返回新闻
工业AI Understanding 简报

《卫报》报道称,随着人工智能数据中心的扩张,美国天然气发电建设速度超过中国

据《卫报》报道,随着人工智能数据中心推动电力需求,美国正在开发的燃气发电能力激增,引发了人们对排放、能源成本和当地反对的担忧。

5 min readRead the original reporting
Source-provided image accompanying The Guardian reports US gas-power buildout is outpacing China amid AI data-center expansion
归因报告来源记录
出版商
theguardian.com
来源链接
theguardian.comhttps://www.theguardian.com/us-news/2026/aug/25/us-gas-power-china-ai-datacenter
来源类型
新闻媒体的报道——不是第一方文件。

我们无法独立确认的内容: 此声明归因于指定的商店。我们没有根据第一方文件对其进行验证。 (theguardian.com)

背景60 秒内了解这一点

从这里开始

关键术语

XAI(可解释的人工智能)
使人工智能预测更加透明和易于理解的技术和实践。
管道
预处理、模型步骤和后处理阶段的有序工作流程。
测试一下自己人工智能测验的未来

发生了什么

The Guardian reports that the United States is now developing twice as much gas-fired power capacity as China, reversing a long-standing pattern, as a surge in AI data-center construction drives demand for electricity. The figures come from a new analysis by Global Energy Monitor. AI Understanding has not independently verified the analysis, its projections or the cited estimates.

The Guardian reports that Global Energy Monitor found the United States is building twice as much gas-fired generating capacity as China and more than any other country. Under-construction US gas projects increased by 76% during the first half of 2026, according to the analysis. When announced and pre-construction projects are included, the US total is nearly three times China’s. The report presents this as a reversal of a pattern in which China had previously added gas generation faster than the US.

According to the figures cited by The Guardian, US gas-power capacity in overall development rose from 252 gigawatts in January to 378GW, a 50% increase. That represents about one-third of the global total. If every project were completed, the US gas fleet would grow by roughly two-thirds, at a reported capital cost of more than $647 billion. These are development figures and projections, not a record of completed construction or confirmed final investment decisions.

The Guardian says about half of the new US capacity is directly tied to upcoming data centers, which require large amounts of electricity for AI. The report describes a rush by technology companies, including Google, OpenAI and Amazon, to build data centers. It also says gas-turbine backlogs have led some companies, including xAI, to turn to smaller and less efficient turbines. The Guardian’s account does not independently establish the status, operating performance or emissions of each named project or facility.

Taken together, these figures describe a rather than a completed set of facilities. The reported comparison therefore concerns capacity at multiple development stages and should be read as an account of planned expansion. The analysis, as summarized by The Guardian, distinguishes current construction from projects that remain announced or pre-construction. That distinction is central to interpreting the reported reversal and to understanding why the projected total is larger than the amount already being built.

来源详情: theguardian.com ↗

为什么这很重要

The report links AI infrastructure growth to a major fossil-fuel buildout. If the projects are completed, the expansion could increase US power-sector emissions, expose ratepayers to volatile gas costs and intensify disputes over water, tax incentives, environmental review and local land use.

The report matters because it connects the physical expansion of AI services to a specific energy choice: new gas-fired generation. The Guardian cites an estimate that using gas rather than renewable power for the new data-center load could increase US power emissions by as much as 20%. The article does not provide enough detail in the supplied text to independently assess the estimate’s methodology, assumptions or time horizon.

The projected buildout could affect people who do not use the data centers directly. The Guardian reports that Global Energy Monitor’s project manager warned that new gas plants could lock in decades of pollution and dependence on volatile fuel costs, with those costs potentially passed to ratepayers. The article also identifies concerns about climate emissions, rising electricity bills, tax breaks for AI companies and water shortages. Those consequences remain contingent on projects being completed and operated as described.

The infrastructure debate also has a public-governance dimension. The Guardian reports that more than 800 data centers are planned and that two-thirds are in drought-stricken areas. It cites a Heatmap poll in which three-quarters of Americans reportedly said they would not want to live next to a data center. The article says dozens of cities and counties have imposed restrictions, while New York became the first state to enact a temporary ban on new hyperscale data-center permitting and construction in July. The polling, project count and legal status are attributed to The Guardian’s report and have not been independently confirmed here.

These reported effects depend on choices and outcomes that have not yet been settled. The source describes potential consequences associated with the projects and the way their costs are allocated, rather than documenting every consequence as an accomplished result. The same uncertainty applies to emissions, bills, water use and local responses. Reading the report this way keeps the significance clear while preserving the distinction between reported projections, attributed concerns and independently established outcomes.

Interactive Mechanism

互动机制:它实际上是如何运作的

以交互方式探索这一发展背后的基础技术。

Model Parameter Size:8B Parameters
VRAM Required5.5 GBGPU memory footprint
Target HardwareMacBook / Single GPUDeployment tier
Privacy100% Air-GappedLocal device capability
Core takeaway: Small, quantized models (3B–8B) now run directly inside smartphones and laptops with complete data privacy, while mammoth 400B+ models remain the domain of datacenter clusters.
交互式概念检查+10 Points
Future of AI Quiz

What should a useful AI forecast state?

接下来看什么

The key questions are how many proposed gas plants and data centers will actually be built, whether renewable or lower-emissions power can replace some of the planned gas capacity, and how regulators and communities respond. The Guardian also reports that turbine shortages, local opposition and new permitting restrictions could change the projections.

The first indicator will be the conversion rate from proposals to construction and operation. The Guardian’s headline figures combine projects at different stages, including announced and pre-construction developments. Local opposition, permitting limits, environmental review, turbine availability and financing could all reduce the eventual buildout. The supplied source does not identify which projects have secured permits, contracts or financing.

The choice of power source will remain consequential. The Guardian reports that the Trump administration has promoted new data-center construction and removed environmental reviews to speed development, while local governments have adopted restrictions. It also reports that the International Energy Agency expects US spending on coal- and gas-fired plants to exceed China’s for the first time in decades. AI Understanding has not independently verified the IEA forecast or the article’s characterization of federal policy.

Further reporting should establish whether data-center operators are securing dedicated gas generation, using grid power, contracting for renewable energy or relying on a mixture of sources. It should also examine who bears construction and fuel costs, how drought conditions affect proposed sites, and whether smaller turbines materially change emissions. The Guardian’s report establishes a significant reported shift in planned capacity, but it does not show the final energy mix, completed emissions or long-term electricity prices that will result.

The projections will become clearer as projects move through the stages identified in the report. Evidence about permits, construction, contracts, financing and operation would help distinguish announced capacity from capacity that reaches the grid. Comparable information about renewable procurement, gas supply and site conditions would clarify the eventual mix. Until that evidence is available, the reported development totals provide a signal to follow, not a final account of facilities, emissions or costs.

相关指南和测验

AI 的未来人工智能培训人工智能模型解释测试你所知道的——尝试免费的人工智能测验在我们的词汇表中查找人工智能术语关注 AI 资金追踪器
觉得这有用吗?