August 25, 2026

AI Watch

California’s largest AI data center is suing for access to 287 million gallons of Colorado River water a year– www.techspot.com

pgnewser July 30, 2026 0

California’s largest AI data center is suing for access to 287 million gallons of Colorado River water a year– www.techspot.com
News Source
EXCERPT:

A hot potato: Once again, an AI data center is embroiled in controversy. This time, it’s not land, electricity, or tax breaks under the spotlight, but water – specifically, water from the Colorado River. And the company behind the facility is suing to get it.

Imperial Valley Computer Manufacturing (IVCM), the company behind what could become California’s largest AI data center, is suing the Imperial Irrigation District (IID) after the publicly owned utility rejected its application for water.

The proposed $10 billion, 950,000-square-foot, 330-megawatt facility would require roughly 750,000 gallons per day for cooling, or about 287 million gallons annually.

That amount equals around 880 acre-feet, just 0.03% of IID’s 3.1 million-acre-foot annual entitlement. However, the Colorado River is Imperial Valley’s only source of fresh water and supplies between 35 million and 40 million people across the West. Federal officials are preparing new operating rules as reservoirs remain under pressure from prolonged drought and weak runoff.

Developer Sebastian Rucci argues that the project wouldn’t increase overall demand. IVCM has leased 160 acres of nearby farmland and plans to stop irrigating it, redirecting roughly the same amount of water to the data center.

AI Capability Races Ahead of the Business Payoff | American Enterprise Institute– www.aei.org

pgnewser July 29, 2026 0

AI Capability Races Ahead of the Business Payoff | American Enterprise Institute– www.aei.org
News Source
EXCERPT:

The state of the AI revolution can be confusing. At times, it can seem as if the latest artificial intelligence models are showing huge capability increases. One example is when two OpenAI models broke out of the lab, accessed the internet, and broke into another AI company. Or, as a Wall Street Journal headline put it, “The Day the Bots Broke Loose.”

But then you might come across another headline from the Wall Street Journal: “Big Companies Are Starting to Hire Again, Defying Predictions of AI Wipeout.” And from that piece:

The push to expand head count, at least modestly, is a reversal from the prevailing corporate messaging during much of the AI era. Major employers largely held back on adding people due to economic uncertainties or a belief that artificial intelligence could shoulder more tasks on the job. But some executives say the costs and limitations of AI now demand that more people be added; others want to hire people back following layoffs.

Graphics card prices jump by up to 20% in China as shops stockpile Nvidia and AMD GPUs– www.techspot.com

pgnewser July 29, 2026 0

Graphics card prices jump by up to 20% in China as shops stockpile Nvidia and AMD GPUs– www.techspot.com
News Source
EXCERPT:

Ripple effect: Just when it seemed like graphics card prices couldn’t get much worse, the memory crisis has delivered another unpleasant surprise. Nvidia and AMD GPU prices have surged in China, with some models rising by as much as 20% in a single week and premium cards now selling for almost 60% above their official MSRPs.

Distributor lists from MSI and Colorful reveal the increases. MSI raised the RTX 5080 Shadow 3X OC and Ventus from 9,999 yuan ($1,477) to 11,999 yuan ($1,773), a 20% jump in seven days. Its RTX 5070 Ti Shadow 3X OC rose 19.5%, while the RTX 5090D V2 increased 13%. What’s worse is that these cards were already well above their suggested prices before the latest adjustment.

Comparing the new figures with Nvidia’s Chinese MSRPs makes the situation look even worse. MSI’s RTX 5070 Ti Gaming Trio OC is now 54% above the base MSRP, while the RTX 5080 Gaming Trio models are around 57% higher. Colorful’s RTX 5080 Vulcan W OC is priced at 13,199 yuan ($1,949), 59% above MSRP, and its RTX 5070 Ti Vulcan W OC carries a similar 58.7% premium.

The increases aren’t just limited to Nvidia. A market snapshot showed Radeon cards climbing across Chinese retailers, with some RX 9000 models seeing double-digit increases. AMD reportedly warned Asus, XFX, Sapphire, and other partners earlier this month that its GPU-and-memory bundles would become 10% more expensive.

The culprit is, of course, the continuing rise in GDDR6 and GDDR7 costs, combined with more expensive PCBs, coolers, and packaging. Nvidia supplies board partners with GPU and memory kits, so higher VRAM costs are passed directly to manufacturers. Reports claim the latest adjustment adds around 600 yuan ($89) to 8GB cards, 900 yuan ($133) to 12GB models, and 1,200 yuan ($177) to 16GB products.

AI is Going into Trillion Dollar Debt

pgnewser July 23, 2026 0

A surge in hidden debt among AI giants like Oracle and Meta is setting off alarms in the industry. The hidden debt comes from future payment operations for expenses not yet online to make money. For instance, leases and property taxes must be paid on sites that are earmarked for databases that haven’t been built or completely built yet. The hidden debt amounts to more than $1.65 trillion.

The AI Illusion: Big Tech’s ‘Invisible Debt’ Surges To $1.65 Trillion On Opaque Funding – NDTV Profit– news.google.com
News Source
EXCERPT:

The hidden $1.65 trillion stems from future payment obligations-such as long-term operating leases for data centers not yet built, bulk GPU supply contracts, and joint ventures. Under current accounting standards, companies are not required to record these obligations as liabilities on their balance sheets until the assets are delivered or the facilities go live.

Instead, these financial commitments are buried in the small-print footnotes of their financial statements. While perfectly legal, this practice effectively obscures the true financial leverage of the companies. By shifting costs to future periods, tech giants can maintain balance sheets and inflate current profit figures, all while locking themselves into hundreds of billions of dollars in future payouts.

While all five giants are utilizing these financial structures, the report reveals that the scale of hidden debt at Meta and Oracle is particularly striking:

Meta: The social media giant’s off-balance-sheet debt has reached an estimated $420 billion, nearly three times its officially recorded debt. A significant driver is its joint venture with Blue Owl Capital to build a massive data center in Louisiana, a project whose expected investment recently ballooned from $27 billion to over $50 billion.

Oracle: The company’s hidden debt has surged more than 30-fold in four years, reaching $273.3 billion. Much of this is tied to concentrated, single-counterparty exposure, such as its massive “Stargate” supercomputer data center project with OpenAI.

Meanwhile, Alphabet, Amazon, and Microsoft have collectively disclosed backlogs of roughly $1.45 trillion for cloud services and related business commitments as of March, according to the report.

 

AI can become more biased than humans while choosing who gets hired; study finds |– timesofindia.indiatimes.com

pgnewser July 23, 2026 0

AI can become more biased than humans while choosing who gets hired; study finds |– timesofindia.indiatimes.com
News Source
EXCERPT:

Artificial intelligence is increasingly being trusted to help employers screen CVs, rank candidates and even conduct early-stage interviews. The promise is simple: AI can process thousands of applications quickly, reduce administrative workloads and, in theory, make hiring decisions more objective than humans. According to a recent review by MIT Technology, based on the 2024 study published in the Journal of Experimental Psychology titled “Costly Exploration Produces Stereotypes With Dimensions of Warmth and Competence”, large language models (LLMs) may not only inherit human biases from the data they are trained on but also develop entirely new stereotypes based on their own experiences. Rather than simply reflecting existing prejudice, these systems can create fresh patterns of discrimination as they attempt to optimise decision-making over time.The findings are particularly significant because businesses are rapidly integrating AI into recruitment. If these systems begin making assumptions about groups of applicants after only a handful of hiring decisions, they could gradually reinforce unfair employment practices without any explicit human instruction.The research, conducted by scientists at Princeton University and the University of Chicago, builds on earlier psychological work showing how stereotypes can emerge from repeated decision-making. It suggests that the same learning strategies that make AI effective at solving complex problems can also make it unusually prone to stereotyping job candidates.

U.S. pledges $5 billion to boost AI in government-backed scientific research– www.scientificamerican.com

pgnewser July 23, 2026 0

U.S. pledges $5 billion to boost AI in government-backed scientific research– www.scientificamerican.com
News Source
EXCERPT:

Overall federal funding for scientific research is set to decline in 2027, the most senior White House science figure said on Wednesday. The government, however, pledged to spend $5 billion on promoting artificial intelligence in science.

Michael Kratsios, the White House’s science and technology adviser, told the House Committee on Science, Space, and Technology that the money will be drawn from the National Science Foundation (NSF) and NASA, as well as other federal agencies.

The goal is to produce “the premier AI for science ecosystem in the world,” he said. The funding will be earmarked for “research awards, funding opportunities, specialized scientific datasets and research facilities,” the White House said in a statement.

China Explores Elevating Cognitive Effects into Its Vision of Warfare– warontherocks.com

pgnewser July 23, 2026 0

China Explores Elevating Cognitive Effects into Its Vision of Warfare– warontherocks.com
News Source
EXCERPT:

In March 2025, China’s military researchers published a concept describing how to rapidly capture and subdue the city of Taipei. Distributed in restricted military channels, this approach argues Beijing can achieve victory “by blocking the city’s ability to receive external assistance, restricting its ability to enable operations, breaking its operational system, collapsing cognitive cohesion, and eliminating its ability to recover.” Its objective is to break the Taiwan people’s will to fight.

While this concept is not yet doctrine, it is an important and rarely seen example of how China’s military strategy and research inform the way it plans to fight a war over Taiwan. The authors’ framework operationalizes China’s new military strategy of “national total war,” referring to whole-of-nation mobilization, and “cognitive domain operations,” referring to influencing the mental space where people and organizations make decisions. Details on both remain sparse, particularly in the context of how they may support the Chinese military’s primary mission of capturing Taiwan.

“All-domain megacity incapacitation warfare” (大型城市全域失能战) describes in remarkable detail how China’s military might capture Taipei by integrating military firepower with cognitive effects. An article from the March 2025 edition of Military Art, China’s leading internal journal on military theory, proposes this idea. The Central Military Commission oversees and approves this journal, and its official circulation is limited to restricted military channels. Because the journal is closely held within China’s military, the article’s authors have liberty to speak in specifics and without varnish as they inject their concept into China’s internal debate on how to plan a war on Taiwan.

OpenAI admits its agent went rogue and hacked AI startup Hugging Face– www.scientificamerican.com

pgnewser July 23, 2026 0

OpenAI admits its agent went rogue and hacked AI startup Hugging Face– www.scientificamerican.com
News Source
EXCERPT:

An OpenAI autonomous agent went rogue and hacked into another artificial intelligence (AI) startup’s infrastructure, the ChatGPT maker said in a blog post.

The agent, which was powered by some of OpenAI’s most advanced models, ran amok during a security test. It freed itself from confinement—a protocol AI labs use to insulate tests from the wider Internet—and get onto the internet. Once online, the agent tried to hack into Hugging Face, an AI startup that hosts open-source models and datasets.

“I think this is interesting as it shows the problem of mis-specified goals,” says Philip Torr, a professor of engineering science and AI safety expert at the University of Oxford. “The model wasn’t malicious it was just doing what it was optimized to do.”