Is it True That “Jobs Will Increase”?

Local governments are rushing to attract AI data centers. “Job creation” and “increased tax revenue”—it sounds appealing. However, what is happening behind the scenes? Local governments are forced to sign NDAs (Non-Disclosure Agreements), and information is not disclosed to residents. Electricity costs rise, water is drained, and taxes are exempted. The ones profiting are the large corporations, while local SMEs and residents bear the brunt.

This is a story currently unfolding in the United States, and it is not irrelevant to Japan. Let’s break down the structure.

The Right to Know Disappears with NDAs

The first major issue is the NDA.

In the U.S., it has become commonplace for local governments to enter into NDAs with companies when attracting AI data centers. A 2024 investigation by the U.S. Senate Committee on Commerce, Science, and Transportation revealed that multiple municipalities had entered into contracts with tech giants that kept details about electricity usage, water resource consumption, and tax incentives confidential.

What does this mean? Local SMEs and residents cannot even know how much of their local resources are being drained. They have no idea how much electricity is being used, where the water is going, or how much tax is being exempted.

“Lack of information” means an inability to negotiate. This is fatal for SMEs. Even if electricity costs rise, they cannot verify “why it has risen.” When they inquire with the administration, they are met with “We cannot respond due to the NDA.” The fundamental principle of democracy, information disclosure, is sealed by a company’s confidentiality agreement.

Electricity Costs Rise—A Direct Hit to SMEs

Next, let’s talk about electricity.

The electricity consumption of a single AI data center is said to be equivalent to that of approximately 50,000 households. According to an environmental report released by Google in 2024, the electricity consumption of its data centers increased by 17% compared to the previous year. This is due to the explosive increase in GPU utilization with the spread of generative AI.

Where does this electricity come from? The local power grid.

Data from the U.S. Energy Information Administration (EIA) indicates that in Loudoun County, Virginia, where data centers are concentrated, industrial electricity rates have risen by about 15% over the past five years. The same applies to residential rates. If electricity demand exceeds supply, prices will rise. This is a natural consequence.

What does this 15% increase in electricity costs mean for SMEs? In manufacturing, electricity costs account for 3-8% of sales. If electricity costs rise by 15%, profit margins could be reduced by 2-3 percentage points. For a small factory operating on a 5% profit margin, this could easily lead to a loss.

Large corporations have the financial power to procure renewable energy on their own. They can secure electricity at long-term fixed prices through Power Purchase Agreements (PPAs). However, a manufacturing business with 10 employees lacks such negotiating power. When electricity costs rise, they simply absorb the increase. This is the structure.

The “Exploitation of the Region” Under the Guise of Tax Incentives

Even more serious is the issue of tax incentives.

Data center companies demand reductions in property taxes, exemptions from sales taxes, and reductions in corporate taxes as conditions for their attraction. In the U.S., Virginia has implemented tax incentives for data centers amounting to approximately $100 million (about 15 billion yen) annually. Similar movements are spreading in Texas, Ohio, and Georgia.

A 2024 Senate investigation reported a case where a data center company received about $200 million (about 30 billion yen) in tax deductions over ten years while employing only about 50 full-time workers locally. The tax deduction per person amounts to about 40 million yen annually. Meanwhile, local SMEs spend about 4 million yen annually to employ one person, while giant tech companies are exempted from 40 million yen in taxes per employee.

Who will fill this tax revenue gap? The remaining residents and SMEs. Increased property tax rates, reduced public services, and postponed infrastructure development—the costs will be passed on to the entire community. Although it was expected that the arrival of data centers would increase tax revenue, the reality is the opposite.

Water Disappears—A Lifeline for Agriculture and SMEs

AI data centers consume large amounts of water for server cooling. According to Microsoft’s 2023 environmental report, the water consumption of its data centers increased by 34% compared to the previous year, reaching approximately 6.4 million cubic meters annually. This is equivalent to the annual usage of about 12,000 households.

It has been reported that the data center built by Meta (formerly Facebook) in Mesa, Arizona, consumes about 57 million gallons (approximately 216,000 cubic meters) of water annually. This is a significant amount in a dry region.

Water is life for local agriculture. The same goes for small food processing businesses. If data centers suck up large amounts of water, it will become difficult to secure agricultural water, and water rates will also rise. As a result, the production costs of agricultural products will increase, directly impacting local food-related SMEs.

Moreover, due to the NDA, local residents cannot know how much water the data center is actually using. They remain unaware as the water levels in wells continue to drop.

The Illusion of “Job Creation”

The biggest selling point for attracting data centers is “job creation.” But what is the reality?

During the construction phase, indeed, hundreds to thousands of jobs are created. However, once the operational phase begins, the number of full-time staff decreases to just a few dozen. This is due to the highly automated nature of the facilities. Moreover, the skills required for those few dozen positions are primarily specialized roles such as network engineers and security specialists, which are difficult to source from the local labor market.

As a result, reports from various regions indicate that while “jobs have increased, local people were not hired.” Once construction is complete, temporary jobs disappear, leaving only towns burdened with higher electricity and water bills.

Implications for Japan—Breaking Away from “Attraction First”

This structure is likely to be replicated in Japan as well.

Starting in 2024, a construction rush for AI data centers is beginning in Japan. Large-scale data center plans are underway in Hokkaido, Chiba, Osaka, and Kyushu. The government is also positioning AI infrastructure development as a pillar of its growth strategy, offering subsidies and tax incentives.

However, what the U.S. examples show is the cold reality that “attraction does not equal regional revitalization.”

Local SME owners need to consider three things:

1. If there are data center plans in your area, check the impact on electricity and water resources immediately.
You should request information disclosure from the local government through the Chamber of Commerce. It will be too late once it is hidden by an NDA.

2. Incorporate the risk of rising electricity costs into your business plan.
If a data center comes, electricity demand will certainly increase. Simulate what would happen if your electricity costs rose by 15-20% now.

3. Verify the “benefits of attraction” with concrete numbers.
When figures such as “X number of jobs” or “X billion yen in tax revenue” are presented, question whether these are full-time positions or temporary jobs, and whether they are net increases after tax deductions. Do not be deceived by abstract “economic effects.”

In the End, Who Benefits from the Attraction?

Attracting AI data centers is merely a “search for locations” for tech giants to secure cheap electricity, water, and tax incentives. For them, local governments are nothing more than “cheap land.”

On the other hand, for local governments, it is a vessel for the desire that “something big will come and revitalize the region.” However, if that “something big” drains local resources, seals information with NDAs, is exempt from taxes, and creates little to no employment, then it is not attraction but exploitation.

What local SMEs truly need is not the attraction of giant data centers, but how to use AI in their own businesses, how to reduce their costs, and how to increase their productivity.

Instead of calling tech giants with a 30 billion yen tax deduction, it would have a much greater impact on the local economy to support 100 local SMEs in adopting AI with just 3 billion yen, which is 1% of that amount.

When the topic of attraction arises, before thinking “this is great,” ask “So, how much will our electricity costs go up?” A region that can ask this question will be the one that survives.