Policy Deep-Dive
Energy, water & taxes

Oregon built the cloud.Was it worth the bill?

Roughly 125 data centers now run in Oregon, drawn by no sales tax, cheap hydropower, and $450 million-plus a year in tax breaks. In 2026 the bargain blew open: a moratorium, a lawsuit, a first-in-the-nation rate overhaul — and a governor saying Oregon should “stop being a cheap date.” This deep-dive argues both sides at full strength, then runs the test that actually matters: under what conditions is a data center a win for the town next door?

~125

data centers in Oregon

Hillsboro, the Columbia River, Central Oregon

$450M+

yearly tax breaks to tech companies

the country's most generous data-center terms

$19M

AWS pays Morrow County anyway

more than the next 16 taxpayers combined

+30% / −1.3%

the 2026 rate split

data centers pay more, households pay less

2026: the year it broke open

Why everyone is suddenly fighting about server farms

Data centers grew quietly in Oregon for twenty years. Then, in six months, the state froze their tax breaks, a teachers' union sued two governments, and regulators rewrote how they pay for power. Three fights, one question: who's getting the better end of this?

Why here at all? No sales tax on billions of dollars of servers, cheap Columbia River hydropower, a cool dry climate that's free air-conditioning, and eight trans-Pacific fiber cables landing at Hillsboro (Oregon Citizens' Utility Board). Communities have zoned 9,100 more acres for them — enough to quadruple the industry's footprint (Lincoln Chronicle).

June 5

The state hit pause

House Bill 4084 barred new data centers from enterprise-zone property tax breaks — Oregon's largest incentive program — until after the 2027 session. Existing and already-approved deals keep their terms (Davis Wright Tremaine). Gov. Kotek had cut data centers from her own tax-break expansion three months earlier (OPB).

June 24

The lawsuit landed

1000 Friends of Oregon, Tax Fairness Oregon, Tualatin Riverkeepers, and the Oregon Education Association sued Hillsboro and Washington County over ~17 tax-break applications — tied to NVIDIA, CoreWeave, Adobe, Dropbox, QTS, and Flexential — rushed through before the moratorium, allegedly without required public notice (Oregon Capital Chronicle).

All year

The power bill got rewritten

Under 2025's POWER Act — the first law of its kind in the country — regulators moved big data centers into their own rate class. PGE's version raised their bills ~30% and cut household bills 1.3% (OPB, Tom's Hardware). The governor, in July: Oregon should “stop being a cheap date” (KLCC).

The strongest case for

Rural Oregon's best deal in a generation

Set aside the metro fights for a moment and look at Boardman, Hermiston, and Prineville — towns the timber economy left behind. This is the case data-center supporters actually make, at its strongest.

This is a steelman: the best honest version of the argument, built from what county commissioners, chambers of commerce, and the companies themselves point to — not a caricature of it. The counter-case follows.

Even with the breaks, they're the biggest taxpayer by far

AWS pays Morrow County more than $19 million a year in property taxes — more than the next 16 largest taxpayers combined — while three of its five sites are still fully tax-exempt. Amazon's buildings are roughly half the county's taxable value and a third of everything it levies. On top of that, fee-in-lieu agreements send tens of millions to schools, housing, and emergency services (East Oregonian, Baker City Herald).

Real jobs, in places that had lost theirs

Local officials credit data centers with 7,400 full-time-equivalent jobs across Eastern Oregon and $9 billion in regional GDP (East Oregonian). The deals require wages at least 130% of the county average (Baker City Herald). Prineville — near 20% unemployment after the mills closed and Les Schwab left — now anchors its budget on Meta and Apple (Fortune).

This is an industry Oregon actually wins

Rural counties can't recruit chip fabs or biotech campuses. They can win data centers, because the raw ingredients — cheap hydro, cool dry air, fiber landings, open land — are already here (Oregon Citizens' Utility Board). Washington, Idaho, and Arizona offer their own incentives; the choice isn't between a data center with breaks and one without, but between hosting the buildout or watching it cross the river.

The honest counterfactual is a wheat field

The abatements forgive taxes on buildings that otherwise wouldn't exist. Without the deal, the land is dryland farm ground generating a few thousand dollars a year. With it, a county of 12,000 people collects $19M+ in taxes and ~$40M in 15-year fees (Baker City Herald). Since the 2026 POWER Act rules, big users also carry 100% of the grid costs they create — households' bills went down (Oregon Environmental Council).

The strongest case against

You paid for buildings that employ almost no one

The opposition's case doesn't depend on hating tech. It depends on arithmetic: what Oregon gives up per job, who absorbs the cost, and what the buildings consume that never shows up on a tax ledger.

Incentive cost per job created

Oregon's incentive study, by program type

Standard enterprise zone (3–5 yr)$4,200

Typical manufacturing-era deal

Long-term rural enterprise zone (15 yr)$54,500

The main data-center vehicle

The same study found the 15-year deals return 16 cents of personal income tax for every dollar of property tax forgone — an 84-cent loss — while standard zones return $1.35.

The jobs math doesn't survive contact

A typical facility runs on ~30 permanent staff across millions of square feet, many in $35–50K security and maintenance roles (Oregon Center for Public Policy). Oregon's own incentive study: the 15-year rural deals lose 84 cents per dollar forgone and cost ~$54,500 per job — thirteen times the standard program (Governing). One Hillsboro deal penciled out near $520,000 per employee per year (Oregon Center for Public Policy).

Every school in Oregon pays a little for each local deal

The mechanic is subtler than “the district loses money.” Oregon equalizes school funding, so when an abatement cuts local revenue the state backfills that district — and the loss reappears as a thinner statewide pool, approximately $1.50 per weighted student for every $1M removed (Oregon Dept. of Education). Hermiston's superintendent told the committee the enterprise-zone school fee is “not additional support to the local district” (DCAC (June 26, 2026)). Districts took $275M in abatements in 2024 alone (Oregon Capital Chronicle) — spread across everyone, felt sharply by no one.

They were coming anyway — the neighbors prove it

Oregon stacks four subsidies: no sales tax (Amazon saved roughly $3B on $39.3B of eastern Oregon investment), ~$450M/yr in property tax breaks, publicly funded infrastructure, and a 10-year income tax exemption that let Meta pay about 1.3% instead of 7.6%. The decisive fact Tax Fairness Oregon put before the committee: California and Washington offer no data-center property tax break at all — and still have data centers (Tax Fairness Oregon). Even the governor concluded Oregon was underpricing itself (KLCC).

The climate ledger is going backwards

Umatilla Electric — 16,000 customers — now emits 1.8 million tons of CO₂ a year serving Amazon's buildout, making it Oregon's third-largest utility emitter; its power is 2,000% more carbon-intensive than a decade ago (Governing). Regional reports warn data-center gas demand could push Oregon and Washington off their 2050 climate targets (Oregon Capital Chronicle).

The water story only came out in court

Google's The Dalles campuses drank 274.5 million gallons in 2021 — over a quarter of the entire city's water, near 40% in recent reports — and the public only knows because the city lost a 13-month fight to keep it secret (The Register, WaterWatch of Oregon). Statewide reporting requirements are still virtually nonexistent, even as the boom reaches basins already in a nitrate groundwater crisis (Rolling Stone).

When the door was closing, they went around the public

In the weeks before the moratorium took effect, companies filed a flurry of Hillsboro applications — six from one firm in a single April day — and staff approved them, the lawsuit alleges, without public notice or a vote of any elected body (Oregon Capital Chronicle). Whatever you think of the deals, that's not how public money is supposed to move.

Where the cases collide

Three findings that decide the argument

Both sides are quoting real numbers. Put them side by side and the disagreement gets smaller — and more useful. The evidence points to three conclusions neither campaign leads with.

Who pays after the POWER Act

Change in electric bills under PGE's 2026 data-center rate class

Data centers (>20 MW)+30%

New rate class: 10–30 year contracts, 100% of new grid costs, minimum-take rules

Residential customers-1.3%

First decrease after years of increases driven partly by grid buildout

01

Both jobs numbers are true — and that's the point

Amazon directly employs roughly 740 people in its eastern Oregon data centers; boosters' 7,400-FTE figure adds construction, contractors, and ripple effects (Governing, East Oregonian). A 2026 Georgia Tech study of the whole country splits the difference: metro counties gain ~4.1% employment after a data center opens; rural counties see “negligible” job spillover — but wages rise ~5% where clusters form, and electricity prices rise ~5% too (Fortune). Data centers are weak jobs programs everywhere. That was never their real value.

02

Geography flips the verdict

The same deal that's hard to defend in Hillsboro is easy to defend in Boardman. Morrow County traded taxes it could never have collected — the buildings wouldn't exist — for payments that now carry a third of its levy (East Oregonian). Washington County traded real school revenue to companies drawn by fiber cables that were coming ashore regardless (Oregon Center for Public Policy). The question was never “data centers, yes or no” — it's “which county, at what price.”

03

The POWER Act proved the win-win is buildable

For years the debate assumed someone had to lose: either block the industry or let households subsidize its grid. Then Oregon wrote the first law in the country making >20 MW users a separate rate class — and the industry stayed, its bills went up ~30%, and household bills went down (OPB, Tom's Hardware). Priced correctly, the arrangement survives. That's the template for every other term of the bargain.

Our honest read

The win-win test: six conditions, and where Oregon stands

A data center deal is a genuine win-win when the host community keeps more than it gives up — on taxes, power, water, and trust. That's testable. Here's the test, applied to Oregon as of August 2026.

Score today: 1 met, 1 partial, 4 unmet. The met one — grid costs — was considered impossible until 2025. None of the unmet four requires banning anything; they require pricing, disclosure, and process. Scoring updated after reading the committee's own record.

Met

01Large users pay the full cost of the grid built for them

The POWER Act (2025) and PGE's 2026 rate class make >20 MW users sign 10–30 year contracts, cover 100% of new distribution costs, and meet minimum-take rules. Data-center bills rose ~30%; residential bills fell ~1.3%.

Partial

02New load is tied to new clean energy, not backfilled with gas

PGE connects new data centers only if emissions-free power is available — but HB 2021 and the POWER Act reach only about 61% of Oregon electricity sales. Co-op and PUD territory is exempt, and that is exactly where Umatilla Electric's carbon intensity rose 2,000% in a decade serving Amazon's buildout.

Unmet

03Water use is disclosed and capped in stressed basins

State reporting requirements are virtually nonexistent. The Dalles spent 13 months in court trying to keep Google's water figures secret; Google used over a quarter of the city's water in 2021, near 40% in recent reports.

Unmet

04Tax deals return more than they give up

Oregon's own incentive study found 15-year rural enterprise-zone deals — the main data-center vehicle — lose 84 cents per dollar of property tax forgone and cost ~$54,500 per job, vs. $4,200 in standard zones.

Unmet

05Schools and services are held harmless

Hermiston's superintendent told the committee the enterprise-zone school fee is 'not additional support to the local district' — equalization backfills the district and the loss lands on the statewide pool, roughly $1.50 per weighted student per $1M. Negotiated fees do fund real local projects, including a $240M Boardman school bond over 80% covered by one operator.

Unmet

06Deals are approved in public, with notice and a vote

A June 2026 lawsuit alleges Hillsboro and Washington County staff approved ~17 enterprise-zone applications — filed in a rush before the moratorium — without required public notice or governing-body authorization.

So which way does it lean?

As signed before 2026, the bargain leaned toward the companies. That's not an activist's claim; it's the state's own math. The 15-year vehicle that built the boom loses 84 cents on the dollar, the buildings employ dozens where the renderings implied thousands, and the costs that never hit a ledger — carbon, water, school revenue — piled up in the dark.

But “bad deal” is not the honest verdict either. In Morrow County, the alternative to an under-priced deal was no deal: the fees and taxes are real, transformative, and bigger than anything else that region could plausibly have recruited. And the POWER Act showed the terms are negotiable — Oregon raised the industry's power bill 30% and the industry kept building.

The win-win exists, under conditions Oregon has only started to enforce. Full-cost power pricing (done), clean-energy additionality everywhere including co-op territory (half done), water disclosure with caps in stressed basins (not done), abatements repriced toward fee-in-lieu deals that hold schools harmless (not done), and approvals in public (not done). If the 2027 Legislature converts the current pause into those terms, Oregon keeps the industry and the returns. If the moratorium simply lapses, the pre-2026 pattern — private gain, socialized cost, discovered later — comes back with the AI boom behind it.

The framework

How to price a data center deal

The whole fight reduces to three numbers: what the site would pay fully taxed, what the deal actually pays, and how likely the company was to build without a break. Get those three, and “good deal or giveaway” stops being a matter of opinion.

The model prices the deal from the host community's side. It deliberately excludes construction activity (temporary, and present in every scenario where the facility is built) and assumes bare land yields ~no revenue — both stated in the fine print, both adjustable in spirit by moving the sliders.

01

What the site pays fully taxed

Taxable value ≈ 70% of on-site investment — servers are most of the value and depreciate fast, but they're refreshed continuously — times the county's rate, over 15 years, discounted. A $2B campus at 1.1%: about $15M a year, or $171M in present value.

02

What the deal actually pays

Fees in lieu of taxes during the abatement, full taxes after it ends, plus state income tax from the permanent jobs. A Morrow-style deal — ~$2.7M/yr in fees, 100 jobs — pays about $37M in present value: roughly a fifth of full freight.

03

The break-even leverage

Divide №2 by №1 (plus income taxes on both sides): the deal beats holding firm only if the chance they'd build anyway is below that ratio. The Morrow-style deal breaks even at ~21%; a Hillsboro-style deal at ~64% — and the fiber cables put Hillsboro's true leverage well above that.

Price the deal

Start from a region, then move the levers. Every input is a real deal term — the presets track the public record for each region.

On-site investment (bldg + servers)$2.0B
$200M$5B
Effective property tax rate1.10%
0.7%1.4%
Abatement length15 yrs
none15 yrs
Fee in lieu of taxes$2.7M/yr
$0$20M/yr
Permanent jobs100
10500
Average wage$90K
$50K$150K
Leverage: would they build with NO break?25%
nevercertainly

Cheap land and power exist across the river in Washington too — leverage is genuinely low.

The deal pencils out only if leverage is below

21%

At your estimate (25%), signing loses to holding firm by $7.9M in expected 15-year value.

deal-friendly zone ends at 21%▎= your leverage
If fully taxed, 15-yr PV$171M
What the deal pays, 15-yr PV$30M

$94K

taxes forgone per job, per year (state study range: $4.2K–$54.5K)

$6.5M

15-yr PV of income taxes from the jobs (goes to the state, not the county)

Teaching model, not a forecast: 15-yr horizon, 4% real discount, taxable value ≈ 70% of investment (server refresh vs. depreciation), 6.5% effective income tax, baseline land revenue ≈ $0. Construction activity, utility fees, and service costs are excluded on both sides.

This is the policy insight hiding in the arithmetic: the right abatement is not a number, it's a curve. The scarcer a community's advantages, the more it can afford to give; the more unique its advantages, the less it should. A statewide flat program — which is what the enterprise-zone system was — prices Boardman and Hillsboro identically, and therefore misprices both. Benchmarks for the cost-per-job readout come from the state incentive study (Governing).

Oregon already ran this calculation — and published the answer

In June 2026, Business Oregon handed the advisory committee its own return figures by program. The short-term standard enterprise zone returns 29.16 per dollar. The Strategic Investment Program returns 6.24. The 15-year rural enterprise zone — which carries $15.4B of the $15.8B in data-center investment and $233.6M of the $240.9M in abated taxes — returns 1.18 (Business Oregon). The state's economic development agency is not disputing the critics' arithmetic. It supplied it.

Morrow County's assessor put the same point in county terms: $123.6M of property tax exempted against $72.3M actually collected countywide, with $23.9M coming back as negotiated fees (Morrow County Assessor). Both things are true at once — the fees are transformative for a county of 12,000, and the exemption is larger than the entire tax base.

What a fully priced framework looks like — Minnesota, presented to Oregon in May

Oregon's POWER Act allocates cost. Minnesota's 2025 bipartisan package, presented to this committee by its Department of Commerce, also directs what gets built: utilities must show a new data center won't raise other customers' rates or impede the state's 100%-clean-by-2040 standard; 15-year agreements run at 80% take-or-pay with upfront collateral and exit fees; a large-user fee funds low-income weatherization; rules reach down to 5 MW so projects can't shrink under the threshold; and the electricity sales-tax exemption was repealed. Google — testifying as the customer — called Oregon's law a missed opportunity by comparison (DCAC (June 26, 2026)).

The transferable lesson for pricing: Minnesota converts every soft condition into a contract term with a number attached. That is what makes a deal checkable rather than promised — and it is why the same approach works for water disclosure and abatement length, not just electricity.

The same math, on a map

Run the framework across Oregon's regions and a siting policy falls out: where deals can pencil, where they can't, and where the answer isn't price at all. The state has started acting on the same logic — in July 2026 the Governor pulled 32 state-owned acres in Salem from a proposed data center (Governor's Office).

HillsboroThe DallesBoardmanPrinevilleSalemBurns
Good at the right priceConditionalPull backNot viable

Schematic, not a survey — each verdict comes from the region's leverage, water, and grid position. Same variables as the calculator.

Hillsboro / Washington County

Pull back

Full freight. With leverage this high, almost no abatement pencils out — the state's study and the 2026 lawsuit both point the same way.

Leverage ~85%Water Tualatin basin pressure; Riverkeepers now in courtGrid PGE territory — POWER Act guardrails apply

The Dalles / Columbia Gorge

Conditional

Conditional. Deals can pencil at mid leverage — but only with mandatory water disclosure and caps written into the agreement.

Leverage ~50%Water Google used >25% of city water in 2021, ~40% recentlyGrid Adjacent hydro; municipal water is the limit

Columbia River East

Conditional

The strongest case for deals — and the weakest guardrails. Pencils at low leverage, but fees near Morrow's ~$2.7M/yr run close to the break-even line, and clean-power terms don't exist here yet.

Leverage ~25%Water Groundwater already in a nitrate crisisGrid Co-op territory: carbon intensity up 2,000% in a decade; POWER Act doesn't apply

Central Oregon

Conditional

Conditional, tightening. The cluster is established; abatement terms should ratchet down with each expansion, not roll over.

Leverage ~40%Water High desert: cooling water is scarce by definitionGrid Pacific Power territory — rate case pending

Willamette Valley

Pull back

The state has already answered: in July 2026 the Governor withdrew 32 state-owned acres in Salem from a proposed data center. High-value farmland is the wrong trade at almost any price.

Leverage ~60%Water Farm and municipal demand already competeGrid Constrained; no surplus story

South & Southeast Oregon

Not viable

Not a pricing question. Without fiber and grid, no subsidy attracts a serious project — chasing one means giving away the most for the least.

Leverage ~5%Water Closed basins; some already over-allocatedGrid No major fiber routes or transmission headroom
Who writes the rulebook

The seven people drafting Oregon's answer

Everything above — pricing, siting, water, power — lands on one desk: the Data Center Advisory Committee Gov. Kotek convened in January 2026. Its recommendations, due by October 2026, are the blueprint the 2027 Legislature will work from.

Charge, membership, and schedule from the Governor's announcement (Governor's Office) and the Oregon Dept. of Energy's committee page (Oregon Dept. of Energy), where agendas, recordings, and materials are posted.

The seven members

  • Margaret HoffmannCo-chairOregon member, Northwest Power & Conservation Council
  • Michael JungCo-chairEnergy & climate policy professional
  • Dan DorranChair, Umatilla County Commission
  • Greg DotsonAssociate professor of law, University of Oregon
  • Bill EdmondsAdjunct professor, University of Portland
  • Tim MillerDirector, Oregon Business for Climate
  • Jean WilsonOperating partner, Sandbrook Capital

A deliberately mixed table: grid planning, environmental law, a county in data-center country, climate-minded business, and private capital — the coalition any deal framework has to survive.

What it must deliver

  1. 01Encourage responsible siting that supports economic development, especially in rural communities
  2. 02Understand data centers' effects on Oregon's climate, clean-energy, and natural-resource goals
  3. 03Ensure data centers get reliable energy without burdening other ratepayers
  4. 04Protect Oregon's limited water resources as cooling demand grows
  5. 05Identify the policy framework the state needs to guide growth responsibly

Deliverable: recommendations for a comprehensive regulatory framework, due to the Governor no later than October 2026 — the direct input to whatever replaces the moratorium in the 2027 session.

One topic per month, in public

  • Feb 27, 2026Economic development & workforce
  • Mar 27, 2026Water resources
  • Apr 24, 2026Land use
  • May 29, 2026Energy
  • Jun 26, 2026Energy affordability, revenue & incentives
  • Jul 31 & Aug 4, 2026Deliberations & draft report
  • Sep 23, 2026Public listening session on preliminary findings
  • Oct 2026Final report & recommendations to the Governor

You can weigh in — here's how

  • Meetings run on Zoom with recordings posted to YouTube; agendas, slides, and summaries are published for every session.
  • Each meeting reserves ~30 minutes for public comment — 15 first-come slots. Email datacenter.ac@oregon.gov with your name, the meeting date, and a two-minute summary.
  • A dedicated public listening session on the preliminary findings is scheduled for September 23, 2026 — the last realistic moment to move the report.
Inside the record

Thirteen things the committee's own documents establish

We downloaded all 77 documents and recordings the committee has posted — agency decks, industry slides, tribal testimony, and the facilitators' summaries of what was actually said in the room. The most useful findings aren't the contested ones. They're the numbers the state produced about itself.

Everything below is attributed to a specific presentation, all of them linked at the bottom of this section. Where presenters disagree, both figures are kept — that disagreement is usually the most informative part.

01

The 15-year program Oregon used for nearly all data-center value barely breaks even

Business Oregon's own table: the long-term rural enterprise zone carries $15.4B of $15.8B in investment and $233.6M of $240.9M in abated taxes — and returns 1.18. The short-term standard program returns 29.16. The state is not disputing the critics' arithmetic; it published it.

Business Oregon, June 26 session

02

Morrow County exempts more property tax than it collects

$7.25B of exempt data-center value against $4.20B of total certified value; $123.6M in exempted tax against $72.3M actually collected countywide, with $23.9M returning as negotiated fees. Umatilla exempts $119.7M and receives $14.2M.

Morrow County Assessor, June 26 session

03

The school 'support fee' does not add money to the school

Oregon equalizes school funding, so when an abatement cuts a district's local revenue the state backfills that district. The loss shows up instead as a thinner statewide pool — about $1.50 per weighted student per $1M — meaning every district in Oregon pays a little for each local deal.

Oregon Dept. of Education & Hermiston School District

04

Permanent jobs are small; construction jobs are not

ECONorthwest counts 2,630 direct operating jobs statewide, 0.2% of Oregon employment. A future 30 MW facility supports 730 construction jobs and 46 permanent ones. Business Oregon's larger 7,600-job figure counts construction and payroll effects — both numbers are real, they measure different things.

ECONorthwest & Business Oregon

05

Electricity demand is the number that actually changes the state

Data-center consumption rises from 14.0 TWh in 2025 to 24.8 TWh by 2030 on ECONorthwest's projection — pushing toward 40% of Oregon's electricity demand. Average IT power per facility more than tripled, from ~11 MW pre-2010 to 39.5 MW for recent vintage.

ECONorthwest, July 31 session

06

There are 2,482 diesel generators behind the cloud

6,328 MW of permitted backup capacity across 39 campuses — more than the state's largest power plants combined — up from 135 MW in 2012. Emission controls are not required, and DEQ's voluntary incentive to install them has had no takers.

Oregon DEQ, July 31 session

07

Growth is moving west, where the case for subsidy is weakest

The installed base is eastern Oregon, but the next five years of physical growth run 80% west and 12% east. Hillsboro's advantages — eight trans-Pacific cables — are exactly the ones no abatement created.

ECONorthwest, July 31 session

08

Data centers took two-thirds of Oregon's new industrial land

9,093 acres of identified demand, 6,472 of it needing UGB expansion or rural upzoning. In 2025–26, urban-growth-boundary expansions for data centers hit 3,243 acres — roughly two-thirds of all new industrial land brought inside UGBs statewide.

DLCD, April 24 session

09

Connection requests already exceed the entire Northwest grid

Bonneville told the committee its queues hold roughly 60 GW of generation requests, 60–65 GW of transmission service requests, and about 40 GW of line-and-load — of which some 30 GW is data centers. Regional peak demand is about 30 GW. The asking is larger than the system that exists.

Bonneville Power Administration, May 29 session

10

Oregon's ratepayer protections cover about 61% of the state

HB 2021's clean-energy mandate and the POWER Act's cost rules bind investor-owned utilities — roughly 61% of Oregon electricity sales. Much of the data-center buildout sits in co-op and PUD territory, where neither applies. Committee members flagged the risk that weak-protection territory simply attracts the next wave.

ODOE & facilitator summary, May 29 session

11

Google told Oregon its own law was the weaker model

Presenting on energy, Google called the POWER Act a missed opportunity next to Minnesota's 2025 package — which pairs cost protection with affirmative clean-energy obligations, covers facilities from 5 MW up, and funds weatherization from a large-user fee. Oregon's law allocates cost; Minnesota's also directs what gets built.

Google & Minnesota Dept. of Commerce, May 29 session

12

Everyone agreed on disclosure — including the cities burned by secrecy

DEQ called withheld information its 'biggest challenge' in permitting. The Dalles conceded that redacting Google's water figures 'created skepticism and hurt public trust.' Tribes, advocates, academics, and agencies all converged on mandatory reporting as the floor.

March 27 water session, facilitator summary

13

The committee has already conceded the central point

The facilitator's June summary records that 'the DCAC agreed incentives need to be reviewed and modernized,' with several members flagging that school support fees mainly change who pays. The open question is no longer whether to reprice, but how far.

Facilitator summary, June 26 session

Who showed up

Every camp that testified, and its hardest number

Seven months of hearings drew agencies, tribes, utilities, unions, county assessors, the companies themselves, and the advocates suing them. Laid out by camp, the coalition structure explains the politics better than any single argument does.

Notice who is missing from the committee itself: no tribal member, no utility, no operator, no ratepayer advocate. Those groups got a presenter's slot, not a seat at the table where the report gets written.

State agencies

Mike Gorman

Morrow County Assessor

Reports the numbers without taking a side — and they are the sharpest in the record.

Morrow County exempts $7.25B of data-center value against $4.20B of total certified value: $123.6M in exempted tax versus $72.3M actually collected countywide, with $23.9M coming back as fees. Umatilla exempts $119.7M and collects $14.2M in payments.

Alex Albertine & Michael Held

Business Oregon

Incentives leverage investment the state would not otherwise capture.

The agency's own return figures split hard by program: short-term standard enterprise zones return 29.16, while the 15-year rural program that carries nearly all data-center value — $15.4B of $15.8B invested, $233.6M of $240.9M abated — returns 1.18.

Nolan Moser & Bret Stevens

Oregon Public Utility Commission

Growth pays for growth — the POWER Act is now operational.

PGE's May 2026 order sets a 90% minimum demand charge, 10–30 year contracts, exit fees, and a 1¢/kWh surcharge on facilities over 100 MW funding efficiency for energy-burdened households. Rates: +29% data centers, −1.3% residential, −3.7% small business.

Leigh McIlvaine

Dept. of Land Conservation & Development

The land-use system is strong but straining; should siting be UGB-only?

9,093 acres of identified demand, 6,472 of it requiring urban-growth-boundary expansion or rural upzoning. In 2025–26, data centers accounted for roughly two-thirds of all new industrial land brought inside UGBs.

Pete Wyckoff

Minnesota Dept. of Commerce

Here is a package that does what Oregon's doesn't.

Minnesota's 2025 bipartisan law requires utilities to show a data center won't raise other customers' rates or impede the 100%-by-2040 clean standard; sets 15-year agreements at 80% take-or-pay with collateral and exit fees; funds low-income weatherization from a large-user fee; and removed the electricity sales-tax exemption.

Vanessa Clark

Oregon Dept. of Education

Explains who actually loses school money — and it isn't the host district.

Equalization backfills the abating district; the loss lands on the statewide pool instead, diluting every district's per-student rate. Roughly $1.50 per weighted student for every $1M added to or removed from the State School Fund.

Ania Loyd

Oregon DEQ

Reports the diesel fleet nobody was counting.

2,482 large diesel backup generators totaling 6,328 MW across 39 permitted campuses, up from 135 MW permitted in 2012. Emission controls aren't required, and a 2025 streamlined-permit incentive for installing them has had zero uptake.

Academic / expert

Ellen Harpel

Smart Incentives (national expert)

Reprice and condition; don't abolish. And speed can beat money.

38 states offer data-center incentives, but only 11 have statewide property tax breaks. The 2026 trend is pause-and-adjust — moratoria in Ohio, Arizona, Illinois; conditions in Pennsylvania and Utah. Fast-track permitting often matters more to operators than subsidy dollars.

Elaine Hart

Sylvan Energy Analytics

Whether the region stays reliable depends on whether data centers can be curtailed.

E3 projects a ~9 GW regional capacity shortfall by 2030. Data-center flexibility cuts new 2030 capacity needs from 1–5 GW to 0–3 GW — but baseline growth implies 7 to 9 days of large-load curtailment in a January-2024-type winter.

Terry Wirkkala & João Ferreira

ECONorthwest / UVA Weldon Cooper

Commissioned, explicitly neutral: here is the size of the thing.

2,630 direct operating jobs statewide — 0.2% of Oregon employment. Electricity demand rising from 14.0 TWh in 2025 to 24.8 TWh by 2030. A future 30 MW facility supports 730 construction jobs but only 46 permanent ones.

Local government

Supt. Tricia Mooney

Hermiston School District

The school fee doesn't actually reach the school.

The enterprise-zone School Support Fee is offset by the state equalization formula — it is 'not additional support to the local district.' Districts often choose the 15% minimum to leave room for negotiated community benefits instead.

Torrie Philippi-Griggs

Boardman Chamber of Commerce

Locally negotiated deals buy things a small town could never fund.

Fees flow through a local nonprofit board: more than $2.5M in homebuyer incentives over 15 years, a $6M business incubator, and a $240M school bond with a single data-center partner covering over 80%.

Steve Forrester & Mayor Jason Beebe

City of Prineville

Data centers rebuilt a town the timber economy left for dead.

Unemployment fell from 20% in 2009 to about 6%; county wages went from near-worst to second-highest in Oregon. Franchise fees run $10–12M a year against roughly $30,000 a month before the data centers arrived.

Mark Morgan

City of Hermiston

Per acre, a data center is the lightest water user in town.

Annual use per acre: schools ~1.7M gallons, food processing ~1.3M, homes ~850K, irrigated agriculture ~815K, hyperscale data center ~180K. With aquifer storage, he calculates a 191M gallon/year net gain for the Lower Umatilla basin.

Niki Iverson & Dan Dias

City of Hillsboro

Not a water problem here, and we planned for decades.

Data centers used 112 million gallons in 2025 — 1.8% of total city consumption and 3% of industrial use. 'Data centers are not considered large water users in the City of Hillsboro.'

Utilities

Robert Echenrode

Umatilla Electric Cooperative

Consumer-owned utilities already solved this locally; don't preempt us.

Data-center load pushed co-op sales from about 1 million to 8.4 million MWh by 2024 while residential rates stayed roughly 30% below the national average. Utility plant grew from ~$140M in 2010 to ~$775M in 2025.

Hamody Hindi

Bonneville Power Administration

The queue has blown past anything our planning model can handle.

Roughly 60 GW of generation requests, 60–65 GW of transmission service requests, and about 40 GW of line-and-load — some 30 GW of it data centers — against a Northwest peak of about 30 GW.

Industry

Beau Schilz

Amazon Web Services

Our water use is small, efficient, and more than offset.

284 million gallons statewide in 2024 — 3–11% of host utilities' production — at a water-use effectiveness of 0.16 L/kWh, about 70% below the North American average, alongside roughly $400M invested in community water infrastructure.

Jeff Omelchuck

Infrastructure Masons (retired)

Court traditional hyperscalers, not gigawatt AI campuses — and plan for the bust.

Oregon has ~123 data centers and zero AI-scale campuses. A Stargate-class project runs 6,000 construction jobs against 300 operating jobs. Warns to require decommissioning plans and financial surety: 'we don't want a hi-tech ghost town.'

Ellen Zuckerman

Google

We'll pay full incremental cost — and Oregon's law asks for less than Minnesota's.

Points to a fleet PUE of 1.06, a 24/7 carbon-free commitment, and an Xcel Minnesota deal pairing 1,400 MW of wind and 200 MW of solar with a 300 MW/30 GWh iron-air battery — estimated at $1.1B of net benefit to other customers over 15 years. Calls Oregon's POWER Act a missed opportunity.

Advocates

Jody Wiser

Tax Fairness Oregon

End the property tax breaks outright in 2027.

Oregon stacks four subsidies: no sales tax (Amazon saved ~$3B on $39.3B of eastern Oregon investment), ~$450M/yr in property tax breaks, publicly funded infrastructure, and a 10-year income tax exemption — Meta paid about 1.3% instead of 7.6%. California and Washington offer no data-center property tax break at all.

Anahi Segovia Rodriguez

Verde

Put residential affordability at the center; communities bear the burden.

PGE and Pacific Power rates rose more than 50% between 2019 and 2025. Disconnections hit 71,190 in 2025, with 9,839 in the first quarter of 2026 alone.

Kelly Campbell

Columbia Riverkeeper

The only voice asking for a pause.

A 100 MW data center consumes roughly 100 million gallons a year — about what 2,500 people use at home. Only half of data centers track water use at all. Asks for mandatory reporting plus a temporary moratorium until rules catch up.

Sam Diaz

1000 Friends of Oregon

Prepare the land you already have; tie abatements to outcomes. No moratorium.

A 66-jurisdiction survey found 9,746 acres already zoned industrial but only 21% development-ready, and 57.5% of jurisdictions reported missed opportunities for lack of ready land.

Ben Gordon

Central Oregon LandWatch

Slow down and review cumulatively — not a ban.

Asks for NEPA-style cumulative fiscal, infrastructure, and environmental review, statutory definitions for hyperscale and exascale, and stronger UGB-expansion standards before Oregon expands data-center land supply again.

Tribal

Julie Carter & Elijah Cetas

Columbia River Inter-Tribal Fish Commission

Treaty rights attach to places; assess cumulatively, not project by project.

Salmon runs fell from 15–20 million fish a year to under 500,000 by the late 1970s. Asks for legally protected instream flows, groundwater and surface water managed as one system, and a true-cost analysis that counts treaty-resource damage.

Trustee Lisa Ganuelas

Confederated Tribes of the Umatilla Indian Reservation

Presenting to this committee is not consultation.

Draws a formal line: CTUIR requires government-to-government consultation with the Governor's office when recommendations are finalized, not a presenter's slot. Flags siting opacity, waste heat, and small modular reactor co-location as unresolved.

Labor

IBEW, LiUNA, OPEIU & the Building Trades

Oregon construction unions

Explicitly opposed to any moratorium.

Twenty-plus crafts work a single campus; one Central Oregon IBEW local grew from roughly 40 to about 500 electricians. Apprenticeships are debt-free and five years long.

The incentives behind the incentives

Why the debate is shaped the way it is

Read the record and the alignments stop looking like opinions. They look like positions people hold because of where they sit. Eight structural features of this process explain most of what gets said — and what never comes up.

01

Tax policy was not in the original charge

The Governor added it after the charge was drafted. That is why the money question landed in month five of a seven-month process, compressed into a single June session, while water and land use each got a full day earlier in the calendar.

02

The people most affected aren't on the committee

The seven members include no tribal representative, no utility, no data-center operator, and no ratepayer advocate. Those voices appear as presenters — heard for a scheduled slot, then absent from the deliberation that follows.

03

Only new data centers are in scope

The charge covers siting of new facilities. The roughly $450M a year already committed to existing deals is structurally outside what the committee can recommend changing — the largest number in the debate is the one it cannot touch.

04

Benefits concentrate; costs disperse

Morrow County collects about a third of its property tax from data centers, so its commissioner sits on the committee and defends the deals. The offsetting school-funding loss is spread thinly across 197 districts, where no single actor feels enough pain to organize. That asymmetry explains almost every alignment in the record.

05

Counties negotiate alone against trillion-dollar counterparties

Assessors approve exemptions; the state does not. A county of 12,000 people negotiates with Amazon's real-estate team. One presenter asked the state to fund negotiating help for small towns — an admission of the mismatch that current law builds in.

06

Secrecy is structural, not incidental

NDAs and shell LLCs are standard, and utilities are often bound to confidentiality. DEQ says withheld information is its biggest permitting obstacle; The Dalles fought a newspaper for 13 months over water figures. Regulators are negotiating against parties who know far more than they do.

07

Power had a forcing mechanism; taxes don't

The POWER Act worked because one regulator with jurisdiction could rewrite one tariff and bind every large user at once. Tax policy has no equivalent lever — it is 36 counties making separate deals, which is why reform there requires the Legislature rather than an agency.

08

The labor coalition rules out the loudest option

Building trades unions testified against any moratorium, and neither 1000 Friends nor LandWatch asked for one. With construction jobs real and concentrated, a pause has almost no constituency — which pushes the debate toward pricing and conditions instead of prohibition.

Put together, these push toward one destination. A moratorium has no constituency — labor opposed it and the major land-use groups never asked for it. Abolition has no path — the money is already committed and out of scope. What's left is repricing and conditioning, which is exactly what the facilitator recorded in June: the committee agreed incentives need to be reviewed and modernized (DCAC (June 26, 2026)). The fight in 2027 will not be over whether to change the deal. It will be over how much, and who gets to decide — the state, or 36 counties negotiating one at a time.

The primary record

Every document, linked

The committee posts everything: agendas, presenter slides, facilitator summaries, and full meeting recordings. This is the complete index as of August 2026 — the raw material for anyone who wants to check our reading against the source.

Hosted by the Oregon Department of Energy (Oregon Dept. of Energy). Written comment goes to datacenter.ac@oregon.gov.

Apr 17, 2026Public listening session

Recording

Recording only — no slides posted.

Aug 4, 2026Deliberations, continued

The decision points

What happens next

The moratorium didn't settle the fight — it scheduled it. Beyond the committee's report, three places to watch.

2027 session

The moratorium forces a decision

HB 4084's enterprise-zone ban for new data centers expires 90 days after the session adjourns. Lawmakers must renew, replace, or let the old rules return.

In the courts

The Hillsboro / Washington County lawsuit

1000 Friends of Oregon, Tax Fairness Oregon, Tualatin Riverkeepers, and the Oregon Education Association want the pre-moratorium approvals voided.

At the PUC

Pacific Power's data-center rate case

PGE's rate class was first. Pacific Power's proceeding decides whether the same guardrails cover the rest of the state's investor-owned territory.

And if you live near a proposed site: the enterprise-zone sponsor — your city or county — is where the next deal actually gets decided. The committee section above has the dates and the comment address.

Sources & method

Where these numbers come from

Every figure links to a public document, a named study, or on-the-record reporting, pulled in August 2026. Both cases were built from what each side's strongest advocates actually cite — and the committee sections come from reading all 77 documents and recordings it has published, indexed in full above.

A live story

The advisory committee reports in October 2026, the Hillsboro lawsuit is in early stages, and Pacific Power's rate case is pending. Time-sensitive figures should be re-checked after each.

Oregon governor says state is a 'cheap date' for data centersKLCC · newsNew Oregon law bars new data centers from enterprise-zone tax breaks (HB 4084)Davis Wright Tremaine · analysisData centers are cut — for now — from a bill expanding Oregon tax breaksOPB · news“We've been very foolish”: inside Oregon's data center boomOregon Center for Public Policy · analysisOregon's own incentive study: what data center tax breaks returnGoverning · analysisPoliticians, teachers union sue Hillsboro, Washington County over data center tax breaksOregon Capital Chronicle · newsMorrow County approves $1 billion in tax breaks for Amazon data centersBaker City Herald · newsEastern Oregon leaders praise data centers, look to futureEast Oregonian · newsOregon Legislature passes 'POWER Act,' targeting industrial energy users like data centersOPB · newsNew PUC rules protect Oregonians from data center-caused rate increasesOregon Environmental Council · primaryPGE data center bills rise 30%, residential rates fall 1.3% under POWER ActTom's Hardware · newsOregon's rural power utility has become a big polluterGoverning · analysisData centers are driving demand for gas from Northwest utilities, reports findOregon Capital Chronicle · newsGoogle's The Dalles water use revealed after city drops records suitThe Register · newsData centers are hogging The Dalles' waterWaterWatch of Oregon · analysisHow Oregon's data center boom is supercharging a water crisisRolling Stone · newsData centers boost jobs 4% in cities; rural economies barely feel a dent (Georgia Tech study)Fortune · analysisOregon Data Center Advisory CommitteeOregon Dept. of Energy · primaryOregon communities envision 9,100 acres for new data centersLincoln Chronicle · newsWhy is Oregon a hot spot for data centers?Oregon Citizens' Utility Board · analysisData Center Advisory Committee charge from the GovernorGovernor's Office / ODOE · primarySession 5 facilitator summary — affordability, revenue & incentivesDCAC (June 26, 2026) · primaryData center incentive programs: investment, abatement & returnBusiness Oregon · primaryCounty assessor data on exempt value and in-lieu paymentsMorrow County Assessor · primaryThe four stacked subsidies, and what to end in 2027Tax Fairness Oregon · primaryUnderstanding Oregon's data center industry (preliminary findings)ECONorthwest · analysisHow property tax abatements move through school fundingOregon Dept. of Education · primaryAir-quality permitting: data center backup generatorsOregon DEQ · primaryData centers and Oregon's land-use systemDept. of Land Conservation & Development · primarySession 2 facilitator summary — water resourcesDCAC (March 27, 2026) · primaryAWS water use and community water investment in OregonAmazon Web Services · primaryWhat other states do with data center incentivesSmart Incentives · analysisGovernor Kotek convenes statewide Data Center Advisory CommitteeGovernor's Office · primaryGovernor Kotek directs state to withdraw land for proposed Salem data center siteGovernor's Office · primary