On September 17, ECOnorthwest and the University of Virginia's Weldon Cooper Center released the most complete public picture yet of Oregon's data center industry. It is careful work. The authors say plainly that the goal is to inform, not to advocate, and they show their method.
The number that traveled was this: data centers accounted for about 22.8 percent of Oregon's retail electricity sales in 2025.
Here is what happened to it. The same day, one outlet reported it as about 23 percent of retail electricity sales, and that outlet's own morning newsletter described it as 23 percent of statewide electricity demand. Two days later, a broadcast station's post said data centers now consume 22.8 percent of the state's electricity. Headlines rounded it to "almost 25 percent" and "nearly a quarter."
Each step was small. Together they turned a share of one thing into a share of something else.
Why the denominator matters
Retail sales, total consumption, and total demand are different quantities. A figure that is correct against one can be wrong against another. When a denominator changes silently, a correct number becomes an incorrect claim — and no one along the chain did anything more than paraphrase.
What the preliminary figures show
The preliminary findings presented to Oregon's Data Center Advisory Committee on July 31 lay out how the estimate is built: installed IT capacity, adjusted for utilization, cooling overhead, hours in the year, and load factor, drawn from a commercial facility database. It is a modeled estimate, not metered consumption. That is a standard and reasonable approach, and the authors say so.
Those preliminary figures put data center demand at 14.0 terawatt-hours in 2025, rising to 24.8 by 2030 — about 1.8 times. A widely shared post said demand could "nearly triple" by 2030. The preliminary figures don't support that. If the final report revised the projection, we have not seen it; the coverage cites the same roughly 25 terawatt-hour endpoint.
One more detail the coverage dropped: the projected 2030 share compares future data center demand against Oregon's 2025 statewide demand. If statewide demand grows, the actual 2030 share will be lower. The preliminary deck says this on its face.
Four counts, one month
How many data centers does Oregon have? This month's public sources say 111 operating; about 123; between 123 and 144 depending on definitions; and 98 operating plus 36 planned. Each is defensible under its own definition. None is wrong. Each needs its definition attached to travel safely.
The rule
State the denominator. A figure without one isn't wrong — it's unfinished.
That matters right now. Oregon's advisory committee is taking public comment through October 24, and counties across the state are deciding whether to pause, regulate, or wait. A number that drifts on its way into a public hearing can end up supporting a decision it was never measuring.
None of this is a criticism of the researchers. Their work shows its method. The qualifiers were lost downstream, one reasonable paraphrase at a time.
- Oregon Data Center Advisory Committee, Preliminary Learnings and Questions Document, September 10, 2026 — read in original
- ECOnorthwest and UVA Weldon Cooper Center, preliminary findings presented to the DCAC, July 31, 2026 — read in original
- Coverage by OPB, KLCC, and KGW, September 17–19, 2026
- The final 135-page report, to which the 22.8 percent figure is attributed, has not yet been read. That figure has not been checked against it.
Calyx Intelligence maintains a public register of data center siting, power, and public finance provisions — each entry marked for whether it was read in the original document and whether it reconciles. register.calyxos.ai