Blue Prysm · free tool
What does spectrum cost?
Every US spectrum benchmark we can compute from public data, in one comparable unit, dollars per megahertz per person covered, with the working shown for each one.
Free · no sign-up · result never gated
Estimate
This is a comparables range, not a valuation. It tells you what other buyers paid for roughly comparable spectrum, per person covered. What your spectrum is worth depends on encumbrances, build-out obligations, licence term, adjacency and who else wants it, none of which is in this data.
The finding most people miss
CBRS priority licences cleared at $0.198 per MHz-POP nationally. The median county paid $0.032, and only 16% of counties paid above the national average at all.
Auction 105 sold 10 MHz priority licences county by county, which makes it the one modern auction where the whole price distribution is visible. It is extremely skewed: 22 counties out of 3,123 account for half of the $4.5B raised. Quoting the national average at a rural operator overstates their cost by roughly six times; quoting it in Chicago understates it by four.
$0.032
Median county
$0.198
National weighted average
16%
Counties above that average
22
Counties = half the money
| What a large county paid, Auction 105 | Population | $/MHz-POP |
|---|---|---|
| Cook County, Illinois | 5,182,617 | 0.769 |
| New York County, New York | 1,660,664 | 0.599 |
| Los Angeles County, California | 9,757,179 | 0.534 |
| Harris County, Texas | 5,009,302 | 0.382 |
| Maricopa County, Arizona | 4,673,096 | 0.382 |
| Miami-Dade County, Florida | 2,838,461 | 0.309 |
| King County, Washington | 2,340,211 | 0.180 |
| Loving County, Texas | 48 | 241.667 |
Every benchmark and its basis
| Transaction | Year | Total | MHz | $/MHz-POP | Basis |
|---|---|---|---|---|---|
| Auction 113: AWS-3 | 2026 | 3.57B | n/a | 2.658 | per licence |
| EchoStar → AT&T: 600 MHz + 3.45 GHz | 2026 | 20.25B | ~50 | 1.191 | national-equiv |
| EchoStar → SpaceX: AWS-4, H, AWS-3 | 2025 | 19.62B | 65 | 0.887 | national-equiv |
| Auction 107: C-band | 2021 | 81.17B | 280 | 0.852 | national-equiv |
| Auction 110: 3.45 GHz | 2022 | 22.51B | 100 | 0.662 | national-equiv |
| Auction 105: CBRS PAL (weighted) | 2020 | 4.59B | 10/licence | 0.198 | per licence |
| Auction 105: CBRS PAL (median county) | 2020 | n/a | 10/licence | 0.032 | per licence |
| Auction 108: 2.5 GHz overlay | 2022 | 0.43B | 17.5–49.5 | 0.016 | per licence ⚠ |
⚠ Auction 108 sold overlay licences covering only the unassigned whitespace inside each county, so its denominator is county population while its actual coverage is far smaller. The figure is real arithmetic on real data and still not comparable with the rest, which is why it is on the list.
Method and limits
Where the numbers come from. Gross winning bids from the FCC’s published auction results files, retrieved 2026-09-17. County populations are the US Census 2024 vintage estimates. Market populations for CMA and BEA licences come from the FCC’s own market table in the ULS bulk licence files. Private transaction values come from EchoStar’s Forms 8-K, not from press coverage.
Two ways of computing, and they are not the same. Where a licence’s market population is available, we compute price per licence and aggregate: the “per licence” basis. Where it is not, we divide the total by the bandwidth and by the whole US population: the “national-equivalent” basis, which assumes every licence sold and every person was covered.
What we cannot compute, and why. The C-band and 3.45 GHz auctions were licensed by Partial Economic Area. The FCC’s market table leaves PEA populations empty, and the reference file that would fill the gap is not machine-fetchable. So those two appear only as national-equivalent figures.
Low band. The 700 MHz (Auction 73) and 600 MHz (Auction 1002) result files return 404 on the FCC’s current path, so low band appears here only inside the blended EchoStar–AT&T figure and the estimator returns no separate low-band range.
Not a valuation, and not investment advice. A comparables range built from public transactions. It does not account for encumbrances, build-out obligations, remaining licence term, adjacency, incumbent clearing costs or competitive dynamics.
Every figure is reproducible from the sources named. Show us one that is wrong and we will change it and say what changed.