ai data centers

Bilingual

Demand from AI data centers drives up computer memory prices
2 months ago
- AI data center demand is causing computer memory prices to skyrocket, doubling or more for some products.
- This price surge contradicts Moore's law, which predicted that technology would become cheaper and more powerful over time.
- Moore's law faces physical limits at the nanometer scale, requiring more creativity and investment to maintain progress.
- Supply and demand dynamics are driving the price increase, with AI companies like OpenAI needing massive amounts of memory—such as 40% of the world's supply for the Stargate project.
- The long-term viability of Moore's law is uncertain, as technological breakthroughs must keep pace with surging AI demand.
Brockovich Data Center Reporting – U.S. AI Data Center Awareness and Issue Map
2 months ago
- AI data centers are being built across the US, with varying community reactions—some welcomed, others delayed or contested.
- An interactive map shows operational, under construction, proposed, and community-reported data centers, with details and source citations.
- Key concerns include high energy and water consumption, e-waste, location risks, scalability issues, and noise pollution.
- Recent news highlights political defeats, rural resistance, bipartisan opposition, and new regulations against data centers.
Five tech giants are hiding $1.6T in AI debt, using the trick that toppled Enron
2 months ago
- Alphabet, Microsoft, Amazon, Meta, and Oracle have $1.35 trillion in reported debt, but a Nikkei study reveals an additional $1.65 trillion in off-balance-sheet debt, up eightfold in four years.
- The hidden debt uses off-balance-sheet vehicles similar to Enron's, but is now legal due to tighter rules and fuller disclosures, though it echoes Enron's structure.
- Companies package debt for chips, servers, and power into separate legal entities like joint ventures to avoid recording it on their own accounts, e.g., Meta's Hyperion data centre structure with $27 billion debt.
- Examples include Oracle's $260 billion in future lease commitments, Nvidia's $119 billion in purchase obligations, and Alphabet and Microsoft also keeping vehicles off-book.
- The industry is expected to spend over $3 trillion through 2028 on AI data centres, with much financed against chips, but the hidden debt could trigger losses if AI demand fails.
- Off-balance-sheet debts become real when data centres go live, potentially leading to credit rating downgrades (e.g., Oracle's S&P cut) and flagged risks from Moody's and Morgan Stanley.
- The disclosures exist in financial footnotes, but investors see less than half the actual leverage, making it a quiet risk amid bubble talk.
A 600-acre AI data center could cost some Wisconsin residents their land
2 months ago
- A 600-acre AI data center project in Port Washington, Wisconsin threatens private land through eminent domain for power lines.
- Tom Uttech, an 83-year-old artist, faces losing his 52-acre property, which inspires his work, to a transmission line for the data center.
- The data center is part of a $15 billion Stargate partnership under the Trump administration, aiming to advance AI and compete globally.
- AI data centers are highly energy-intensive, with single facilities potentially using as much electricity as a large U.S. city.
- Local officials, like Mayor Ted Neitzke, support the project for expected job creation, tax revenue, and community transformation.
- Residents and activists protest over eminent domain concerns, potential electricity bill increases, and environmental impacts.
- OpenAI and Oracle pledge to cover energy costs and use renewable sources to avoid raising local electricity prices.
- The final transmission line route will be decided by Wisconsin's Public Service Commission, with alternatives minimizing landowner impact.
- Uttech is legally challenging the project with a conservative law firm to protect his property rights and land's natural beauty.
Microsoft Reports Maasive 25 Percent Jump in Emissions
2 months ago
- Microsoft's greenhouse gas emissions increased by about 25% last year, driven largely by the expansion of data center infrastructure.
- Similar emission increases were reported by Google (18%) and Amazon (16%), highlighting a trend of rising tech company emissions.
- Scope 2 emissions, from purchased energy, accounted for 13% of Microsoft's total increase.
- Microsoft matched 100% of electricity consumption with carbon-free sources but has made recent deals involving gas-powered data centers.
- Partnerships with Chevron and other projects could emit millions of tons of CO2 annually.
- Microsoft stopped purchasing unbundled renewable energy certificates to avoid greenwashing, prioritizing investments in new clean electricity.
- Despite emissions rise and AI investments, Microsoft aims to be carbon negative by 2030.
The Lies They're Telling Towns and Tribes About the Benefits of AI Data Centers
2 months ago
- AI data centers are falsely promoted as economic saviors for rural towns, but they create few long-term jobs and often rely on out-of-area contractors.
- Claims that data centers attract innovative businesses due to lower latency are misleading, with minimal real-world impact on AI service quality or local entrepreneurship.
- Water consumption by data centers is downplayed by comparing it to other uses like golf courses, ignoring severe local strains in drought-prone areas.
- Corporate-backed think tanks, like the Progressive Policy Institute, spread pro-industry narratives under a veneer of objectivity, masking their lack of genuine expertise.
- Tribal lands are targeted for data center projects to bypass regulatory hurdles, exploiting sovereignty for quicker permits despite risks of unfulfilled promises.
- U.S. regulatory systems for labor, environment, and public safety are largely dysfunctional, leaving communities unprotected against corporate abuses.
- Public opposition to data centers is widespread and justified, driven by concerns over pollution, higher energy costs, and broken accountability mechanisms.
AI data centers and the concentration of wealth
2 months ago
- Opposition to AI data centers unites communities across party lines, focusing on local costs like housing misallocation, energy pressures, and environmental impact, but may obscure larger issues of AI companies' power and wealth concentration.
- AI companies aim to capture value across entire industries, targeting fields like enterprise software, creative design, and legal services, while deflecting resistance to infrastructure debates.
- Data center opposition has mixed success, often failing against well-funded projects, and may be a temporary spike as technology shifts toward smaller models and on-device AI, reducing demand for centralized computing.
- The core risk is the concentration of power and wealth in AI companies, exacerbating inequality, with corporate influence over politics posing existential threats that require solutions like public AI, taxation, and regulation.
- Broader concerns include economic instability from AI-driven debt spirals, heat pollution from data centers, and long-term impacts on education and R&D, highlighting the need for a populist agenda against corporate power.
Data Centers Are Quietly Taking over Texas. The Pollution Could Be Catastrophic
2 months ago
- OpenAI's Stargate data center in Abilene, Texas was built next to Omaira Garcia's ranch without prior notice, causing dust and pollution concerns.
- Texas regulators allowed data centers like Stargate to use minor air permits ('permits by rule') to avoid extensive environmental reviews and public input, leading to quiet construction of gas plants and diesel generators.
- A Floodlight investigation found that minor permits have enabled dozens of Texas data centers to operate onsite power sources, emitting significant pollutants, with over 2,100 backup diesel generators permitted statewide.
- Stargate's developers used a 'small first, big later' strategy, starting with minor permits for 10 turbines and 62 generators, then filing for a major expansion with 41 more turbines and 18 generators, potentially making it one of Texas's largest fossil-fuel plants.
- Former regulators and watchdogs criticize the Texas Commission on Environmental Quality (TCEQ) for lax enforcement, noting that public participation often comes too late, after facilities are built, and that the agency has a backlog of over 1,400 cases.
- Research indicates Texas could surpass Virginia as the top data center market by 2030, with AI-driven construction leading to a 'shadow grid' of gas plants, risking long-term fossil fuel dependency and significant greenhouse gas emissions.
- Residents like Garcia face reduced quality of life, including traffic, trash, and health worries, with property values affected and limited options to move, highlighting the human impact of rapid, unregulated AI infrastructure growth.
IBM claims first sub-1 nanometer chip technology
2 months ago
- IBM developed a new chip architecture integrating nearly 100 billion transistors at a density twice that of its previous generation.
- The technology, termed the 'world’s first sub-1 nanometer chip technology,' promises significant performance improvements and energy efficiency for AI data centers.
- IBM’s 'nanostack' architecture aims to achieve the computing benefits expected from theoretical sub-1nm chips without physically shrinking features below that limit.
- The chip is described as being built at the 0.7-nanometer node, labeled as the 7 angstrom node, but node numbers no longer directly correspond to physical dimensions.
- Historically, chip node names matched physical feature sizes, but modern processes like 3nm or 2nm are marketing terms, not literal measurements.
Google's AI buildout drove 37% increase in electricity use in 2025
3 months ago
- Google's annual electricity consumption rose by 37% in 2025, the largest increase in company history, largely due to AI data center expansion.
- Total electricity usage has grown over 250% since 2019, driven by Google Cloud, YouTube streaming, and data center operations for AI services.
- Data centers consumed over 42 million MWh in 2025, rivaling the electricity consumption of entire countries like New Zealand, Denmark, and Nigeria.
- Despite the surge in electricity use, Google reduced operational carbon emissions by 2% through continued purchases of clean energy.
- Supply chain emissions increased by 25% due to Asia-Pacific grids lacking sufficient carbon-free energy, leading to an 18% rise in total 'ambition-based emissions'.

ai data centers

Demand from AI data centers drives up computer memory prices

Brockovich Data Center Reporting – U.S. AI Data Center Awareness and Issue Map

Five tech giants are hiding $1.6T in AI debt, using the trick that toppled Enron

A 600-acre AI data center could cost some Wisconsin residents their land

Microsoft Reports Maasive 25 Percent Jump in Emissions

The Lies They're Telling Towns and Tribes About the Benefits of AI Data Centers

AI data centers and the concentration of wealth

Data Centers Are Quietly Taking over Texas. The Pollution Could Be Catastrophic

IBM claims first sub-1 nanometer chip technology

Google's AI buildout drove 37% increase in electricity use in 2025