字号 ·· | 护眼
国会山报

揭穿围绕数据中心的政治恐慌言论Debunking political fearmongering over data centers

点「原文对照」整页切到原文,或双击某段只看那段的原文。

新技术的出现带来了新的担忧,而这些担忧往往被出于政治目的而加以利用。当前关于人工智能(AI)和数据中心的争论就是一个典型的例子。人们确实有合理的疑问,比如数据中心的用水量、能源消耗以及土地使用情况;美国的农民、牧场主和其他民众提出这些问题是完全正当的。然而,与以往许多情况一样,这场争论的真正目的与其说是为了寻找真相,不如说是为了推动某种特定的政治议程。

The advent of new technologies brings about new anxieties, which are too often exploited for political reasons. Current contentions over AI and data centers are a perfect example. There are legitimate questions concerning water usage, energy requirements and land use; America’s farmers, ranchers and other citizens are justified in asking them. But like so much these days, the debate has less to do with honestly discovering answers than with pushing an agenda.

这场讨论越来越多地被某些激进的政治势力所操控——这些势力对美国农村地区来说再熟悉不过了;他们反对数据中心的建设,实际上掩盖了他们对自由市场、私营企业以及美国全球主导地位的普遍反对态度。让我们客观地分析一下这些问题吧。

The conversation is increasingly being hijacked by an activist political faction that rural America is all too familiar with, whose opposition to data centers masks their general opposition to free markets, private enterprise, and the very idea of American dominance. Let’s honestly examine the issues.

在用水量方面,劳伦斯伯克利国家实验室发布的《美国数据中心能源使用报告》显示:2023年,美国的数据中心直接消耗了约660亿升水(相当于每天4780万加仑)。这个数字听起来确实很大,但如果没有具体的背景信息,这些数据其实并没有太大意义。作为对比,美国地质调查局估计2015年(该机构最后一次进行全国灌溉用水量统计的年份)农业灌溉的用水量为每天732亿加仑。

In terms of water usage, Lawrence Berkeley National Laboratory’s U.S. Data Center Energy Usage Report informs us that in 2023, American data centers directly consumed approximately 66 billion liters of water — roughly 47.8 million gallons per day. That sounds like a lot of water; but the figures are meaningless without context. By comparison, the U.S. Geological Survey estimates that agricultural irrigation used 73.2 billion gallons per day in 2015, the last year the federal agency conducted an estimate of national irrigation water consumption.

换句话说,数据中心的用水量仅占农业用水量的0.065%左右。不过,这些数据仍然没有揭示全部真相:数据中心的用水量还受到多种因素的影响,包括当地的气候条件、可用的水源以及数据中心所采用的技术类型。例如,许多数据中心采用了“闭环系统”,这种系统并不需要持续供水。那么能源消耗情况呢?

In other words, data center water consumption is equal to about 0.065 percent that of agriculture. Even those numbers don’t tell the full story; water use by data centers is dependent on several factors, including local climate, available water source, and the type of technology the data center incorporates. Many data centers, for example, incorporate closed loop systems that do not require a continual supply of water. What about power requirements?

在需求增长与目光短浅的公共政策共同作用下,我们的电网已经不堪重负。这些政策包括所谓的“可再生能源”和电气化强制令,它们在让本已较为充裕的电源退出的同时,人为地将需求推高至可用容量之上。我们是否有理由新增如此庞大的用电负荷?尽管数据中心确实需要巨量能源,但认为这些能源必须由现有电网供应,或通过电力用户补贴来提供,是一种误导。

Our electrical grid is already becoming overwhelmed amid a combination of increased demand and short-sighted public policies, including so-called “renewable” and electrification mandates that artificially increase demand beyond available capacity, even while taking more abundant sources offline. Can we justify an enormous new electrical draw? Although data centers do require a tremendous amount of energy, it is misleading to suggest that it must automatically be supplied by the existing grid or ratepayer subsidization.

事实上,我们的国家政策规定,任何造成大量新增用电需求的实体都应当自行承担相应成本。特朗普政府的“电力用户保护承诺”要求科技和人工智能公司为其所需的任何电力输送基础设施付费,不得将网络升级成本转嫁给家庭,并且必须为任何新增发电能力和合同购电付费,即使最终实际使用量低于合同约定量。众议院刚刚通过了由科罗拉多州共和党众议员加布·埃文斯提出的《电力用户保护法案》;如果参议院也采取相应行动,该承诺将被写入法律。

In fact, it is our national policy that any entity creating substantial new electrical demand ought to pay for it. The Trump administration’s Ratepayer Protection Pledge calls for tech and AI companies to pay for whatever power-delivery infrastructure they require, not pass network-upgrade costs on to households, and to pay for any new generation and contracted electricity, even if they end up using less than they contracted for. And the House of Representatives just passed the Ratepayer Protection Act, sponsored by Rep. Gabe Evans (R-Colo.), which will codify the pledge into law if the Senate follows suit.

除了现有电网升级措施外,数据中心还可以利用多种新技术。模块化核反应堆和其他模块化发电机、余热回收技术以及改进的电池技术,都有助于缓解当地电网的压力。这种现场发电方式构成了微电网的骨干,代表着高耗能行业和高需求社区未来电力配送的发展方向。就占地面积而言,房地产服务公司仲量联行估计,一座容量超过1吉瓦的大型数据中心可能需要占用500至800英亩土地(也就是每英亩0.5至0.8吉瓦)。

Data centers also have access to new technologies beyond existing grid improvements. Modular nuclear reactors and other modular generators, as well as waste heat capture and improved battery technology, can help alleviate pressure on local grids. This type of on-site generation forms the backbone of micro-grids, representing the future of electrical distribution for high-demand industries and communities. In terms of land footprint, real estate services company JLL estimates that a large data center with a capacity of greater than one gigawatt could require between 500 and 800 acres (that is, between 0.5 and 0.8 gigawatts per acre).

美国能源部2026年7月发布的研究显示,到2030年,新增数据中心容量的需求最高将达55吉瓦。

A July 2026 study by the Department of Energy informs us that there will be a demand for up to 55 GW of additional data center capacity by 2030.

每吉瓦(GW)数据中心建设需要0.8英亩的土地,这意味着总共需要44,000英亩的土地来建设数据中心。鉴于美国目前拥有8.7395亿英亩的农田,只有0.005%的农田可以被用于数据中心建设。因此,我们必须认真考虑数据中心驱动的人工智能技术对农业、社区以及整个国家所带来的潜在益处。农业科学技术委员会(Council for Agricultural Science and Technology)的一项新研究指出,人工智能技术对农业和粮食安全的重要性已经变得极为突出;该技术已经从一个具有前景的研究课题,发展成为一种正在重塑农业及整个食品系统的基础性技术。

At 0.8 acres per GW, that means 44,000 acres will be required for data center development. At America’s current 873.95 million acres of farmland, only 0.005 percent of U.S. farmland may be used for data center development. Ultimately, the potential benefits of data-center-powered artificial intelligence to agriculture, our communities and nation need to be considered. A new study by the Council for Agricultural Science and Technology reveals just how important AI has become for agriculture and the security of our food supply, pointing out that the technology “has moved beyond a promising research topic to a foundational technology that is reshaping agriculture and the broader food system.”

对于美国的农民和牧场主来说,人工智能在精准灌溉、化肥和农药施用、作物病虫害检测、牲畜健康监测、产量预测、天气及市场风险分析等方面发挥着不可或缺的作用。对于美国农村地区而言,数据中心不仅创造了新的就业机会,还为当地经济注入了资金。作为国家,我们需要问自己:我们是想利用自身的优势来主导这一新兴领域,还是将其拱手让给那些可能并不真正关心美国利益的其他国家?

For our farmers and ranchers, AI can be indispensable in providing precision irrigation, fertilizer and pesticide application, crop-stress detection, livestock-health monitoring, pest and disease identification, yield forecasting, weather and market-risk analysis, and a host of other services. For rural America, data centers offer an economic lifeline, creating new jobs and injecting new capital into a community. As a nation, we need to ask ourselves: do we want to employ America’s inherent strengths to dominate this new field, or cede it to someone else, another country who may not have America’s best interests at heart?

美国人完全可以像以往一样应对这一挑战——依靠我们对自由市场的支持、企业家精神,以及对私有财产和法治的尊重;但前提是我们必须拒绝让政治议程主导讨论的方向,而是要客观、准确地分析这些挑战。斯科特·蒂普顿(Scott Tipton)议员曾在2011年至2021年间代表科罗拉多州第三选区担任国会议员,并担任众议院农业、畜牧业和自然资源委员会(House Agriculture, Livestock, and Natural Resources Committee)的成员。

Americans can meet this new challenge as we always have, leveraging our embrace of the free-market, entrepreneurial spirit, and institutional respect for private property and the rule of law — but only if we refuse to let the conversation be hijacked by a political agenda, and examine the challenges honestly and accurately. Congressman Scott Tipton served Colorado’s third congressional district from 2011 to 2021 and was a member of the House Agriculture, Livestock, and Natural Resources Committee.