The financial case that deserves a public hearing.
Every number here comes from the District's audited financial statements or its own billing and planning records. Together they show a profitable segment, a wireless line that's about a third of its revenue, and a one-time upgrade cost that's small against what's at stake. The deeper question they raise: stepping back is easier for the District — but that isn't the same as better for the county.
This is a lower-risk decision than the day the network was built — and a more necessary one.
Before the line-by-line finances, here is the whole case on one screen. The District is not being asked to gamble on something new. It's being asked to refresh a network it already owns, that already pays its own way — a far safer bet than the one that created it, and a more urgent one than it was in 2010.
Projected monthly rate — a cost-based public rate that holds, vs. a for-profit bill that raises a little every year. The shaded wedge is the customer's growing value.
ILLUSTRATIVE Both lines start at the District's proposed $40 to isolate the effect of rate increases alone — in reality the monopoly fallback already starts far higher (Starlink ~$120). The public line holds flat, which is conservative: by law a PUD can only recover cost, and the durable share of its costs falls as debt retires. The for-profit line rises ~5% a year — in line with real broadband hikes (AT&T raised home-internet prices in 2023, 2024, and 2025), and below the 6.1%/yr the FCC documented for cable as a long-run analogy (DA 14-672); at that pace a bill doubles in about 14 years. Over 15 years the wholesale gap compounds to roughly $3,000–$4,000 per connection at the floor. We chart the rate the PUD controls (wholesale); each ISP adds its own retail markup — but because the cost-based floor underneath doesn't ratchet, the public option's bill still climbs far slower.
Lower risk than at first install
Every unknown from 2010 is now a known.
- The market exists now. In 2008 no ISP sold on the network — the District built the towers and hoped resellers would come. Today ~8 retail ISPs and 3,800+ paying customers already ride it. The gamble is now a going concern.
- Demand is proven. The first build bet millions that rural households would buy. Today the network is “severely congested with no possibility for expansion” — its own memo. You don't congest a network nobody wants.
- The technology is proven. Fixed wireless was new and untested across this terrain in 2010. The upgrade path (ngFWA / Tarana-class) is mature and deployed across rural America today.
- The asset is built and paid down. 2010 meant ~$9.2M of new debt to build ~200 miles from nothing. 2026 is a ~$1.2M refresh of radios on towers that already stand — ~⅛ the size, a fraction of the risk.
- It already earns. The first build had zero revenue on day one. Today the segment runs a surplus, and the upgrade pays itself back in a modeled ~5–6 years.
More necessary than at first install
And the reasons to act have only grown.
- A monopoly is now pricing by captivity. There was no Starlink in 2010. Today it's the dominant fallback and, by its own surcharges, charges more where rural customers have nowhere else to go. The public network is the check that holds that in line.
- The fiber alternative got more expensive, not less. Fiber here runs ~$40K/home (~9× denser counties), and ~$50M in federal fiber grants for the county were lost or returned. Wireless is the only last mile that pencils for the rural majority.
- The price is cost-based, and the District controls it. Its own Tarana plan proposes a single wholesale rate of $40/mo for 250 Mbps — about the old top-tier price for several times the speed. A for-profit prices to a margin and ratchets; the public utility prices to cover cost, so the gap only widens.
- The network is full — and slipping from neglect. It can't sign new customers, and reinvestment fell ~92% since 2021. Waiting doesn't hold the line; it loses ground that costs ~$40K/home to ever rebuild.
- Walking away is the irreversible move. Keep the asset and the District can still adjust later; abandon it and a rebuild from scratch may never come. The reversible, low-regret choice is to maintain what you already own.
The cautious move and the bold move have switched places: spending ~$1.2M to keep a paying asset running is the conservative choice — letting it go is the gamble.
The District has fears. The record has answers.
The pause rests on four claims. We requested the analysis behind them: there is none — no market study, no financial analysis, no consultation with the seven providers who hold the customer data. These aren't findings; they're fears. And they're being weighed against an asset the District already owns — built with $9.2M of federal money, earning $1.1M a year, sold by a channel it doesn't pay. We took each fear seriously anyway: researched the strongest case for it, and tested it against the District's own records.
FEAR 01“NCI is investing — it will take the market.”
Evidence offered: none in the recordWhat the records show: the experiment already ran. NCI has operated faster towers across this county for years — and the local providers held (−5% over 5½ years) while NCI itself paid the District more every year for wireless, fiber, and transport. A network that held at a thirty-fold speed disadvantage has no reason to do worse at speed parity. This fear comes true in exactly one scenario: the frozen one.
FEAR 02“Starlink is cheaper for more bandwidth — it will flood the market.”
Evidence offered: a price comparison — no studyWhat the records show: this is the serious one — and it's an argument for the upgrade. Starlink is also a different product: self-install, self-support, no guaranteed speeds — a DIY utility, where the local service is installed, supported, and accountable, for about the same monthly dollar. Even against our broken 20 Mbps gear, only ~15% of cancellations chose it; where local providers are good, the world's best market data (New Zealand's regulator) shows they keep growing under maximal Starlink pressure. Starlink beats one thing every time: a network left standing still.
FEAR 03“Even if no one rival wins, the combination will make our share spiral.”
Evidence offered: none in the recordWhat the records show: both rivals have been here the whole time — Starlink selling county-wide since 2021, NCI's faster towers throughout — and under both at once the local providers grew to their peak, plateaued for three years, and drifted only after the District stopped supplying radios (new hookups: 229 a year → 2). A spiral needs a mechanism — fixed costs forcing rate hikes as customers leave — and this network's build is sunk, its rates cost-based by law, its sales channel free. The arithmetic that makes networks spiral doesn't exist here.
FEAR 04“The system has been in active decline for years.”
Evidence offered: subscriber counts — the one real datumWhat the records show: half true — and the District is citing the wrong half. The counts fell after the budgets cut wireless reinvestment 92% and the District all but stopped connecting new customers; more than half the recent drop is one reseller winding down its resale channel. Meanwhile revenue grew into the District's largest telecom line, customers climbed toward the 20 Mbps ceiling, and total last-mile connections rose. What's declining is the hardware — end-of-life gear from a vendor now in going-concern distress. That isn't a reason to stop. It's the deadline.
They didn't run out of money. They stopped reinvesting — and the network started shrinking.
Here's the distinction that matters: we do know where the money went — it's in the District's own capital budgets. What we don't know is why, and the District hasn't explained it. So we're not guessing at motives — just reading the record, and the pattern is hard to miss.
Budgeted reinvestment in the wireless network fell about 92% — from roughly $330,000 a year in 2021 to about $26,000 in 2026 — and the subscriber base, which had been growing, turned and began falling right as that investment hit the floor. This isn't a network that ran out of money: the telecom segment's audited net worth more than doubled over the same period. It's a profitable network that was quietly under-reinvested until it started to slip.
So where did the reinvestment go? The telecom fund's own capital tilted toward fiber distribution — a few hundred thousand dollars a year — while the wireless side fell to near zero, and a surplus was set aside every year. None of that is hidden; it's in the District's own budgets, and we're not second-guessing the fiber work.
The honest question is just about the other side of the ledger. The District is investing about $4.3 million in fiber to Conconully — a worthwhile build for a community that genuinely needs it — while the ~$1.2 million wireless upgrade — the first phase, modernizing 878 customers on seven towers within the ~2,700-home wireless network — is not in the 2026 budget. The upgrade is affordable from the segment's own surplus, and at its June 22, 2026 meeting the Board agreed to take it up in a workshop. We're asking it to follow through. Follow the money → · Who the wireless network serves →
Want to share it with neighbors? Grab the print-ready community flyers → — one per argument, plus a three-page version — over on the action page.
Hearing the usual objections — "government shouldn't compete," "municipal broadband fails," "just use Starlink"? See Answering the Critics →, the honest response to each, using the skeptics' own evidence.
Want the numbers in CFO terms? The full discounted-cash-flow analysis → — NPV, IRR, a 10-year pro-forma, break-even and sensitivity, built from the District's audited financials and its own Tarana cost workbook.
Most decisions can be revisited. This one can't.
The pause is being treated as a deferral — step back now, step back in if it's ever needed. But the District's ability to stabilize rural broadband isn't a switch it can flip back on. It's a 25-year stock: a working network, a base of paying customers, a half-dozen small local providers, a trained local crew, public trust, and the one-time federal money that built it all. Spend that down and it doesn't come back on demand. Everything else on this page is a reason the network is worth keeping. This is the reason you can't afford to find out what losing it costs.
This isn't rhetoric — it's how markets with sunk costs behave. Economists call it hysteresis: when both entering and leaving a market cost money you can't recover, exit and re-entry aren't symmetric. You give up the network at conditions far worse than the ones you'd need to justify rebuilding it, so a "temporary" retreat becomes a permanent loss (Dixit, Journal of Political Economy, 1989; Baldwin & Krugman, Quarterly Journal of Economics, 1989). What's actually at stake, link by link:
- The customers. A household that gives up and buys a ~$350 Starlink dish has sunk a cost that works against ever coming back — and the research on won-back telecom customers is blunt: most never return, and the ones who do are the likeliest to leave again (Journal of Marketing Research, 2018).
- The local providers. The seven retail ISPs on this network own no towers of their own — the open-access platform is their business. Starve it and they lose customers, then they fold. You cannot reconstitute a half-dozen local companies, with their customer relationships and on-the-ground knowledge, on command.
- The workforce. A crew that knows these towers is hard to reconstitute — and the labor pool is thin: the industry is short an estimated 178,000 broadband workers (Fiber Broadband Association), and training a technician takes 12–24 months. A local crew that disperses isn't quickly rehired.
- Public trust. "Trust is fragile. It is typically created rather slowly, but it can be destroyed in an instant" (Slovic, Risk Analysis, 1993). A quarter-century of credibility with ratepayers and providers is not rebuilt on a budget cycle.
- The money that built it. This network exists because of one-time federal grants (NoaNet's ~$84M stimulus award; the USDA-funded 2010 build). That windfall is not on tap: about a third of the FCC's $9.2B RDOF program collapsed in defaults, BEAD's rules were rewritten mid-stream in 2025 — and the District just walked away from a $30M fiber grant it already held (being rescinded unspent — Resolution 1863). "We'll get a grant and rebuild later" assumes a door that may not open — and the District just let one close on money already in hand.
So the real question for the Board isn't only "does the upgrade pay for itself?" It's "is this the decision we can least afford to get wrong?" The cheap, low-risk time to act is while the network, the customers, and the providers are all still here — a second chance, if it comes at all, comes slower, costlier, and on someone else's terms.
A profitable public asset — not a money-loser.
From the Washington State Auditor's Office report on the District's 2024 financial statements (Note 9 — Telecommunications):
The audit reports one wholesale total. We split it.
The auditor reports the wholesale telecom line ($2,808,173 in 2024) as a single figure — it does not separate wireless from fiber. Using the District's own monthly billing reports (2021–2026, by public-records request), wireless works out to roughly $1.1 million per year — about a third of telecom operating revenue, and it has held steady for five years. This is exactly the analysis that should have been in front of the Board when it voted.
One honest limit, stated plainly: that's revenue, not standalone profit. The audit doesn't break out wireless's own cost-to-serve, so neither the District nor we can call the wireless line independently profitable — only that it's roughly a third of the segment's revenue and the segment as a whole runs a surplus. If the District holds a wireless-specific margin analysis, it should put that on the record too. See the full data and methodology →
The risk isn't investing. It's not investing.
The District's plan — upgrade only "should future subscriber enrollment reach a level sufficient to offset the costs" — treats the upfront cost as the risk. From an accounting and stewardship standpoint, that's backwards: it treats a one-time, bounded number as the risk while ignoring the recurring revenue and public asset that "wait and see" puts at risk.
The District's own stated cost — a one-time, bounded number you can plan around. It's on the order of one year of the revenue the wireless line brings in (~$1.1M/yr), and about 1.5 years of the segment's $808K annual surplus. A full deployment could run somewhat higher; either way it's a single capital decision, not a recurring drain.
You're weighing a one-time cost against a recurring revenue line. No need to assume the network goes to zero: even a partial, gradual decline of a third-of-segment revenue stream adds up — over a few years — to more than the one-time upgrade, while the segment keeps carrying its fixed costs. Behind it sits a public asset whose wireless layer was built with ~$9.2M in 2010 federal funds (the loan refinanced in 2020, now a small balance).
"Wait and see" isn't the cautious choice — it's the expensive one. And it's a trap: subscribers won't climb on a slow, unsupported, end-of-life network, so the District's own condition for upgrading can never be met. Inaction guarantees the very decline it's trying to avoid.
Even if a modernized network underperformed, it wouldn't fail overnight. A base of ~2,700 customers doesn't vanish in a year, so a one-time cost is small against the revenue that keeps flowing while any decline plays out. The downside of acting is bounded and one-time; the downside of doing nothing is recurring and compounds. And the evidence — a profitable segment, five years of steady revenue, a model that just weathered Starlink's cheapest-ever pricing — points to that revenue being defensible, not doomed.
A prudent steward weighs a one-time cost against a recurring revenue line and a multi-million-dollar public asset — and, on those terms, acts. Treating a bounded, one-time outlay as the bigger risk than an open-ended decline isn't conservative — it's the gamble.
Across Washington, PUDs keep investing in broadband. Okanogan paused the wireless its rural majority relies on.
The District will say it is investing — and it points to fiber. Fair. But its peers are expanding their networks while Okanogan freezes the wireless layer that actually serves the dispersed rural majority — and several of those peers are doing it on weaker financials than Okanogan's profitable wireless line.
Jefferson County PUD
Put about 40% of its 2026 capital budget — over $15 million — into broadband.
Kitsap County PUD
Building now on a $15 million federal award plus a state match — 21 new fiber nodes in early 2026.
Pend Oreille County PUD
A rural, open-access PUD that — like Okanogan — built its network with a federal stimulus grant. It's now expanding fiber on a fresh ~$12.4 million federal award plus a $4.1 million state match, and still runs a wireless layer too. Fitch rates it 'A', stable.
These are all open-access public utilities, like Okanogan — and all are growing their networks. Okanogan is the one pausing its wireless layer, the part that reaches the rural majority fiber won't. And the District's own fiber grant? $30 million awarded in 2023, $0 of it spent two years later, and the District has since signaled it may not proceed — so "we're still investing in fiber" doesn't answer for the ~2,700 customers who depend on wireless now, or the rural addresses fiber won't reach for years. (Jefferson 2026 budget; Kitsap BEAD award + 2026 build; Pend Oreille BEAD award & Fitch 'A' rating, 2025; ReConnect grant per the District's FY2024 audit.)
Yes — and it's worth being precise about what "it" buys.
The letter's core condition is that the District will reinvest only "should future subscriber enrollment reach a level sufficient to offset the costs of upgrading." That's a recovery question, so we answered it from the District's own Tarana cost workbooks — and we'll be exact about the scope, because the scope is the whole point.
What the ~$1.2M actually covers. Not the entire network — seven tower sites (Jackass Butte, Coleman, Nortons, Pickens, Number, Eder, Shellrock) that today serve 878 customers. The cost splits cleanly: about $371,000 of fixed tower hardware, plus a ~$787 radio for each customer served. That structure is the key to the whole recovery question.
It's long-lived utility infrastructure, recovered at wholesale rates. Each upgraded customer costs about $1,200 all-in (~$1,340 if the full 1,035-radio budget is spread over today's 878) and returns the District's proposed $40/mo wholesale rate for the new network — about $480 a year (the retail margin stays with the local provider). At a plausible wireless operating margin, that pays the upgrade back within the equipment's service life — the way the 2010 wireless build did, gear that's still earning ~$1.1 million a year sixteen years later:
What the upgrade buys, tower by tower
From the District's own Tarana upgrade workbook — one sheet per site. The seven add up, from the bottom, to the ~$1.2M the District has stated.
| Tower site | Customers today | Upgrade cost | Per customer | Payback, alone* |
|---|---|---|---|---|
| Eder • phase 1 | 241 | $274,499 | $1,139 | ~5.3 yr |
| Jackass Butte • phase 1 | 132 | $156,713 | $1,187 | ~5.5 yr |
| Coleman • phase 1 | 170 | $227,385 | $1,338 | ~6.2 yr |
| Number Hill | 83 | $112,952 | $1,361 | ~6.3 yr |
| Nortons | 83 | $124,863 | $1,504 | ~7 yr |
| Pickens | 125 | $188,575 | $1,509 | ~7 yr |
| Shellrock | 44 | $93,383 | $2,122 | ~9.8 yr |
| All seven | 878 | $1,178,370 | $1,342 | ~6.2 yr |
*Standalone payback — each tower's cost over its own customers' annual wholesale cash flow, at the proposed $40/mo tier and an estimated ~45% cash margin. Every site clears within the gear's ~10-year life; the cheapest-per-customer sites pay back fastest, so a phased build starts with them.
| If the wireless operating margin is… | …the ~$1.2M pays back in about |
|---|---|
| 24.9% — the District's published net margin (after depreciation) | ~10 years — right at the gear's ~10-yr life (the most conservative reading) |
| 30% | ~8 years |
| 40% | ~6 years |
| 50% | ~5 years |
Even at the District's own published 24.9% net margin — the most conservative reading, struck after non-cash depreciation — the upgrade recovers right at the edge of the equipment's ~10-year service life, and every margin above that pays back faster. The recovery chart below uses the cash margin (the right measure for repaying capital, since depreciation isn't a cash cost), which is why its undiscounted simple payback is the faster ~5–6 years (discounted and risk-modeled, the median recoup is ~9–10 years).
Which margin? The cash one — and the District's own audited operating income lets us estimate it. The widely-quoted 24.9% is the segment's net margin, struck after more than $1 million of non-cash depreciation — the wrong figure for asking whether new capital gets repaid. The State Auditor's 2024 report shows telecom operating income of $647,484, and adding that depreciation back gives roughly $1.5 million of operating cash flow on $3.25 million of revenue — about a 45–50% cash margin. At that margin the District's proposed $40 plan recoups the upgrade well inside its service life — a median ~9–10 years once the cash flows are discounted and stress-tested, against the gear's ~10–11-year life (the gear it replaces lasted about sixteen).
Share of the upgrade cost recovered, over time
Percent of the ~$1.2M recovered over time, scoped to the customers it serves (878 today, up to the 1,035 the District budgeted), at the proposed $40 tier. Where a line crosses 100%, the upgrade is square; the dashed vertical is the equipment's ~10-year service life. One basis note: these curves charge the radios actually built for 878 customers (~$1.06M); on the full ~$1.18M workbook (which pre-funds radios for 1,035) the same scenarios run ~0.8 yr longer — the conservative basis used in the financial analysis.
All six lines start from the District's proposed $40 tier, each adding one lever, and all are undiscounted simple payback — the plainest reading of the cash flows. The conservative case stacks both adversities — a 38% cash margin (below the ~45–50% base) and a full NCI exit as an RSP from the District's network (~685 served) — and still recovers in ~7 years, so a one-time NCI exit is built in. (These are deterministic lines, not the model's floor: the fully discounted, stress-tested simulation puts the District's own recoup at ~88% — counting the separately-billed bandwidth resale conservatively (~76% if bandwidth were zero, ~97% at the documented ~2.8 Mbps) — in the central case, and the county ahead in ~99% (~88–100% across all three scenarios). A District miss isn't a loss — the towers and 200-mile backbone still stand and customers keep paying. See the three scenarios →) Reading the levers off the chart: more sign-ups at the same $40 price — the upgrade reaches more homes — fills toward the budgeted 1,035 radios and crosses sooner while recovering more overall; adding tiered pricing crosses sooner still; and if it's popular — tiering plus growth into that 1,035 capacity at a top-of-range margin, and in the best case premium tiering plus growth into the new homes the upgrade reaches (~1,200 subscribers, returning customers and non-line-of-sight signups) — the earliest crossings on the chart. And the lines don't just cross sooner, they end higher: the popular case recovers nearly three times the cost over the gear's life, the best case ~3.5×. Discounted and risk-modeled, the median payback across these is ~9–10 years, within the gear's ~10–11-year life. Honest caveat: subscriber growth on its own mostly raises lifetime return more than it speeds payback — each new customer brings its own radio cost, so margin and pricing are the faster levers, but growth compounds with both, which is what the top two lines show. One more honest note: every line holds the served base flat for the gear's life — no ongoing attrition or mid-life radio swaps are modeled — but even a steady ~3–5%/yr net loss would still land payback inside that gear life.
And it's affordable today, whatever the exact payback clock. The ~$1.2M is about 1.5 years of the telecom segment's $808,166 annual surplus — a segment whose audited net worth more than doubled to $10.5M, funded from telecom, not borrowed and not drawn from electric ratepayers. And it's a remarkably cheap network to run — the District pays $0 for internet transit (it peers through NoaNet) and just $34,700 a year to lease all nine tower sites (the biggest, Eder, is free). For a utility, this is an ordinary infrastructure refresh, not a bet it can't carry.
The cost follows the customers — and the base is loyal to the network, not to any one reseller.
NCI, the second-largest reseller, may move its book onto its own gear and leave the PUD network. It barely changes this, for two reasons. First, the cost is per-customer: the District buys a ~$787 radio only for a customer it actually serves, so NCI's exit removes its revenue and its cost together. Only the fixed tower hardware — about a third of the total — re-spreads over fewer customers, extending payback by roughly 9%, not 25%. Second, those customers stay in the county. On an open-access network a reseller's book can move — to its own towers, or to one of the other local providers — without the customers ever leaving the network or the valley. The base that pays this back is sticky to the network, whoever bills it.
The honest limits. The District has never published a wireless-specific operating margin — only the whole telecom segment's (24.9% net in 2024, after depreciation and including fiber). Wireless runs on largely-depreciated 2010 plant, so its cash margin is plausibly higher — but we don't assert a number we can't source, which is why the payback above is shown as a range. The other input is the equipment's service life; the last generation lasted about sixteen years. If the District holds a wireless cost analysis, it should put it on the record — it would sharpen this either way. The model and assumptions are downloadable so anyone can check or challenge the math.
And every payback above assumes zero growth.
That's the conservative way to score it — but it understates the case, because the upgrade doesn't just hold the customers it has. It's next-generation fixed wireless that reaches homes the old line-of-sight radios couldn't (trees, terrain), and in the District's own Jackass Butte field test a sector built for a 90° view measured usable coverage closer to ~160°. The capacity to grow is already bought: the $1.2M workbook funds 1,035 customer radios — 157 more (+18%) than the 878 served today. Because the towers are a fixed cost and those radios are already budgeted, each added customer is almost pure acceleration:
| Hypothetical sign-ups | Subscribers | Payback | Added over 10 yr |
|---|---|---|---|
| Today — as served now | 878 | ~6.2 yr | — |
| If the 100+ who said they would return do | 978 | ~5.6 yr | +$216,000 |
| Fill the radios the $1.2M already buys | 1,035 | ~5.3 yr | +$339,120 |
| +25% via non-line-of-sight reach | 1,098 | ~5.2 yr | +$475,200 |
Hypothetical sign-up scenarios at the conservative plan values ($40/mo, ~45% cash margin, $216/yr per customer). Sign-ups up to the 1,035 the budget already funds add ~$0 in new cost. The maps, the field-test speeds, and the full technology write-up are on the Data page →
The real comparison isn't whether the cost comes back — it's this: invest, and the customers and their revenue stay and grow on the network; wait, and the aging gear sends them to satellite, taking the very revenue that would have paid for the upgrade with them.
$4.3 million of new fiber connects 367 homes. A $1.2 million upgrade modernizes 878 wireless customers.
The District's own budget prices the contrast. Its newest fiber project — the Okanogan-to-Conconully middle mile and fiber-to-the-home build — is $4.3 million to connect 367 new homes. The $1.2 million wireless upgrade it declined modernizes 878 customers on seven towers — part of a network already serving ~2,700 rural homes — at a fraction of the cost per home, even doing a full equipment refresh.
What each project actually connects
Cost per home — fiber: $4.3M ÷ 367 ≈ $11,700; wireless: $1.18M ÷ 878 ≈ $1,340.
Honest caveats: the $4.3M fiber figure includes ~22 miles of middle-mile backbone that will serve future homes too, so its per-home cost falls as more connect; and fiber is a faster, longer-life product than wireless. The point is reach per dollar today, not "fiber is worse." Per the District's Tarana workbook, the $1.18M upgrade covers seven towers serving 878 customers — part of a wireless network already serving ~2,700 rural homes; the 367 are new fiber connections.
Who pays for the $4.3 million — and it isn't BEAD
Washington State Public Works Board award (May 2024) plus county ARPA. The District repays the loan portion.
None of this makes the Conconully build wrong — rural homes deserve fiber where it reaches. But it reframes the District's "wait until it pays" stance on wireless: it is willing to take on $2.66 million of debt to wire 367 homes, while calling a $1.2 million upgrade that already serves 878 — and could reach far more per dollar — unaffordable.
The District demanded commitments — then gave no one a way to make one.
Asked what it would take, the General Manager said he'd reconsider only if customers "signed contracts" and enrollment rose. Fair as a condition — but the May 4 letter defined no commitment, set no threshold, and offered no one a way to sign anything. It demanded a test and withheld it. Yet the demand is already there — more than 100 former customers told a local provider they would return if the network were modernized, most citing the network's lack of any forward trajectory, not a single speed or price. And other Washington PUDs have run exactly this play for years.
First — what "a commitment" can even mean. There's a ladder, softest to firmest: a petition (signatures of interest); a service commitment (a subscriber agrees to take service and pay a small construction adder on their monthly bill once it's built — this rides the account, so renters qualify too); a provider commitment (the retail ISPs guarantee the wholesale purchase and carry the risk themselves); and a local-improvement assessment (an owner option, repaid like a small line on the property-tax bill). The District never said which it meant — so here's how every one of them works, and who would be on the hook.
Who carries the risk? Pick any — or split it three ways.
The customer
Subscribers pay a small construction adder on their internet bill once it's built — roughly $5–$15/month over 12 years (less over 20), the way Mason PUD 3 does it at $25/mo. It rides the account, not the property, so renters qualify. Nobody pays a cent until service is delivered.
The providers (RSPs)
The District's actual customers — the local retail ISPs — guarantee the wholesale purchase per tower and shoulder the demand risk themselves; households commit to nothing. The providers have said they're confident enough to do it. This is the cleanest fit for a wholesale network — and it removes the District's capital risk outright.
The District
Or it simply does the upgrade, as it has for 25 years: a one-time ~$1.2M from a profitable segment (about 1.5 years of its telecom surplus), recovered through ordinary wholesale revenue — a bounded, one-time cost against a recurring ~$1.1M/yr line.
All three
Or split it: the District funds the shared tower hardware (~$371K); the providers commit to a minimum take per tower; customers cover only their own radio — about $5.50/month. Divided three ways, every piece is small.
And the customer's share is small — tower by tower
If households funded their tower's full cost (the strongest, all-on-the-customer version), here's what each would commit to — and remember, nobody pays until it's built:
| Tower | Customers | Upgrade cost | Per customer | If customer-funded |
|---|---|---|---|---|
| Eder | 241 | $274,499 | $1,139 | ~$8/mo |
| Jackass Butte | 132 | $156,713 | $1,187 | ~$8/mo |
| Coleman | 170 | $227,385 | $1,338 | ~$9/mo |
| Number Hill | 83 | $112,952 | $1,361 | ~$9/mo |
| Nortons | 83 | $124,863 | $1,504 | ~$10/mo |
| Pickens | 125 | $188,575 | $1,509 | ~$10/mo |
| Shellrock | 44 | $93,383 | $2,122 | ~$15/mo |
| All seven | 878 | $1,178,370 | $1,342 | ~$9/mo |
"If customer-funded" spreads that tower's full cost over its customers as a monthly adder over 12 years; over 20 years it's lower still, and if the District funds the shared tower hardware, each household covers only its own radio (~$5.50/mo). Every figure assumes nobody pays until the upgrade is built.
The District didn't lack a path forward — it declined to offer one. Define what a "commitment" is, set the threshold, put the District on a build timeline, and choose any model above. The providers are ready to sign; the customers are ready to return. The only thing missing is the path the District never drew.
One price for everyone — the quiet decision that caps the upside.
For the upgraded network the District has proposed a single wholesale plan: $40/mo for 250/50 Mbps. The legacy network it replaces carried three to four tiers ($17–$39). To be clear about what we're not arguing: the $40 base is a fair price — about 25% cheaper in real terms than the District's 2009 rate, for vastly more speed — and on cost-based math, going below it is genuinely hard. The problem is that the ladder stops there: hardware built for 500 Mbps with nothing above the base plan to sell — which works directly against the District's own stated worry about recovering the investment.
Tiering is the norm — everywhere, including here. Essentially every major ISP sells a ladder of plans: an entry tier around $30, a mid tier, and a premium gigabit tier at $100–$180 — Spectrum, AT&T, and Xfinity all follow that pattern. The District's own competitor, Starlink, tiers too (residential $55/$85/$130; business $250–$1,500). And the District itself offered tiers on this network for two decades. Selling one plan isn't the simple, safe choice — it's the unusual one.
The economics has a name. A product ladder is what economists call versioning — second-degree price discrimination (Shapiro & Varian, Information Rules). The entire point is that a line of plans lets each customer pick the version that fits — and move up it as their needs grow. It captures more revenue from customers who'll pay for speed, gives providers something to upsell as a household adds cameras, a home office, a VoIP line — and lets an operator manage congestion by steering light users to lighter plans. A single price forfeits all three — it's precisely the structure that leaves money on the table at the top.
It skews recovery
The hardware "could comfortably offer 500 Mbps" (the District's own words), but with no premium tier there's nothing to sell the businesses and heavy users who'd pay for it — pure foregone revenue. And the District's own records prove customers climb when there's a rung: on the legacy network the 20 Mbps top tier grew 635 → 1,033 even as the total base shrank. Capping the ladder caps the very revenue that repays the upgrade — the lever the dashed lines on the recovery chart illustrate.
It handicaps the RSPs
The District sells one wholesale product, so its retail providers can't build a real lineup — no premium plan for the ranch running cameras and VoIP, no business tier for the shop that would pay for symmetry and support, less room to bundle upward with phone or managed IT. The providers are the ones who actually compete with Starlink and know their customers; a single tier takes the upsell tools out of their hands.
And what about below $40? We're not asking for that. On the District's cost-based math a sub-$40 tier is genuinely hard to build — a radio costs what it costs to hang, and the law says rates recover cost. The affordability answer is already in the proposal: $40 buys 250 Mbps — about a quarter cheaper in real terms than what the 2009 wholesale dollar bought, for more than ten times the top legacy speed. Retail affordability from there is the providers' craft — bundles, promotions, annual plans — and if a genuine hardship gap remains, that's a question for the provider workshop, not a reason to cap the top of the ladder.
None of this means the District should be setting tiers itself — just the opposite. Pricing is a retail decision, and the District doesn't do retail. The providers who do — who see willingness-to-pay customer by customer — should be free to build the ladder. Handing them a single wholesale price is the same top-down move as the rest of this decision: a market judgment made over the heads of the people who actually serve the market.
The evidence: customers climb ladders — when there's a rung to climb to
No study directly pits a tiered menu against one flat rate (none exists). But the industry's universal practice, the take-rate record of well-priced public networks, and — strongest of all — the District's own five-year billing record all point the same way:
| What the research shows | The figure | Source |
|---|---|---|
| This network's own customers climbed for five straight years | The 20 Mbps top tier grew 635 → 1,033 while the total base shrank — demand ran out of ladder, not interest | the District's monthly billing reports, 2021–2026 |
| The customers who stayed pay more, voluntarily | Wholesale revenue per subscriber rose ~8% ($28.90 → $31.30/mo) with zero price increases — pure up-tier mix | the District's monthly billing reports |
| A premium tier is mainstream, not niche | ~1 in 3 households buy the top (gigabit) tier when it's offered | OpenVault, 2025 |
| Tiering is the universal norm; one rate is the outlier | Spectrum, Xfinity, AT&T — even Starlink — all sell a 3-tier ladder | provider rate cards, 2026 |
| Networks priced like the $40 base fill up | At ~$40–50, public networks hit 60–70% take (NextLight ~66%, EPB ~70%) — the base rate isn't the obstacle | ILSR / EPB, 2015–2025 |
What the single rate costs — paid back faster, and more, forfeited
If premium and business plans above the $40 base lift the blended wholesale rate even modestly, the upgrade pays back faster and earns far more over its life — at a 45% cash margin, $1,210 all-in per customer, 878 customers, 10-year gear life:
| Blended wholesale rate | Simple payback* | vs. flat $40 | Extra lifetime revenue |
|---|---|---|---|
| $40 — the single tier | ~5.6 yr | — | — |
| $45 blended (modest premium mix) | ~5.0 yr | 11% faster | ~+$0.5M |
| $48 blended (strong premium mix) | ~4.7 yr | 17% faster | ~+$0.8M |
*Undiscounted simple payback, an illustration of the relative gain from tiering at a fixed 45% margin; the fully discounted, risk-modeled median recoup is ~9–10 years, within the gear's ~10–11-year life.
The reasons given deserve a closer public look.
"Subscription rates have steadily declined."Offered as the reason not to reinvest.
Stable for years — and the recent decline is attrition from a starved network, hit by separate one-time shocks.The base held steady from 2021 through 2023, then began falling in 2024 — the same year net new telecom investment dropped to $159,187 and the Cambium gear hit end-of-life. But the headline "~30 a month" is misleading: more than half of it is a single reseller (NCI Datacom), whose RSP subscriber numbers fell sharply after its tie-up with the out-of-area-owned Core Fiber rollup — concentrated in one reseller, not a platform-wide verdict. (The District's data shows those RSP accounts fell — it can't tell us why, or where those customers went.) Every other reseller declined far more gently. Beneath that runs a quieter factor with nothing to do with the market: in a service area whose deaths have run ~20% above the pre-COVID baseline since 2021 — and whose rural districts are the county's oldest — a real share of "lost customers" simply passed away (the demographics). Where customers do leave for Starlink (~15% of cancellations), it's because a frozen, 20 Mbps network can no longer deliver reliable service — not because the public model failed. One reseller's restructuring, record mortality, and a price promo landed in the same short window — on a network running gear six to nine years past its expected service life (a 2010 federal build against the government's 7-year planning standard). That is a real trend with a named, fixable cause — disinvestment — not a market verdict, and the platform held through all of it. The decline is the argument for reinvesting, not against it. See the breakdown →
"A highly competitive marketplace."Implying customers have plenty of alternatives.
It's competitive because the District makes it so.The competition the letter points to — 7+ local providers, plus NCI on its own towers — runs on the District's open-access network. Step back and the "marketplace" narrows toward one private operator answering to out-of-area investors (NCI, under the Core Fiber rollup) plus satellite — billed out of the county, priced by how few choices a customer has. And the network has held its ground: in the providers' own cancellation logs Starlink is the biggest single competitor but still a minority — about 15% — with most customers leaving for moves, nonpayment, or seasonal vacancy, and some simply switching to another local provider on the same network. The platform weathered Starlink's cheapest-ever pricing intact. A "highly competitive marketplace" that depends on the District isn't a reason to step back — it's the reason to stay. See the breakdown →
"The District does not compete on price."Because rates are cost-based, not market-based.
The District doesn't need to — its providers do.This misreads the model. The District sells wholesale at a fixed rate; the retail providers set the customer price and compete on it. Several have already committed to match Starlink's pricing, even at thinner margins, to keep customers — and because the District's wholesale rate doesn't change, it earns the same either way, at no cost to itself. (State law, RCW 54.16.330, requires the District's wholesale rate to be nondiscriminatory and the telecom line to be self-supporting — and as a non-profit it prices at cost, not for profit; the retail price is the providers' to set. A modernized local tier is estimated near $50/mo for 250 Mbps — at or below Starlink's cheapest plan, with more than double the speed.) The District competing on price was never the question; keeping the wholesale network alive so its providers can is.
Upgrade only "should future subscriber enrollment reach a level sufficient to offset the costs."
A bounded, one-time cost — set against a recurring revenue line.The District has stated the upgrade would cost about $1.2 million, consistent with its own workbooks (≈$788,693–$1,064,403 of equipment, plus the usual labor, tax, install, and contingency). We accept that figure — and a full deployment could run somewhat higher. It's a one-time outlay, about 1.5 years of the segment's $808,166 annual surplus, to defend a recurring ~$1.1M/yr revenue line and a federally-funded public asset. "Wait for subscribers to return to an un-upgraded, end-of-life network first" is the chicken-and-egg trap. And from a risk standpoint, "wait and see" is the expensive option →
"The District will continue operating its existing wireless system."Offered as a stable status quo.
On hardware the District itself calls maxed out.The network runs on Cambium PMP 450-series equipment. The District's own memo says its 450m access points are “no longer keeping up with demand” and that sites are “severely congested with no possibility for expansion.” The radios can still be bought today, but support for the 450 line ends December 31, 2027, their maker was delisted from Nasdaq in 2026 under a going-concern warning, and the 2010 build is already six to nine years past its expected service life. The District's own engineers scoped a full Tarana 6 GHz replacement — then chose not to fund it. “No further investment” isn't a steady state; it's a countdown with a published date on it.
The decline, overlaid: investment fell first, customers followed
The clearest test of "did the market take these customers, or did under-investment?" is the timing. Put the two lines on one axis:
Left axis (rust): budgeted wireless reinvestment, by year. Right axis (navy, zero-based — no exaggeration): active subscribers, monthly. Reinvestment was cut 34% by 2023, while the subscriber base sat at its all-time peak; customers didn't begin leaving until mid-2024 — about 18 months later. Starlink arrived in 2021, yet the base grew for three years after. The District's own spending cut set the slide up first — Starlink's promo and NCI's RSP decline triggered the cliff that followed — and the deepest cut came while the network was healthiest, so it was a choice, not a reaction to lost demand. More on the timing →
No one asked the people who actually know the market.
The letter blames "additional providers offering wireless services equivalent to, or exceeding" the District's own. But the retail service providers who sell on this network — the people who track local demand, churn, and margins customer by customer — say they were not consulted before this decision. And no market analysis appears in the records released to us: across 122 board-meeting records and the budget and engineering files we obtained, there is no demand study, competitive analysis, or customer survey — and if one exists, the District should put it on the record. The "competitive marketplace" claim rests on a comparison to Starlink's advertised prices, not on the businesses that actually serve Okanogan County. A market this important deserves more than a price check.
It was never only about speed.
The District is treating this as a spreadsheet of megabits and subscriber counts. That misunderstands what a broadband utility actually provides — and what customers actually value: a connection that simply works, from someone local. Most rural homes need only ~10 Mbps; what makes them stay or leave is reliability, not peak speed.
Internet providers rank near the bottom of every U.S. industry for customer satisfaction — second-worst of 46 industries in the American Customer Satisfaction Index — and the single biggest complaint is the call center: the faceless, take-it-or-leave-it support of a distant company you can't fire.
The PUD's open-access model is the antidote to exactly that. Because the District sells wholesale capacity to competing local providers, a customer here can call a real person in this county, switch providers if they're unhappy, and keep their dollars circulating in the local economy. That — local ownership, competition, accountability — is the product. Not a megabit number.
Plenty of customers don't need gigabit: the network tops out at just 20 Mbps, and many are content on modest tiers — more than 40% on the 7 Mbps plan or slower — because what they value is a connection that works, from someone local. But "modest" is not the same as "frozen." Stuck on end-of-life hardware near 20 Mbps while satellite keeps improving, the network has crossed from good enough to falling behind.
And here's the honest part: Starlink is the biggest single competitor customers name — a minority today (about 15%), but one that gains ground the longer the network stays frozen. That isn't proof the local model failed; it's proof of what happens when a local network is starved instead of modernized. Give those same customers a fast, fairly-priced connection with a local provider to call, and the relationship wins — that's exactly what a funded Tarana upgrade delivers, and exactly what the District voted not to fund.
And remember the scale of who this is: roughly three in five of the county's residents (~27,000 of ~43,000) live in the unincorporated rural areas this network serves — the majority of Okanogan County, and the towns where fiber concentrates aren't growing. Agriculture is the county's largest industry — the tree-fruit sector alone tops $250 million a year and employs roughly a third of residents — and it runs on reliable connectivity for logistics, sales, and day-to-day farm management. Abandoning the wireless platform doesn't just cost speed. It removes the one structure that keeps rural broadband here accountable to the people who use it — and hands most of the county back to distant providers with no local presence and no one to answer the phone.
This market has been competitive for years. The pause is what forfeits it.
Part of the rationale for stepping back was competition — but this was never a monopoly, and the District never lacked rivals. NCI Datacom owns its own towers and has competed here for years. On Eder, the District's biggest site, NCI advertises at least 600/50 Mbps on its own equipment (reseller-reported) while the District sold just 20 Mbps — and the District's subscriber base held at peak levels through late 2023 anyway. That's strong evidence that what holds customers here is service as much as speed.
NCI is the one other operator that owns its towers, which makes it both a reseller on the District's open-access network and a direct competitor on its own. It sits head-to-head with the District on three southern ridgetops and shares or competes on the rest — and the District leases every tower site it uses — all nine, including the seven in this upgrade — owning none of them. There is no fortress here. There is a public network competing in a real market, and winning on service. See the full landscape, mapped from the records →
That competition isn't the problem — it's the point. Washington created its PUDs in 1930 precisely to stabilize markets the private sector underserves, and for two decades the District has done exactly that here: open access let competing local providers keep each other — and NCI — honest on price and service. Freeze the District on end-of-life hardware and it drops out of that competition. What's left is one out-of-area, investor-owned operator (NCI/Core Fiber) as the dominant player, with no public alternative to keep rates in check.
So the decision has the risk backwards. The pause doesn't protect against a monopoly — it risks creating one. Modernizing keeps the District in a competition that has served ratepayers for twenty years; walking away hands the county's broadband market to a single out-of-area operator answerable to investors, not ratepayers. The point isn't that NCI is a villain, or that competition is bad — competition here has been healthy, and the District should want to keep it. It's that the one institution able to keep that market open is the one choosing to step out of it. And the competition isn't what it was — see then vs. now →
Starlink is a backstop — not a replacement for a network we own.
Before the point-by-point comparison, the thing the "just use Starlink" argument misses: satellite is the right answer for a cabin at the end of a dirt road, and a fine backup anywhere. But sold as a county-wide substitute for a network the public owns, it trades local control for dependence on one distant company.
It all hinges on one company — and one rocket
Starlink isn't infrastructure the county owns; it's a subscription to a fleet that must be relaunched forever. Each satellite drops out of orbit in about five years, so the service exists only as long as one company keeps launching — on its schedule, its priorities, its finances. Even the promised "gigabit" rides on a next-generation satellite that flies only on Starship, still in testing. A county that leans on Starlink is betting its connectivity on one company keeping up, indefinitely — with no local recourse if it doesn't.
One company holds the price, rules, and off-switch
With a local network you elect the board, attend the meeting, read the public records, and call a neighbor when it breaks. With Starlink all of that lives at one headquarters: the price has changed repeatedly and is set with no local rate board; the contract says the service is "as is" and "not… a mission-critical or safety-of-life service," with no guaranteed uptime; and the same central control can disable a terminal remotely. On July 24, 2025, a single software failure took ~6 million users across 140 countries offline at once — the kind of all-at-once outage a distributed local network can't have.
Every dollar leaves the county
Starlink's monthly fee plus $349 of hardware goes to an out-of-state company — and the fast Starlink (gigabit, an SLA, a static IP, priority) is the business product at $250+/mo; what a rural family can afford is the basic residential tier. The local retail providers keep the dollars, work, and accountability in the county at cost-based prices — several ready to match Starlink's speed at or below its price — so the local option is the one the lowest-income households can actually afford.
A backstop, not a foundation
Satellite is genuinely the best answer for the truly remote, and a smart backup for everyone. But a resilient county is local-first, with satellite as the backup — not dependent on orbit. Trading a public network you already built and paid down for reliance on one distant company doesn't add resilience; it removes the choice, the local control, and the price-check the PUD exists to provide.
One more, honestly caveated: in 2022 the FCC pulled an $885 million award rather than bet Starlink could deliver at scale (upheld 2023). We don't lean on it — federal policy has since gone technology-neutral and the hardware has improved — but it's a fair marker of how cautiously the government once treated satellite as a primary rural solution.
None of this means Starlink is bad — it's a genuine marvel, and for the truly unreachable it's the best option there is. It means satellite and a locally-owned network are different things. Treating "good enough from orbit" as a reason to let a buildable, fiber-backed public network decay is how a county trades an asset it controls for a service it doesn't. The resilient answer was never local or satellite — it's a local network with satellite as the backup, and pausing removes the primary that backup supports. With that in mind, the point-by-point:
Why don't we all just use Starlink?
It's a fair question — and for a few addresses, satellite is the right answer. But as a county-wide substitute for a public network, it's more expensive, higher-latency, less accountable, and it drains money out of the local economy.
Cost to a household
$55/mo for 100 Mbps, $85/mo for 200 Mbps, $130/mo for up to 400 Mbps. Hardware $349 (or ~$10/mo rental). $660–$1,560/yr leaves the county. (residential rates, June 2026)
An estimated ~$50/mo for a 250 Mbps tier (the District's proposed plan) — at or below Starlink's cheapest price, with more than double the speed and no priority tiers — the same service for everyone. Revenue recirculates locally.
Cost to a farm or business
$250/mo (1 TB) to $1,500/mo (6 TB); High-Performance Kit $1,999–$2,500 up front. $3,000–$18,000+ every year, out of county.
Business-class service at cost-based rates, with a local technician and a local retail provider who answers the phone.
Speed & latency
Median 134 Mbps down / 19 up, ~45 ms latency to a satellite (Ookla, late 2025). A median, not a guarantee.
Engineered to match or beat those speeds at 5–20 ms to a local tower — better for video calls, telehealth, gaming, and VPNs.
Reliability in our climate
Tree obstruction is the #1 documented complaint — outages when obstructions exceed ~5% of sky view. 50–100 W power, up to 165 W in snow-melt mode.
Customer equipment draws under 15 W. No dish to keep clear of snow, no clear-sky requirement.
Local support
Support runs through an app; user-reported response times run from hours to days. No local presence.
A local installer and a phone number answered by a person in this county, through the retail providers.
Who decides the future
One private company. Pricing has changed repeatedly in three years, and the terms are set unilaterally — no local rate board, no public hearing, no appeal.
Elected commissioners, open meetings, public records. When something needs to change, you can show up at a Tuesday-night meeting and say so.
Every Starlink subscription sends money out of state. Every local subscription keeps it here — funding local jobs, local infrastructure, and the PUD itself.
The PUD can stabilize prices here — even against Starlink.
This was never just a balance-sheet decision. The May 4 letter weighs one thing: whether the wireless line pays for itself fast enough. But a public utility holds responsibilities a balance sheet can't show — and on three of them, stepping back costs the county far more than the upgrade ever would. These aren't sentiment. They're the reasons public utility districts exist in the first place.
A price check — even on Starlink
Washington created its public utility districts in 1930, in a Grange-led ballot measure, because private power companies were charging rural families more than double the urban rate — and skipping the countryside altogether. A PUD's founding purpose is to stabilize a market the private sector under-serves — and that is exactly what this District has done with broadband for two decades: kept a competitive, locally-priced network on the ground so no single provider could squeeze the county. That job hasn't changed; the only thing that has is the equipment, left to fall behind while the market it was built to stabilize moved on. Broadband is the same story: where the local network fades, the only option left is a satellite priced and controlled from out of state — no local rate board, free to raise prices at will. If it doubled or tripled its price tomorrow, a county with a working local network has an answer; one without it simply pays. The District can't set Starlink's price, but a public, at-cost option is what keeps the squeeze in check. Let the local network decay, and the county hands a distant monopoly exactly the leverage the PUD was created to prevent.
A life-safety service — in fire country
Across much of Okanogan County — the largest county in the state, about eight people per square mile — there's no reliable cell signal and landlines are mostly gone, so home internet is the phone: Wi-Fi calling, VoIP, and 911. And this is wildfire country. The 2014 Carlton Complex was the largest single wildfire in Washington history; in the 2020 Cold Springs Fire a cell tower burned to the ground and 911 was rerouted out of county for about twelve hours. No one knows this better than the District — after the 2014–15 fires its own crews reported radio, phone, cell, and fiber "all down," with sixty miles of its fiber and wireless lost. No technology is fireproof — but a locally owned, locally maintained network is a layer of resilience this county has needed before, in the one place it's been tested. And where the county's hospitals are small critical-access facilities that mostly stabilize and transfer, that connection is also the link to care — the region's main health system, Confluence Health, launched 24/7 at-home virtual urgent care in October 2024. And when disaster strikes, local crews restore what they own first — the PUD's own line crews reach these towers; a distant satellite operator has no one here. (Any upgrade should harden tower backup power, so the network stays up when the grid goes down.)
An ecosystem — not a product line
This network isn't one company's product; it's a marketplace the District built. Because Washington law for two decades allowed PUDs to sell broadband only at wholesale (retail authority came only in 2021), the District created an open-access network that eight independent local providers compete on — public infrastructure, private competition. It's the model Washington is known for nationally: open-access markets here average about seven providers, and Okanogan has eight — more real choice than most American cities. Seven of them resell on these wireless towers, with no other way to reach those customers. "Pausing investment" isn't trimming a product line — it's pulling the floor out from under a local industry the District spent twenty-five years building. A utility doesn't only serve customers; it stewards the market it created. NoaNet — the open-access network Okanogan helped found in 2000 — still offers "wireless colocation … to allow wireless Internet Service Providers to expand their network's reach without a high capital requirement." The District's own founding partner champions the very model it is now stepping back from.
The letter asks whether the wireless line clears a financial bar. The deeper question — the one only the Board can answer — is whether Okanogan County PUD still sees itself as the steward of a market, a safety net, and a quarter-century public legacy, or as the operator of a line item it can switch off. The first is the reason public utility districts exist at all.
Sources: Washington's PUDs were established by Initiative No. 1 (1930). Starlink's published residential rates and repeated price changes, from Starlink's own materials. Okanogan County Broadband & Digital Equity Plan (2023); HistoryLink; Methow Valley News (2020); Okanogan PUD / NWPPA wildfire case study (2016); FCC guidance that VoIP and Wi-Fi calling cannot reach 911 without power and a working internet connection. Open-access provider counts: NoaNet; Washington PUD Association. Wholesale/open-access authority: RCW 54.16.330.
A legacy, not a line item.
Before the spreadsheet, the history — because it changes what "risk" means here. What this District built is, frankly, a rural marvel: more competition and local choice than most American cities have, on infrastructure other rural counties wish they had. It exists because earlier leaders took real financial risk when rural broadband was anything but a sure thing.
In February 2000, Okanogan was one of about ten Washington utility districts that founded NoaNet — now the state's largest open-access network, 3,800+ miles of fiber — built precisely because private ISPs served the I-5 corridor and skipped the rural east. The District then built its own backbone and launched fixed wireless in 2002.
This wasn't all grants. For the 2010 build the District took on a ~$3.7M federal loan alongside a $5.5M grant — debt it took on and has steadily paid down — and it reinvests its own telecom earnings into the network. Earlier boards bet real money on an unproven model, and the bet paid off.
Eight competing local internet providers serving ~3,800 customers — more real choice than most U.S. cities, where one or two dominate. Cost-based prices, local accountability, a stable market, a $16.9M public asset — and, in the District's own words, "one of the leading broadband solutions for rural communities in Washington State."
So here's the risk the letter never names. The risky, expensive, pioneering work is done — and largely paid down. What remains is a routine ~$1.2M equipment refresh: a fraction of the original bets, on infrastructure that already exists, already earns ~$1.1M a year, and is already de-risked. Weighed against that, the larger danger is letting a 25-year, federally-funded public asset decay on end-of-life hardware — forfeiting the choice, competition, and market stability it took two decades and real financial risk to build. Stepping back doesn't avoid risk; it takes the one risk you can't undo: losing the legacy.
Today's commissioners weren't the ones who founded NoaNet in 2000 or built the network in 2010 — but they inherited the reward, and they're now the stewards of whether that legacy continues. That isn't a spreadsheet decision; it's a decision about whether to protect what this county built.
Decades of public investment — and a pivotal decision.
A fiber backbone reaches the border
The District built an 82-mile fiber backbone from Pateros to the Canadian border at Oroville, and began offering wholesale broadband (wireless from 2002) to retail providers serving homes and businesses.
Federal money builds the wireless network
Through the USDA Broadband Initiatives Program (Recovery Act), the District won a $5,501,782 grant and a $3,667,855 loan — roughly $9.2 million in federal funds — to add wireless access points along the fiber route, building the ~170-site network that still serves rural customers today.
A $30.2M federal grant — for fiber
The District is awarded a $30,195,500 USDA ReConnect grant (a 100% grant, no loan) to build fiber to about 574 funded households (the PUD cites ~750 residences) and 21 businesses across 301.7 sq mi of the northeastern highlands — Chesaw, Molson, Havillah, Nine Mile, Siwash Creek.
A year on, the fiber grant has barely moved
Awarded back in August 2023, the ReConnect grant wasn't put under a signed agreement until Aug 14, 2024 — and per the District's FY2024 audit, $0 had been spent by year-end, the District's own "spring 2024" construction start already missed. Net new wholesale-telecom plant for the whole year was just $159,187 on a $16.9 million base — effectively flat.
The wireless investment pause
Staff present a letter to the Board telling retail providers there are "currently no plans to make additional investments in wireless infrastructure improvements." In the minutes the letter appears under the Manager's Report — no motion, no vote, no resolution; that day's minutes record "No new business" and "Public: None." No supporting market or financial analysis appears anywhere in the records. The General Manager reported signing the letter at the May 18 meeting.
The $30M fiber grant is rescinded
The District moves to rescind the ReConnect grant — Resolution 1863, "Rescind RUS ReConnect Award WA 1707-A73," on the June 22, 2026 board agenda. The funds were never deployed in time and rising costs left the budget short, so the federal money for Molson, Chesaw, Havillah, Nine Mile, and Siwash Creek goes back unspent. The plan to bring fiber to the dispersed rural northeast is abandoned — while the wireless that already serves those same areas is the platform now being cut. (Per the posted June 22, 2026 board agenda; pending the final vote and signed minutes.)
Every public meeting for 3½ years — and where the wireless network appears.
One square per board meeting record, January 2023 through June 2026 — 86 meetings, drawn from the District's own published minutes. A filled square is a meeting whose minutes contain a substantive wireless-investment item. Count them. Before spring 2026 it happened twice — both rate housekeeping. The decision to stop investing appears at square 4 — and it was never voted on.
Dashed squares are special meetings and budget workshops — including five budget workshops where the telecom budget was adopted with no recorded discussion of wireless investment. The monthly subscriber-count boilerplate in staff reports doesn't count as an investment discussion; if it did, the case would be worse — the Board watched the numbers monthly while the minutes record no debate about the line's future. The word “Tarana” — the upgrade the District priced, bought, and surplused — appears in these minutes zero times. Source: Okanogan PUD published board minutes, Jan 2023–Jun 2026 (86 records; obtained and indexed by this project, July 2026).
The same story, told by the documents themselves.
Three exhibits, in chronological order — the budget that starved the network, the meeting record where the decision surfaced without a vote, and the auction listing for the upgrade equipment. No commentary required.
“The District will continue to operate its existing wireless system; however, there are currently no plans to make additional investments in wireless infrastructure improvements. Subscription levels will be monitored for future consideration.”
“Tarana 6 GHz Wireless Equipment — New in Box”
Sources: the District's 2026 budget workbook (records request); published board minutes, May 4 and May 18, 2026; Public Surplus listing (Okanogan County PUD). Full document library: Documents & Sources →
Provide dependable public utility services at a reasonable cost, governed by, and for the benefit of our customers in a safe and environmentally responsible manner.Okanogan PUD mission statement
Under RCW 54.16.330, PUDs are authorized to operate and improve wholesale telecommunications on a nondiscriminatory, open-access basis (and, since a 2021 law, to offer limited retail service in unserved areas). Washington's statutory broadband goals (RCW 43.330.536) call for 25/3 Mbps to all by 2024, 1 Gbps to anchor institutions by 2026, and 150/150 Mbps to all businesses and residences by 2028. The wireless platform is how that public mandate reaches the addresses fiber doesn't — and it isn't “public” only in the abstract: it was built with roughly $9.2 million in federal grant and loan funds.
Back up the case — and act on it.
Four arguments, answered
The claims behind the pause — NCI, Starlink, combined competition, decline — steelmanned, tested against the records, and graded. Printable, for the Board & GM.
Open the brief →Charts & methodology
Every figure, traced to the PUD's own records.
Explore the data →Who the network serves
The lower-income rural majority who rely on the affordable wireless option.
Open the map →Take action
Add your voice and send the letter to the Board.
Take action →