Salem's debate over data centers is moving beyond whether the city should welcome large computing facilities. A proposed City Council resolution raises a more immediate legal question: Does Salem's existing zoning code already prohibit AI and hyperscale data centers, or would declaring them prohibited amount to changing the law?
The distinction could affect a major development proposal at the Mill Creek Corporate Center, the way Salem regulates future projects and the legal defensibility of its land-use decisions.
The City Council is scheduled to consider Resolution No. 2026-46 at a public hearing Nov. 9. The measure would interpret the city's existing classification for data center facilities as excluding AI and hyperscale operations, including facilities principally devoted to large-scale computational processing.
Under the proposed interpretation, those facilities would not fall within any other use classification in Salem's Unified Development Code.
Developer Aaron Stickney has submitted a series of written objections arguing that the resolution would effectively establish a citywide prohibition without the procedures required to amend land-use regulations. He also contends that the interpretation cannot lawfully be applied to an application submitted months before the resolution was proposed.
Salem City Attorney Dan Atchison disputes the underlying premise. In written correspondence with Stickney on Oct. 28, Atchison characterized the resolution as a legislative interpretation of existing regulations, not an amendment, zone change or decision on a specific development application.
Neither position establishes the resolution's legality by itself. The answer depends substantially on what the existing ordinance means, how Oregon courts review local interpretations and whether the council's proposed action clarifies or changes the legal treatment of large data centers.
What Salem's existing code says
The disagreement begins with Salem Revised Code 400.110(e), which establishes a use classification called "data center facilities."
The provision describes facilities that physically house computer systems and associated equipment dedicated to data storage or transmission. It also identifies features commonly associated with data centers, including backup power supplies, communications connections, environmental controls and security equipment.
Its examples include computer centers, server farms, co-location facilities and data storage and hosting operations.
The text does not expressly exclude AI data centers or hyperscale facilities. Nor does the existing classification establish a numerical limit based on electricity consumption, building size or computing capacity.
That absence is central to Stickney's objection. He argues that the city's proposed distinction would remove a category of facilities that falls within the ordinary meaning of the existing classification.
But an absence of express exclusions does not necessarily mean every facility using computers qualifies.
SRC 400.015 directs the city to classify uses according to their principal activities and the characteristics of the relevant use and broader category. SRC 400.010 explains that listed characteristics describe necessary qualities and generally associated impacts, while examples illustrate typical activities without providing an exhaustive list.
The data center classification also contains an exception addressing facilities serving an individual on-site use, treating them as accessory to that use.
Taken together, these provisions require more than identifying whether a building contains servers. They also invite examination of what happens on the property and how that activity fits within the code's structure.
This gives the city a potentially relevant basis for distinguishing a facility principally devoted to storage and transmission from one principally devoted to high-intensity computation.
The question is whether that distinction can reasonably be drawn from the ordinance Salem already adopted.
Can processing be separated from storage?
According to the proposed findings discussed in Stickney's testimony, Salem's interpretation places considerable emphasis on the omission of "processing" from the characteristics listed in SRC 400.110(e).
That distinction is not merely semantic.
Traditional data hosting, cloud computing and modern AI training can involve different computing demands and infrastructure configurations. Facilities designed for large-scale computation may require substantial electrical capacity, specialized cooling and supporting equipment.
Those differences can matter when evaluating infrastructure capacity, neighborhood compatibility and development impacts.
The city's broader regulatory concerns therefore have a factual foundation. Oregon's Data Center Advisory Committee has been examining the economic opportunities and infrastructure challenges associated with rapidly growing data center demand. Electricity affordability, water availability and environmental impacts are among its stated areas of concern.
But the existence of different impacts does not automatically establish different zoning classifications.
Storage and transmission systems also process information. Servers hosting websites, managing cloud applications or retrieving stored data continuously perform computational operations.
Stickney argues that Salem cannot reasonably draw a legal boundary based on processing when the code expressly includes computer centers and server farms.
That argument raises a practical question: At what point does a permitted data center become an excluded computational facility?
A building could contain equipment supporting both cloud hosting and AI workloads. Those workloads could change as tenants and technologies change. The same physical infrastructure might support activities with different purposes at different times.
If classification turns on the intended computing workload, city planners may need evidence that is not ordinarily available through a building's site plan.
Alternatively, Salem could argue that a proposed facility's principal function, designed capacity or operational characteristics provide an adequate basis for determining its use.
The distinction matters because land-use standards must be capable of consistent application. A rule that clearly addresses one development but lacks an administrable boundary for others can create uncertainty for applicants and regulators alike.
Stickney points to separate draft code amendments dated Oct. 19 that reportedly use electrical demand thresholds to distinguish different sizes of data centers. His testimony describes proposals involving 2-megawatt and 10-megawatt thresholds.
Those proposed thresholds would provide a more measurable approach than a classification based solely on broad descriptions of AI or hyperscale activity.
However, those amendments and the proposed interpretation serve potentially different legal purposes. A numerical threshold can establish a prospective development standard without necessarily deciding how the existing code should be interpreted.
The fact that the city is considering more specific rules does not alone prove that its current interpretation is incorrect.
What did Salem intend when it adopted the code?
The proposed resolution also relies on the historical context surrounding Salem's adoption of the Unified Development Code in 2013 and 2014.
As described in the testimony, the city reasons that the classification should reflect the kinds of data centers contemplated when the code was established.
Stickney challenges that reasoning with examples of substantial data center development already underway in Oregon before 2014.
The historical evidence supports an important part of his argument.
Google began operating its company-built data center in The Dalles in 2006. Amazon Web Services launched its US West (Oregon) cloud region in November 2011. Facebook, now Meta, also established a major data center presence in Prineville before Salem's code revision.
Large data centers were therefore not a technology that emerged only after Salem adopted its existing classification.
That history is relevant to any suggestion that the original ordinance could not have contemplated large server facilities.
It does not, however, establish that the architects of Salem's code intended to permit every type of computing operation that might eventually be described as hyperscale.
The terms "AI data center" and "hyperscale data center" are not interchangeable. Hyperscale generally describes a facility's operating scale or computing infrastructure, while AI describes a type of workload. A large cloud facility may support many activities, and AI computing does not necessarily require an entire hyperscale campus.
The historical evidence demonstrates that substantial computing facilities already existed in Oregon. It does not conclusively establish how Salem intended to classify facilities dedicated primarily to today's intensive AI workloads.
Nor does the absence of detailed legislative discussion automatically resolve the question.
Stickney says the city's draft findings acknowledge that individual use categories were introduced without extensive specific analysis. If accurate, that would limit the value of arguments based on a clearly documented legislative intention to exclude certain facilities.
But courts can interpret enacted language using its ordinary meaning, context and purpose even when the legislative history does not specifically address a later technology.
The more consequential question may be whether Salem's proposed distinction is supported by the code's language and structure, not whether officials in 2014 anticipated every application of computing technology in 2026.
The city's previous statements complicate the interpretation
One of the more consequential portions of Stickney's testimony concerns statements he attributes to city staff.
His second supplemental filing quotes summaries prepared for Salem's first two data center town halls, dated Sept. 21 and Oct. 9.
According to the testimony, both summaries described data centers as permitted in several zoning districts without existing size restrictions. They also reportedly identified larger hyperscale AI facilities among the types currently allowed.
If accurately reproduced, those statements would establish that city staff previously described the code in broader terms than those proposed in Resolution 2026-46.
The contradiction would be meaningful, but it would not necessarily make the council's interpretation unlawful.
Planning staff can make preliminary assessments, and those assessments do not automatically bind an elected governing body's later interpretation of its ordinance.
A city may also revisit its understanding of a regulation after further legal analysis.
Still, changes in interpretation deserve explanation, particularly when applicants may have made substantial business decisions based on the city's earlier administration of the code.
Stickney's testimony also cites Salem's decision to pursue a moratorium and prospective code amendments as evidence that the city originally understood large data centers to be allowed.
That reasoning has some force, but it is not conclusive.
Governments sometimes adopt moratoriums and new standards to address uses they already permit. They may also amend ordinances to resolve uncertainty or regulate activities whose impacts are not adequately addressed.
The existence of a moratorium does not prove that every proposed data center qualifies as a permitted use.
It does, however, make the relationship between the city's earlier regulatory approach and its proposed interpretation an important subject for the council's findings.
The original town hall summaries and final draft findings will be particularly important in determining whether the city has adequately reconciled those positions.
Why the notice dispute matters
The legal disagreement became more concrete on Oct. 28, when Atchison answered questions about the city's procedures.
In his correspondence with Stickney, Atchison confirmed that Salem had not submitted post-acknowledgment amendment notice to the Oregon Department of Land Conservation and Development.
He also confirmed that the city had not mailed the property-owner notices associated with qualifying zoning changes under Oregon Revised Statutes 227.186.
Atchison said neither form of notice was necessary because the resolution was not an amendment or zoning change.
Instead, he said Salem was following its legislative land-use procedures under SRC 300.1110 and had provided notice consistent with those requirements.
The dispute is therefore not primarily about whether Salem overlooked a notice requirement. It is about whether the disputed requirements apply to this particular action.
ORS 197.610 establishes notice procedures for changes to acknowledged comprehensive plans and land-use regulations. Oregon's Department of Land Conservation and Development generally requires advance notice of proposed amendments.
ORS 227.186 separately addresses notice to affected property owners when qualifying zoning changes limit or prohibit previously allowed land uses. The statute also provides that legislative acts relating to comprehensive plans, land-use planning or zoning must be adopted by ordinance.
Stickney argues that the practical effect of Resolution 2026-46 would be to remove previously allowed uses from multiple zoning districts. He reasons that an action with that effect should be treated as a regulatory change regardless of whether it is labeled an interpretation.
The city draws a different distinction.
If the council is merely establishing the correct meaning of existing language, Salem could argue that it is not withdrawing permission for a previously lawful activity. It is identifying the boundaries of the classification that have existed since the ordinance was adopted.
That distinction has legal consequences.
A valid interpretation can affect future land-use decisions without necessarily requiring the same legislative process as an amendment. But the power to interpret an ordinance is not unlimited, and a resolution cannot simply replace the ordinance's substantive requirements.
The city's position is therefore strongest to the extent that its interpretation can be justified by the existing code. Stickney's position depends substantially on demonstrating that the resolution adds restrictions rather than explains existing ones.
The legal effect of the council's action, not simply its title, will be central to any eventual review.
The pending Oakline application
The immediate stakes are visible in the city's own planning records.
Alderstone LLC, a Verrus subsidiary, filed application 26 115385 00 PLN on July 31 for the proposed Oakline at Mill Creek data center campus in southeast Salem.
The city's permit portal identifies a development involving two single-story buildings with a combined footprint of approximately 600,000 square feet, together with associated equipment, parking and a proposed lot for an on-site electrical substation.
The project is located in the Mill Creek Corporate Center, where employment-oriented zoning is intended to accommodate economic activity and industrial development.
The application has not been approved.
Salem notified Alderstone on Aug. 28 that the application was incomplete. The city's public portal continues to identify the case as "Application Received," with additional materials submitted during the review process, including documents dated Oct. 26.
The distinction between filing and completeness is legally important.
Under ORS 227.178, Oregon generally requires a city to determine whether an application is complete and to provide an opportunity to submit missing information. For qualifying applications completed through the statutory process, subsection (3) protects the standards and criteria applicable when the application was first submitted.
The statute is commonly known as the goal-post rule because it limits changes to applicable approval standards during review.
Stickney argues that the July 31 filing date prevents Salem from applying a new exclusion adopted in November.
That would be a consequential protection if the resolution creates a new substantive standard.
But if Salem's interpretation is legally valid as an explanation of the code already in effect on July 31, the city could argue that applying it does not move the goal posts.
A separate issue is whether the application ultimately satisfies the statutory completeness conditions. The city's current portal status does not conclusively establish that the application has been deemed complete.
The Aug. 28 notice also identified Jan. 27, 2027, as the deadline for the applicant to respond to the identified deficiencies under the applicable 180-day process.
The ordinary deadline for final city action generally begins when an application is deemed complete, rather than simply when it is first filed. Statutory exceptions and agreed extensions can also affect that period.
These distinctions mean the pending application has procedural protections, but they do not establish that approval is required or that the city has already violated a decision deadline.
Could the resolution decide Oakline's future without naming it?
Stickney's second supplemental testimony raises a related concern about the practical effect of the interpretation.
He attributes two statements to the city's October town hall materials.
One reportedly explained that proposed zoning amendments would not apply to the pending Verrus application. Another reportedly indicated that adopting the code interpretation would prevent the proposed Verrus development from being allowed.
The original town hall summary is needed to establish the precise context of those statements.
If accurately characterized, however, the comparison illustrates why the distinction between interpretation and amendment has become consequential.
Prospective changes to a development ordinance may be constrained by statutory protections for applications already submitted. A valid interpretation of existing law may operate differently.
That creates a substantial incentive for all participants to establish whether the proposed resolution is genuinely interpretive.
It does not, by itself, demonstrate an improper motive.
A citywide interpretation can be prompted by a specific development without becoming an unlawful decision on that development.
Likewise, a developer's interest in protecting a pending application does not make its legal objections invalid.
The proper inquiry concerns the decision's substance and the procedures that apply to it.
Stickney has also questioned whether the council's action could be quasi-judicial rather than legislative in substance because of its potential effect on one pending project.
Atchison expressly describes the resolution as legislative.
Oregon law distinguishes broadly applicable legislative land-use actions from quasi-judicial proceedings applying legal standards to particular circumstances. The fact that one development could be especially affected is relevant context but does not automatically determine the proceeding's legal classification.
If quasi-judicial hearing requirements apply, the relevant statute is ORS 197.797, including its procedures for evidence, public participation and opportunities to respond to new material in the record.
Those protections cannot simply be assumed to govern a legislative interpretation. Their applicability would depend on the nature of the proceeding.
What Oregon courts have said about local interpretations
Both sides' positions must also be evaluated under Oregon's rules concerning deference to local governments.
ORS 197.829 requires the Land Use Board of Appeals to affirm certain interpretations of local comprehensive plans and land-use regulations, subject to limitations.
The Oregon Supreme Court's decision in Siporen v. City of Medford is important because a plausible interpretation by a governing body may receive deference even when another interpretation is also reasonable.
That principle favors neither a particular developer nor a particular city outcome. It establishes the framework for reviewing the council's interpretation.
Salem would not necessarily need to demonstrate that its interpretation is the only possible reading of SRC 400.110(e).
It would, however, need to remain within the boundaries of a legally permissible interpretation.
A construction inconsistent with the ordinance's express language, purpose or underlying policy may fall outside the applicable deference standard.
The city has also cited Save Stafford Road v. Clackamas County, a September 2026 Oregon Court of Appeals decision involving a Portland General Electric transmission-line project.
The decision discusses deference to local interpretations and the obligation to address applicable comprehensive plan policies where those policies form part of the relevant approval criteria.
Stickney contends that the decision reinforces limits on adding requirements not found in an ordinance. He also argues that Salem's findings should address the economic development and employment-land policies relevant to Mill Creek.
The case supports the importance of reading ordinances carefully and addressing applicable approval standards.
But its comprehensive plan holding arose from a specific Clackamas County criterion requiring consistency with relevant plan policies.
It does not automatically establish that every general interpretation must separately analyze every potentially relevant comprehensive plan policy.
The decision therefore provides useful legal principles without resolving Salem's particular dispute.
Economic development and infrastructure are separate questions
Much of Salem's broader data center debate concerns the consequences of accommodating large computing facilities.
Potential economic benefits include private capital investment, construction employment, property-tax revenue and activity associated with supporting infrastructure.
The proposed Oakline development has been presented as a substantial investment in the Mill Creek Corporate Center.
But projections of investment or employment should not be confused with verified economic outcomes. The number and quality of permanent jobs, public costs, tax treatment and infrastructure commitments would all matter when assessing the project's eventual net benefit.
The infrastructure side also requires careful analysis.
Large computing facilities can place substantial demands on electrical systems, while cooling technologies and operating designs can affect water requirements. Backup generators, noise, utility upgrades and nearby land uses create additional questions.
The extent of those impacts depends on the specific project. They cannot be established solely by applying the label "AI data center."
Salem's Data Center Task Force was formed to consider these issues and develop recommendations balancing economic, environmental and community interests.
That work has a public-policy purpose distinct from determining what the current zoning ordinance already permits.
A city can reasonably conclude that its existing data center standards are inadequate for modern facilities and pursue new safeguards.
A developer can reasonably insist that those safeguards be adopted through the procedures required by law and applied consistently with statutory protections.
Those positions are not inherently incompatible.
Indeed, the city could ultimately adopt more specific regulations while the pending Oakline application is evaluated under whatever standards are legally applicable to it.
The more difficult question is whether Resolution 2026-46 can independently establish that the existing code never permitted the proposed category of development.
The questions the council must resolve
The November hearing is likely to turn on several issues that require more precise answers than a general debate over whether data centers are desirable.
The first is textual: What language in SRC 400.110(e), considered alongside the city's principal-use classification rules, supports excluding AI and hyperscale facilities?
The second is administrative: How will planners distinguish an excluded computational facility from a permitted server farm or computer center, particularly when tenants and computing workloads may change?
The third is procedural: If the resolution changes which uses may be developed in Salem's zoning districts, does Oregon law require an ordinance and the notice procedures associated with amendments or qualifying zone changes?
The fourth concerns the existing application: Will Salem apply Resolution 2026-46 to Oakline, and if so, on what legal basis does that application remain subject to the interpretation?
The fifth is evidentiary: How will the city's final findings address its earlier staff descriptions, the historical presence of large data centers in Oregon and the competing readings of the code?
A well-supported decision need not satisfy every participant. It does need to identify the governing standards, explain the reasoning and accurately characterize the record.
For Salem's business community, that process matters beyond the outcome of one project.
Businesses considering major investments must evaluate land costs, infrastructure requirements, permitting timelines and regulatory risk. Residents and local governments must also be able to understand how new development will affect public services, utilities and surrounding land.
Neither investment predictability nor protection from infrastructure impacts eliminates the other concern.
The relevant legal procedures exist in part to make those competing interests subject to defined standards and review.
A consequential decision still ahead
As of Oct. 30, Resolution 2026-46 remains a proposal. The City Council has not yet adopted the interpretation, and the pending Oakline application has not received final land-use approval.
Stickney has requested that the council decline to adopt the resolution and instead pursue the code-amendment process already underway. If the council proceeds, he wants revised findings and an express statement that the interpretation will not apply to the July 31 application.
Atchison's written position is that Salem is acting within its authority to interpret existing regulations and is following the procedures applicable to that action.
The unresolved issue is not simply which interpretation better serves the city's data center policy objectives.
It is whether the interpretation the council adopts can be reconciled with the words Salem placed in its development code, the procedures required by Oregon law and the standards applicable to a development application already under review.
The council's Nov. 9 hearing will provide the next opportunity for those distinctions to be addressed before a final decision.