hot rock geothermal, a field of interest in my earlier days of research.  Read my posting of this possible pathway.  According to this publication, this technology could supply up to 20% of the electricity in the U.S. by 2025. From Google AI:

Geothermal energy is experiencing a massive revival, transitioning from a niche power source into a major clean-energy contender. While conventional geothermal only taps naturally occurring hot water, new tech is unlocking the enormous, near-limitless potential of "hot rocks at depth". [1, 2, 3, 4, 5]

The industry is divided into two main categories: [1, 2]

1. Enhanced Geothermal Systems (EGS)

EGS targets moderately hot, dry rock roughly 2 to 3 miles deep. Because the rock is dry, operators use oil and gas-style hydraulic fracturing to create artificial reservoirs. Water is pumped down, heated by the rock, and brought back up to spin turbines. [1, 2, 4, 5]

• Status: Advancing rapidly. Private investment has surged, and the U.S. Department of Energy (DOE)has ongoing initiatives, such as the Frontier Observatory for Research in Geothermal Energy (FORGE)in Utah, to test commercial-scale EGS.
• Players: Companies like Fervo Energy are successfully deploying EGS, bringing high-efficiency, 24/7 carbon-free power to the grid. [1, 2, 3]

2. Superhot Rock Geothermal (SHR)

SHR aims to drill up to 12 miles deep into the Earth's crust to reach extreme temperatures (\(>300^{\circ}C\)). At these depths, water transforms into "supercritical" fluid—holding both liquid and gas properties—which yields 5 to 10 times more energy per well than conventional systems. [1, 2, 3]

• Status: In the developmental and demonstration phase. Conventional mechanical drills cannot withstand the extreme heat and pressure, so developers are testing next-generation technologies like millimeter-wave drills that vaporize rock.
• Players: Startups like Quaise Energy are actively developing this technology, with pilot plants planned in Oregon. [1, 2, 3, 4]

The Industry Shift

• AI & Tech Demand: The rise of energy-intensive data centers has spurred immense interest in geothermal. Tech giants are directly backing hot rock startups to secure 24/7 baseload power. [1, 2, 3]
• Government Support: Governments recognize hot rock as a massive untapped domestic energy source. In early 2026, the U.S. Department of Energy announced a $171.5 million funding opportunity to expand next-generation geothermal field tests. [1, 2, 3]

-