Voters approved 40 of 41 proposed ballot measures that would restrict some of the more than 2,000 new AI data centers planned nationwide, with water and power topping ratepayers' concerns. Measuring site-level water use is hard because operators usually report companywide figures, and Shaolei Ren of UC Riverside says a 1-gigawatt AI data center can consume up to 10 million gallons of water per day, depending on hardware, location and cooling method. Modern AI data centers use closed-loop liquid cooling, but total water use depends largely on how heat is released outside the building. Dry cooling uses far less water but can need much more power in hot climates, while evaporative cooling uses little power yet can lose large amounts of water on the hottest days. Equinix consumed 1.2 billion gallons of water across its operations in 2025, up 10% from a year earlier.
Summary written by Upside AI from the original article. Read the source for full details.
Keywords
data centersartificial intelligenceelectricityregulation
Open a keyword to see other news on the same topic.
Related tickers and likely impact
NVDA
• NeutralNvidia
Relevance: low
| Market | US stocks |
|---|---|
| Sector | Information Technology |
| Industry | Semiconductors |
| Market cap | $5.1T |
| Close on 11/04/2026 | 200.28 |
| Change that session | -3.10% |
Impact is scored by AI from the headline and may be wrong. It is not investment advice.
Related news
- Musk Says AI Is Short on Power. Here’s Why Constellation Energy Still Has a Rules Problem
Constellation Energy (CEG) closed up 4.60% at $270.74 on October 21, 2026, a day after SpaceX CEO Elon Musk said the constraint on deploying superintelligence lies more in power, cooling and infrastructure than in chips. The bigger issue came on October 9, when FERC rejected most of PJM's data center framework, three days after Google signed a 20-year, 890-megawatt uprate deal with Constellation. Management had said regulatory ambiguity was the worst thing for deal execution, but added that already signed contracts did not depend on the PJM and FERC outcomes. The mean analyst target slipped to around $338 from around $343, below Morgan Stanley's new $378. Third-quarter results, expected in early November, will show whether the FERC decision has slowed new contracting.
- Trump Wants Data Centers to Pay a “Dividend” to Their Neighbors
On October 21, 2026, President Donald Trump posted on Truth Social that big data centers should pay host communities a “dividend” and that he had “mandated” them to make their own electricity. No executive order, rule, dollar amount or enforcement program has been issued, and the posts do not say who pays or who qualifies. Cash offers have not always won local support: developers of a proposed Amazon Web Services campus in Hazle Township, Pennsylvania, offered each resident $10,000, which residents rejected in October. The article says Bloom Energy’s fuel cells would benefit most if on-site power spreads, though its $70.25 billion market value already assumes it wins; GE Vernova’s $176 billion backlog depends less on the policy, since data centers make up only 20% of its contracted gas gigawatts. The Federal Energy Regulatory Commission also declined in October to require data centers to bring their own generation.
- Trump says data centers should generate their own electricity — but experts say only states can mandate that
President Trump said data centers should generate their own electricity, but experts say only states can mandate it. His remarks come as voter backlash against the facilities intensifies ahead of the midterm elections.
- Microsoft (MSFT) Is Facing Fresh Regulatory Pressure
Microsoft agreed a 20-year power supply deal with NiSource for 1,100 MW to serve its expanding AI data centers. Meanwhile, the UK Competition and Markets Authority is preparing to label parts of Microsoft's business software as having strategic market status, and Microsoft, Apple and Palantir face possible new US Congressional oversight and hearings on their partnerships and operations. The article says the power deal helps reduce execution risk for the roughly US$175b to US$190b of 2026 capex that Microsoft's investment narrative relies on, while the regulatory moves add political and regulatory risk. Key items to watch are future disclosures linking data center expansion and power contracts to AI and Azure usage, along with any quantified costs from UK rules or US hearings.
- America's growing digital economy requires power grid updates. Who will foot the bill?
US lawmakers are debating who should pay for grid upgrades needed by large data centers. The Ratepayer Protection Act fell short of the 60 votes needed to advance in the Senate, with 57 in favor and 43 opposed, though the House had passed it 417–3. The bill would require facilities with peak demand of at least 100 megawatts to cover the full incremental costs of power generation and grid upgrades, with each state deciding whether to adopt the standards. The author from the Crypto Council for Innovation argues that households and small businesses should not bear these costs, and points to the Ratepayer Protection Pledge signed by major tech companies and a White House expansion bringing in over 200 more participants, while calling for faster permitting and infrastructure construction.
- AI data centers have a bigger problem than water
Host Ejaaz Ahamadeen argues that concerns about AI data centers' water use are overstated, noting that facilities now recycle the same water and that the author of an earlier viral claim later corrected herself. Citing a SemiAnalysis review, he says one or two In-N-Out restaurants use more water than Elon Musk's Colossus 2, one of the largest AI data centers. He says electricity demand is the more valid concern and points to Microsoft and, he believes, Meta pledging to cover and subsidize local electricity bills if they rise. He predicts the hyperscalers will simply pay those costs, though he says he will believe the pledges when he sees them.
Related markets
Latest news
- Sam Altman Says It's 'Fucking Annoying' His RAM Comments Were Taken Out of Context
OpenAI CEO Sam Altman said it was 'fucking annoying' that his comments about computer memory were taken out of context after Elon Musk mocked them. Altman had said it was 'quite terrible' for the world's limited DRAM supply to sit in home computers used 5% of the time, and he said he has long argued for more memory production. He briefly renamed his X account 'Ram Altman' in response to users' jokes. The remarks come amid a global memory shortage driven by AI data center demand, with contract prices for 16GB DDR5 chips reportedly rising from $8.70 in October 2025 to $48 in September 2026, an increase of more than 550%.
- Blockchain is ‘perfect solution’ for AI agent trust, VeChain CEO says
VeChain co-founder and CEO Sunny Lu said AI agents could drive blockchain's next adoption cycle, since autonomous agents would need systems to verify their identities and reputations, calling blockchain 'the perfect solution'. He spoke on Cointelegraph's Chain Reaction show while promoting VeWorld.ai, an agent-building platform that uses VeChain's blockchain. In his vision, agents would carry digital 'passports' within a decentralized credibility system that tracks their reputations and those of their creators, with smart contracts settling transactions between agents or with people. Lu also argued that agent-building tools could let non-developers turn their expertise into businesses, comparing this to mobile app stores, while Animoca Brands chairman Yat Siu warned that users risk building strategies around services they do not control.
- Apple cuts staff AI usage in scramble to rein in costs
Apple has restricted employees' AI use to cut costs. According to Bloomberg, daily usage caps for engineering staff were reduced from $300 (£223) to $200, and the company is reportedly considering lowering the engineering cap to $100 and cutting non-technical staff's daily budgets from $100 to $25. Apple joins other companies that have curbed AI spending, including Revolut, Accenture, Uber, Tesla, Microsoft and Meta, amid concerns about costs and productivity. The limits apply to 'tokens', units of AI consumption, even as Apple pushes staff to use more AI to boost productivity. The changes are among the first by CEO John Ternus, who succeeded Tim Cook in September, and the trend has raised concerns about the planned IPOs of Anthropic and OpenAI, with investors saying Anthropic may list at a valuation of $1.5tn or less, below earlier hopes of $2tn. Apple did not comment.
- Chinese AI lab Moonshot turned to Europe for Nvidia Blackwell access amid U.S. China curbs, sources tell CNBC
Moonshot AI, a Chinese AI lab that the White House said accessed Nvidia's Blackwell chips via a facility in Thailand, has reportedly sought access to such chips through data centers in Europe, according to CNBC sources. Talks this year reportedly covered renting capacity from non-Chinese neocloud operators, including clusters of 8,000 or more Blackwell GPUs for inference and possibly training. One source said Moonshot was proposed to pay 30% to 50% more upfront than a non-Chinese customer. U.S. export controls restrict shipments of Nvidia's most advanced chips to China, but remote access through overseas cloud services is not broadly regulated, and the Remote Access Security Act passed the House in January but has not cleared the Senate. The Financial Times reported that Nvidia told cloud providers to sell at least 51% of their services to non-Chinese customers. Moonshot and Nvidia did not respond to requests for comment.
- TSMC’s AI Opportunity Is Expanding. Here’s the Question That Matters for Its Stock Price
TSMC posted its third straight earnings beat, with third-quarter EPS of $4.60 against a $4.46 consensus and revenue of $46.72 billion, up 41.2% in dollar terms. It guided fourth-quarter revenue to $52.3 billion to $54.3 billion. Even so, shares fell 10.97% over the past month and closed at $406.13. CFO Wendell Huang said that if the company executes well, revenue growth will outpace capex growth. The 24/7 Wall St. price target is $516.16, implying 27.09% upside, with a bull case of $536.47 and a bear case of $432.68, and the article flags margin dilution from overseas fabs as a risk, with gross margin down 1.3 points sequentially and guided margins of 65% to 67%.
- Two in three US-listed companies pay no dividend, a census of SEC filings finds
A census of SEC filings by Intrinsiqq, a free stock research site, found that only 35.2% of 4,631 US-listed companies paid a dividend as of October 16, 2026, with Amazon, Berkshire Hathaway and Tesla among the non-payers. Utilities had the highest share of payers at 69.2%, and 52.6% of those had raised dividends for at least ten consecutive years, yet only 20.3% of utility payers (15 of 74) generated enough free cash flow to cover the payout. Across dividend payers outside banks, insurers and REITs, 76.5% had free cash flow covering their dividends. Nearly half of US-listed companies (46.7%) were unprofitable over the past year. In the S&P 500, 81.3% pay dividends according to S&P Dow Jones Indices for Q4 2025, a figure that counts share issues rather than companies.
Get the briefing on Telegram or Zalo
Join the Upside Telegram channel or Zalo group for the Vietnam stocks and gold briefing at 8:00 and 19:00 (Vietnam time) on trading days, plus a weekly wrap on Saturday morning.