Category: Power & Infrastructure

Data centres, energy, water, finance, ownership, public costs and the physical systems supporting artificial intelligence.

  • Google Uses About a Third of This City’s Water. Who Decides What Comes Next?

    Google Uses About a Third of This City’s Water. Who Decides What Comes Next?

    INVESTIGATION | RESPONSIBILITY & RISK

    Google’s data centres used about one-third of the water consumed in The Dalles, Oregon, in 2024. The company has helped fund the city’s water system. The city says its long-term water plans are for the whole community. The unresolved question is what rules protect that community as one customer’s demand continues to grow.

    In The Dalles, Oregon, a question about artificial intelligence has become a question about water.

    Google opened its first data centre in the city in 2006. By 2024, its three local sites were using about one-third of all the water consumed through the city system.

    That year, Google used 434.4 million gallons. That is about 1.19 million gallons a day.

    The figure does not prove that one-third of the city’s water was used for AI. Google’s facilities support many computing services, and the public record does not identify what share of their local workload is related to AI.

    But the physical infrastructure supporting the wider digital and AI economy is expanding. Google completed another building in The Dalles in 2025 and was expected to complete another in 2026.

    The public question is therefore no longer whether the company uses a significant amount of water. It is who decides how much more water can be committed, what evidence must be public, and whose needs take priority when supply tightens.

    The number grew while the city changed around it

    Google’s local water use was about 104 million gallons in 2012, equal to roughly 12 per cent of the city total. By 2024, it had risen to 434.4 million gallons and about one-third of city consumption.

    The Dalles grew too, but more slowly. Portland State University estimates cited by Oregon Public Broadcasting put the city’s population at about 16,200 after an increase of roughly 1,700 people over 15 years.

    These figures do not establish that Google caused every change in the city’s water planning. They do establish that its demand is now too large to be treated as a minor industrial detail.

    Google is the city’s largest water customer. Its current annual use is close to the one million gallons a day assigned in the city’s 2024 master plan to a future unnamed industrial user. The city says that user is not Google.

    That distinction matters. Immortal AI has found no public evidence proving the unnamed user is Google, and the article should not imply otherwise.

    It also leaves a wider question. If the city is planning for another industrial customer of similar scale while Google’s existing campus is still expanding, has the public seen a complete cumulative test of future demand?

    The city’s case is stronger than a simple corporate giveaway story

    The City of The Dalles rejects the claim that its watershed plans are being driven by Google.

    It says long-term reservoir expansion was considered decades before data centres arrived. Its current watershed proposal would transfer about 150 acres of federal land surrounding the Crow Creek reservoir and related infrastructure to the city.

    The city says the proposal does not expand its water rights, authorise construction or create a dedicated water supply for any private company. Any future reservoir work would require separate environmental review, permits, Tribal consultation and public engagement.

    The city also says its municipal watershed provides around 70 to 80 per cent of annual supply and that long-term planning must address population growth, reduced snowpack, drought and wildfire risk.

    Those are legitimate public responsibilities.

    Google has also contributed material infrastructure. The city says the company paid normal industrial water rates, transferred water rights, funded an aquifer storage and recovery system, and delivered about $28.5 million in publicly owned improvements including wells, reservoirs, a pump station and a sewer lift station.

    Those contributions complicate the story. They can improve public capacity and reduce the amount other customers must fund. They can also deepen the city’s dependence on a customer whose future demand is commercially important to both parties.

    Paying for infrastructure is not the same as proving that future allocation is fair, resilient or transparent.

    The dispute is about purpose, not only volume

    The city says the watershed land transfer and possible future reservoir expansion serve the whole community. Mayor Richard Mays told OPB that Google was “really not a factor” in the proposal.

    Environmental groups and some residents are not convinced. They point to Google’s growing use, the planned campus expansion and the ecological importance of the Dog River system for threatened fish.

    The proposed legislation itself would not increase withdrawals. The harder question comes later: what additional storage may be built, what water would fill it, what environmental conditions would apply, and how future supply would be allocated among homes, public services, existing businesses and large industrial customers.

    This is where a public decision can become difficult to examine.

    The city operates one municipal system. Water added to that system may benefit the whole community. It can still be planned, financed or reserved around the needs of a very large customer.

    The correct test is not whether a project has Google’s name on it. It is whether the decision records show what demand drove the project, who receives reliable access, who pays, what happens in drought, and whether approval reopens when demand changes.

    The records were once treated as a trade secret

    Public trust is also shaped by how the one-third figure became known.

    The City of The Dalles initially refused a public-records request for Google’s local water use and argued that the information was a trade secret. After the Wasco County district attorney ordered disclosure, the city filed proceedings seeking to prevent release.

    The dispute ended in a settlement. The city agreed to disclose a decade of Google water-use records and similar annual information in response to future requests. Google later said it would no longer seek to protect site-level water figures as trade secrets.

    The records then showed how concentrated the demand had become.

    This history does not prove that current city statements are wrong. It does explain why assurances alone are insufficient.

    A community should not need litigation to learn how much of a shared resource is being used by its largest customer.

    One-third is serious, but it is not a universal data-centre number

    The Dalles should not be used to claim that every data centre consumes one-third of its host city’s water.

    Hillsboro, also in Oregon, reported that its data centres used about 111 million gallons in 2025. That was 1.76 per cent of total city demand.

    Virginia’s legislative audit found another pattern. Data centres used less than 0.5 per cent of statewide withdrawals in 2023, but at six utilities reviewed they accounted for between 2 and 21 per cent of system use when reclaimed water was excluded.

    The contrast is the lesson.

    National and statewide percentages cannot decide whether a particular city, aquifer or watershed can absorb a project. Cooling design, climate, project scale, source water, reclaimed-water capacity and competing demand all change the result.

    A corporate water-positive target cannot answer that local question either. Replenishment elsewhere may have value, but it does not guarantee water in the host community during a hot summer or drought.

    What a defensible decision would show

    The Dalles dispute points to a standard that should apply before any community commits land, incentives or infrastructure to a large data-centre project.

    The public record should show:

    • The project’s average, peak-day and peak-season water demand.
    • The water source, legal entitlement, spare capacity and competing committed uses.
    • Cumulative demand from every existing, approved and reasonably foreseeable large customer.
    • The cooling design, expected direct consumption and the trade-off between water and electricity.
    • Drought rules identifying who must reduce use, when restrictions begin and whether drinking water can become a fallback supply.
    • The cost and ownership of new wells, reservoirs, treatment, pipes and monitoring.
    • Actual facility-level use reported at least annually, with enough detail to test projections.
    • Enforceable limits, audit rights and a requirement to reconsider approval if the project or its demand changes materially.

    This is close to the approach Texas adopted in August 2026. Its governor directed regulators to obtain projected annual and peak water consumption, supply sources and cooling technology before data-centre projects could move forward through the grid-connection process.

    The Texas process is new and its effectiveness remains untested. Its importance is the decision gate: provide the evidence before approval, not after the infrastructure is committed.

    Who decides what comes next?

    The Dalles does not present a simple choice between a technology company and household taps.

    Google pays for the water it uses. It has transferred assets and water rights to the city. The city has long-standing reasons to plan for drought, growth and wildfire. The data centres also bring jobs, tax revenue and economic activity.

    But payment does not settle priority.

    When one customer uses about one-third of a city’s water, its future demand becomes a public governance issue. Decisions about added storage, new supply and drought protection can affect residents, other industries, future housing, downstream users and ecosystems for decades.

    The people making those decisions should be publicly accountable. The evidence should be disclosed before commitments are locked in. The agreement should say what happens when the assumptions fail.

    If the city is right that future water projects serve the whole community, transparent cumulative modelling and enforceable allocation rules should help demonstrate that.

    If those records remain incomplete, the public is being asked to trust a conclusion it cannot test.

    What communities can do now

    • Ask the utility for the largest industrial users’ annual and peak demand, subject only to narrow security or privacy redactions.
    • Request a cumulative water assessment covering existing, approved and proposed data centres and other major users.
    • Separate infrastructure that is funded and operational from future promises, replenishment targets and voluntary goals.
    • Require drought, expansion and project-delay scenarios before approvals or public spending are final.
    • Record who pays, who owns the assets, who receives priority and what triggers a new public decision.

    Use the Data Centre Community Hub to structure the evidence request and record what remains unanswered.

    The question is not whether Google should receive water

    A city can decide that a data centre’s economic and infrastructure benefits justify substantial water use.

    That decision should be made openly, using local evidence and enforceable rules.

    The Dalles shows why averages and sustainability pledges are not enough. A company can pay its bills, fund infrastructure and still become large enough to reshape the choices available to everyone else.

    Google used about one-third of the city’s water in 2024. The next question is not whether that number sounds large.

    It is what the city has promised, what the company can ask for next, and what protection remains for the people and places sharing the same supply.

    Continue the investigation

    This article extends Immortal AI’s Data Centre Community Impact investigation. Read AI’s Buildings Are Private. The Costs Do Not Always Stay That Way and Meta Needed a Data Centre. BlackRock Found the Money. Who Carries the Risk? for the wider questions of infrastructure costs, finance and responsibility.

    For practical tools, evidence questions and the host-community decision framework, visit the Data Centre Community Hub.

    Sources

    Oregon Public Broadcasting: As Google’s water demands grow, The Dalles aims to pull more from Mount Hood forest

    Oregon Public Broadcasting: The Dalles’ mayor called OPB’s data center story inaccurate. Here are the facts

    City of The Dalles: The Dalles Watershed Act, what you need to know

    City of The Dalles: Google’s contributions to the city water system

    Reporters Committee for Freedom of the Press: Settlement over Google water-use records

    Virginia JLARC: Data Centers in Virginia

    City of Hillsboro: Data Center Water Use

    Office of the Texas Governor: Comprehensive data-centre audit direction

    Lawrence Berkeley National Laboratory: US Data Center Energy Usage Report

    Editorial note: The one-third figure refers to Google’s share of water consumed through The Dalles city system in 2024. It does not establish what share supported AI workloads. The city’s stated reasons for the watershed proposal and Google’s infrastructure contributions are included. Disputed claims are identified as disputed.

    Editorial disclosure: Immortal AI uses AI-assisted research and drafting. Material claims were checked against city records, government reports, court-related records and detailed local reporting. Final editorial decisions remain the responsibility of Immortal AI.

  • Data Centre Tax Breaks: How Many Jobs Are Actually Permanent?

    Data Centre Tax Breaks: How Many Jobs Are Actually Permanent?

    INVESTIGATION | POWER & INFRASTRUCTURE

    Official evaluations often combine temporary construction work, permanent operations roles, contractors and modelled spillovers when describing the employment created by data centres.

    By Andrew McDonald · Immortal AI

    Data centres create substantial construction employment, but far fewer permanent operating jobs. Virginia estimates illustrate the difference: approximately 1,500 workers during peak construction compared with around 50 full-time roles at a typical operating facility.

    The problem is not that data centres create no work. They do. The problem is that “jobs” can mean several different things: a temporary construction workforce, permanent employees inside the facility, contractors who service it, and jobs estimated elsewhere in the economy. Those categories describe different public benefits, but incentive debates often present them as if they were interchangeable.

    The number changes when the clock changes

    Virginia’s legislative oversight agency, JLARC, estimated that data centres supported about 74,000 jobs annually, including direct, indirect and induced employment. About 59,000 were associated with construction and 15,000 with operations. Of the operations total, about 4,400 were direct jobs.

    The difference is easier to see at the scale of one facility. JLARC reported that a typical 250,000-square-foot data centre may employ about 50 full-time workers once operating, roughly half of them contractors. At peak construction, the same project may put about 1,500 people on site for 12 to 18 months.

    Both figures matter. Construction jobs can be valuable, well paid and locally significant. But they are not the same promise as decades of permanent employment. A public claim that combines them should say so plainly.

    The tax benefit is easier to count

    Virginia’s retail sales and use tax exemption delivered about US$928 million in tax savings to the data-centre industry in fiscal year 2023, according to JLARC. About 90 percent of the industry used the exemption. The public cost was therefore concrete enough to estimate. The employment return required more interpretation: which phase, which employer, which geography and which economic model?

    Georgia shows the same tension from another direction. Its programme sets investment and “quality job” thresholds that vary with county population. Depending on location, a qualifying project may need to create as few as five, ten or 25 quality jobs while investing between US$25 million and US$250 million. That structure may be intentional: data centres are capital-intensive rather than labour-intensive. But it also means the scale of the investment can dwarf the minimum direct-employment requirement.

    Would the project have happened anyway?

    The hardest question is causation. An incentive can coincide with a project without being the reason the project exists. Georgia’s 2025 evaluation estimated that only 30 percent of data-centre construction activity was attributable to the exemption; in its model, 70 percent would have occurred without it. That estimate is not a universal fact about every project. It is an analytical assumption used to test the programme’s effects, and it should be read as such.

    The same evaluation also illustrates why simple verdicts are misleading. It calculated a negative direct fiscal impact for state government while estimating positive economy-wide value added. A concession can therefore look costly in the public ledger and beneficial in a broader economic model at the same time. The result depends on what is counted, over what period, and which benefits would not otherwise have occurred.

    A better public bargain

    Communities do not need a single magic jobs number. They need a public ledger that keeps unlike things separate. At minimum, incentive agreements and annual reports should distinguish construction jobs from ongoing jobs; employees from contractors; local hires from workers brought in temporarily; direct jobs from modelled indirect and induced effects; and commitments from outcomes.

    Reporting should also show duration, pay bands and the date on which each job count was measured. If a benefit depends on a minimum headcount, the public should be able to see whether the threshold was maintained. If the project misses the requirement, the agreement should explain whether tax benefits can be suspended, reduced or clawed back.

    Oregon’s enterprise-zone system offers a useful governance lesson: annual reporting and public-agreement requirements can make a promise more auditable. The exact design will vary by jurisdiction, but the principle is portable. The public should not have to infer performance from a ribbon-cutting announcement years after the tax treatment became certain.

    What residents can ask

    • How many jobs are construction, how many are permanent operations roles, and how long is each category expected to last?
    • How many roles are direct employees, contractors, local hires and modelled spillovers?
    • What wages, hours and benefits qualify a position to be counted?
    • Which figures are contractual commitments, which are forecasts, and which have been independently verified?
    • What happens if the investment or job thresholds are missed after the exemption has been claimed?

    The honest answer is not that every data-centre tax incentive is a bad deal. Some projects may broaden the tax base, support construction trades, improve infrastructure or generate wider economic activity. The honest answer is that those benefits should be tested against a clear counterfactual and reported in categories the public can understand.

    The tax break may be certain. The public return should be no less visible.

    Related Immortal AI investigations

    Principal sources

    AI disclosure: Immortal AI uses AI-assisted research and drafting. Sources, claims, framing and final editorial decisions remain the responsibility of Immortal AI.

  • AI’s Buildings Are Private. The Costs Do Not Always Stay That Way.

    AI’s Buildings Are Private. The Costs Do Not Always Stay That Way.

    NEWS & ANALYSIS | POWER & INFRASTRUCTURE

    Meta and BlackRock have created a $14 billion venture to build an AI data-centre campus in Texas. The deal shows how private capital is spreading the cost of the AI buildout. It does not answer who ultimately pays for power, water, tax concessions and stranded infrastructure.

    By Andrew McDonald · Immortal AI

    The artificial-intelligence boom is often described as a race between models. Its physical reality looks different.

    It requires land, concrete, servers, cooling systems, transmission lines, substations, water, roads and large amounts of finance. Before an AI system produces a single answer, someone must build the infrastructure that allows it to exist.

    On 28 July 2026, Meta and BlackRock announced a venture to develop and own a data-centre campus in El Paso, Texas. The project is expected to cost about $14 billion.

    BlackRock-managed funds will own 80 per cent. Meta will retain 20 per cent and contribute land and construction assets valued at about $2.3 billion. BlackRock will contribute about $4.9 billion in cash. Much of the structure will be financed through approximately $12.5 billion of debt.

    This is private finance at extraordinary scale. But “privately financed” does not necessarily mean every cost stays private.

    Immortal AI’s earlier analysis examined the Meta-BlackRock deal itself. This article follows the costs beyond the financing structure, into the grids, tax systems and communities that host the infrastructure.

    The first cheque is not the final burden

    The developer, operator, technology company, lender and investor may write the first cheques for a data centre.

    Other costs can travel through different systems. A utility may build new generation, transmission or distribution assets. A water authority may expand capacity. A council may provide roads, emergency services or planning staff. Governments may offer tax abatements or exemptions. Residents may experience noise, land-use conflict or pressure on scarce resources.

    Whether those costs reach households and taxpayers depends on contracts, tariffs, regulation and local conditions.

    It would be wrong to assume that every data centre raises household bills. It would be equally wrong to assume that a private project cannot shift risk onto the public.

    The real question is whether the people who benefit from the project are required to pay the costs it causes.

    The electricity scale is no longer marginal

    A Berkeley Lab report estimated that US data centres used 176 terawatt-hours of electricity in 2023, equal to 4.4 per cent of national electricity consumption.

    Its scenarios place 2028 demand between 325 and 580 terawatt-hours, or about 6.7 to 12 per cent of US electricity use.

    Those figures are estimates, not guarantees. They depend on accelerator shipments, utilisation, cooling, efficiency and the pace at which proposed facilities connect.

    The range itself is important. Utilities can be asked to plan expensive infrastructure years before final demand is known. If a project is delayed, reduced or cancelled after new assets are committed, someone still has to pay for what was built.

    A strong large-load tariff can place that risk on the applicant through upfront study payments, minimum bills, long contracts, credit support and exit charges. A weak arrangement can leave other customers carrying part of the stranded cost.

    Jobs are real. So is the difference between construction and operations

    Data centres create economic activity.

    Virginia’s Joint Legislative Audit and Review Commission estimated that the industry supported about 74,000 jobs and $9.1 billion in annual gross domestic product in the state during the period it studied. Most estimated employment came from construction rather than continuing operations.

    Several industry representatives told the commission that a typical 250,000-square-foot facility may employ around 50 full-time workers, about half contractors. That is an industry-reported benchmark, not a universal average.

    The distinction matters because public announcements often combine temporary construction work, indirect employment and permanent site jobs. All three can be valuable. They are not interchangeable.

    A community committing land, power, water or tax support needs to know which benefits remain after construction crews leave.

    Tax concessions are public spending through the tax code

    Governments often compete for data centres by exempting expensive servers, electrical equipment and other purchases from sales or use taxes.

    That can attract investment. It can also create a large public cost.

    Virginia’s legislative analysts reported that the state’s data-centre sales and use tax exemption accounted for $1.7 billion, or 42 per cent, of economic-development incentive spending over the decade examined in its 2024 report.

    That figure does not prove the exemption was bad policy. The correct comparison is what activity, tax revenue and strategic benefit occurred because of the concession, and what would have happened without it.

    But the cost must remain visible. A tax exemption is not free because no agency writes a cheque. It is revenue the public chooses not to collect.

    The same discipline should apply to property-tax abatements, discounted public land, infrastructure grants and special utility arrangements.

    Communities can receive large benefits

    The evidence is not one-sided.

    Where data-centre equipment and property are taxable, local revenue can be enormous. Mature clusters can fund schools, services and infrastructure at a scale few other land uses match.

    Construction can create well-paid work. Operations can provide skilled jobs and local contracts. Large customers can support new generation and spread fixed utility costs when tariffs are designed properly. Reclaimed water and less water-intensive cooling can reduce pressure on drinking supplies.

    These are genuine benefits. They are also location-specific.

    A result in a mature Virginia cluster cannot simply be applied to an arid Texas community, a constrained power market or a town offering a different tax structure.

    Who receives the upside?

    The Meta-BlackRock venture makes the ownership chain unusually visible.

    Meta will gain access to computing capacity. BlackRock-managed funds will own most of the venture. Lenders will receive interest. Contractors and suppliers will be paid to build it. The operator and technology tenant may create long-term revenue from the facility.

    The host community may receive wages, contracts, taxes and community investments.

    But gross project value does not show how much benefit remains locally after tax concessions, infrastructure costs and resource impacts are counted. Nor does a large capital figure tell us who bears the risk if technology changes, demand falls or the tenant leaves.

    Layered ownership can spread financial risk among sophisticated investors. It can also make public accountability harder because the landowner, operator, tenant, utility and financier may be different entities with different contracts.

    The community test

    A government should require a project-specific answer to five questions.

    1. Will the project pay the full cost of the electricity, transmission, water and public services it causes?

    2. Are permanent local jobs, taxes and procurement benefits stated separately from temporary construction and modelled indirect effects?

    3. Would the project proceed without the proposed tax concessions, and can those concessions be recovered if promises are not met?

    4. Are power, water, land and amenity impacts within local limits under realistic drought, heat and growth scenarios?

    5. Who pays if the project is delayed, reduced, cancelled or abandoned?

    If those questions cannot be answered before approval, the uncertainty does not disappear. It is transferred.

    AI may be global. Its infrastructure is local.

    The economic value created by AI may flow through products and markets around the world. The physical burden lands somewhere specific.

    A substation is built in one service area. Water is drawn from one system. Noise reaches particular homes. A tax exemption affects a particular budget. A cancelled load leaves assets on a particular grid.

    The Meta-BlackRock agreement shows that private markets are developing increasingly sophisticated ways to finance AI infrastructure. Communities need equally sophisticated ways to protect the public interest.

    The standard should be simple: project-caused costs should be assigned to the project, public support should produce an auditable public return, and local people should not be asked to insure private upside without knowing it.

    Principal sources

    Editorial disclosure: This article was developed with assistance from artificial intelligence. Its sources, claims and conclusions were reviewed by Immortal AI’s editor before publication.