---
格式版本: 2
标题: "House Bill Would Put Federal Electricity Tax on Data Centers"
原文链接: "https://www.datacenterknowledge.com/regulations/house-bill-would-put-federal-electricity-tax-on-data-centers"
发布日期: "2026-08-19"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:configured_publication_date_rule"
发布时间证据: "datacenterknowledge-jsonld-date-published html:original: \"datePublished\":\"2026-08-19T14:03:01.000Z\""
发布时间校准原因: "信源发布日期识别规则直接确认发布时间"
发布时间校准置信度: "high"
发布时间候选数量: 2
发布时间严格候选数量: 2
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-20T01:36:46+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-20T01:34:59+08:00"
入库时间: "2026-08-19T17:36:46.759Z"
来源平台: "Data Center Knowledge 搜索"
搜索渠道: "source_template"
搜索词: "https://www.datacenterknowledge.com/search?q=AI%20Rack"
匹配关键词:
  - "AI Rack"
  - "AI"
相关厂家:
  []
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 25
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "文章核心为美国联邦对数据中心征收电税的议案，讨论税收政策、成本影响及资金分配，未涉及超节点/AI Rack架构、互连、供电散热或产品落地，与项目主题无关。"
AI质检模型: "ali-deepseek-v4-flash"
AI质检时间: "2026-08-20T01:36:56+08:00"
AI主题相关性: 2
AI来源权威性: 10
AI新颖性: 5
AI技术细节: 0
AI商业部署信号: 3
AI完整性: 5
AI摘要: "美国众议员Salinas提出《数据中心社区再投资法案》，拟对用电容量超1兆瓦的数据中心征收每千瓦时1美分的联邦消费税，预计年收入17.6亿美元，分配给五个联邦基金，包括住房、保护与基建等项目。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-09-07T03:16:20.362Z"
采集批次: "2026年8月20日0点36分19秒"
采集批次ID: "20260820-003619-923"
去重键: "https://www.datacenterknowledge.com/regulations/house-bill-would-put-federal-electricity-tax-on-data-centers"
---

H.R. 10102 would tax data center electricity at 1 cent per kWh, raising an estimated $1.76 billion annually for housing, conservation, and infrastructure.

Getty Images

A new House bill would impose a federal excise tax on electricity consumed by data centers, targeting facilities with more than 1 MW of power capacity and directing the resulting revenue toward five federal funds.

Rep. Andrea Salinas, D-Ore., has introduced the [Data Center Community Reinvestment Act of 2026](https://www.congress.gov/bill/119th-congress/house-bill/10102), which would amend the Internal Revenue Code to establish a 1-cent-per-kilowatt-hour federal excise tax on electricity used by qualifying data centers. The measure was referred to the House Ways and Means Committee, with additional referrals to the Energy and Commerce Committee and the Science, Space, and Technology Committee.

The proposal comes as rapidly expanding AI and cloud computing workloads are driving [large increases in electricity demand](https://www.datacenterknowledge.com/energy-power-supply/pjm-monitor-ai-data-center-growth-reshaping-power-markets) and intensifying debates over who should pay for the infrastructure and [community impacts](https://www.datacenterknowledge.com/data-center-site-selection/can-growing-community-backlash-quiet-the-ai-data-center-boom-) associated with data center growth.

Salinas said that the bill is intended to ensure communities are not left bearing the costs of rising data center demand and share in the benefits generated by that growth. “Every community deserves to drive a hard bargain with data center developers, and at the end of the day, every community deserves to share in the benefits from data center growth,” she said in a press statement.

The proposal would build on a state-level approach that [Virginia adopted](https://www.datacenterknowledge.com/regulations/virginia-approves-first-ever-data-center-power-tax) earlier this year. Virginia became the first state to impose a tax directly on data center electricity consumption, setting the rate at 1.1 cents per kilowatt-hour beginning July 1 and applying it to electricity supplied by utilities and retail providers as well as self-generated power, including [behind-the-meter generation](https://www.datacenterknowledge.com/energy-power-supply/why-data-centers-produce-their-own-power). State budget documents projected about $600 million in annual revenue for the general fund.

The federal bill would set a slightly lower rate but apply it nationwide. Unlike Virginia’s tax, however, Salinas’ proposal would allocate the revenue among five federal funds rather than sending it to a general fund.

## What Operators Would Pay at Scale

Neil Osnato, founder of Persistence Analytics Group, said the tax could be significant for large data center operators, but its broader implications depend on whether electricity consumption is an appropriate proxy for the costs individual facilities impose on the grid.

At continuous full load, a 100 MW data center would incur about $8.76 million annually under the tax; at 500 MW, about $43.8 million; and at 1 GW, about $87.6 million. At a 90% annual load factor (actual average usage as a percentage of maximum), those figures would be approximately $7.9 million, $39.4 million, and $78.8 million, respectively.

The tax could affect data center siting and power procurement at the margin, particularly when operators compare jurisdictions with otherwise similar power economics, Osnato said. For the largest AI and hyperscale projects, however, access to executable megawatts, transmission availability, interconnection timing, generation supply, and development certainty may matter more than the tax alone.

If a facility’s underlying electricity price were 5¢/kWh, the proposed tax would add 20% to the energy component. At 10¢/kWh, it would add 10%, Osnato said.

The bill defines a data center as a facility that primarily contains electronic equipment used to process, store, or transmit digital information and has a maximum rated power capacity or total peak power load exceeding 1 MW.

## Flat Tax vs. Grid Cost Causation

Osnato said the proposal should not be confused with a mechanism for assigning electricity system costs to the customers that cause them.

“The bigger issue is cost causation,” Osnato said.

Two 500 MW data centers could consume the same amount of electricity while imposing very different costs on the grid, depending on where they locate, how their demand coincides with system peaks, what transmission and substation upgrades they require, whether new generation must be procured, and how flexible their operations are, he said.

“A flat federal consumption tax does not distinguish among any of those conditions,” Osnato said.

Data centers already pay through multiple layers of the electricity system, including energy, transmission, distribution, and capacity costs, as well as [customer-specific interconnection or construction charges](https://www.datacenterknowledge.com/energy-power-supply/amazon-pushes-virginia-to-let-ai-data-centers-fund-their-own-grid-upgrades) in some cases. The policy question, Osnato said, is whether existing mechanisms fully capture the electrical consequences of new large loads.

“A federal tax does not substitute for that analysis,” he added.

## Revenue Split: Five Federal Funds

The legislation would distribute amounts equivalent to one-fifth of annual receipts to each of five destinations:

1. The Land and Water Conservation Fund
2. The Housing Trust Fund
3. The Hazardous Substance Superfund
4. The Highway Trust Fund
5. A new Energy Technology Trust Fund established by the bill

Salinas’ office projects about $1.76 billion in annual revenue based on current nationwide data center electricity use, allocating roughly $352 million per year for each of the five funds. The release says the Energy Technology Trust Fund would support federal loan guarantees for clean energy, advanced nuclear, and grid infrastructure projects.

## Tax Would Apply Broadly

The bill does not specify a separate rate for AI data centers. Instead, the tax would apply based on electricity consumption by facilities meeting the bill’s 1 MW threshold.

That threshold means the proposal reaches well beyond the largest hyperscale AI campuses. It could also apply to smaller colocation, enterprise, and institutional data centers with more than 1 MW of maximum rated capacity or peak load.

The legislation also covers electricity generated onsite, according to Salinas’ office.

Osnato said policymakers should distinguish between electricity consumption and its consequences for the grid.

“The questions that matter are: Is the projected load real? When will it actually materialize? How durable will it be? What generation, transmission, and local infrastructure does it cause?” Osnato said.

He also said policymakers should determine which costs are customer-specific versus genuinely regional and whether a facility’s operating characteristics increase or reduce system burden.

“Megawatts are not interchangeable,” Osnato said.

## Tax Revenue Doesn’t Assign Actual Grid Costs

Osnato said customer-specific costs should follow the customer that causes them, while genuinely shared system costs should be shared to the extent that the benefits are systemwide. The bill instead uses electricity consumption as a proxy for the costs and impacts associated with data center growth, while directing the revenue to five federal programs that are not specifically tied to the electricity infrastructure serving the facilities.

Only the proposed Energy Technology Trust Fund has a direct energy connection, and its proceeds would support federal loan guarantees rather than specifically funding the local or regional grid infrastructure created by the taxed facility.

“Policymakers could collect money from the right industry while still allocating the underlying grid costs incorrectly,” Osnato said.

If enacted, the bill’s amendments would apply to electricity used and taxes received after the date of enactment.

The proposal remains at the introduced stage after its Aug. 13 referral to the three House committees.