Seventeen years of carrier-owned DAS shows up exactly where it should: indoors. AT&T holds the strongest indoor 5G, Verizon delivers the cleanest signal quality, and T-Mobile — with no in-building path — is the carrier the neutral-host cutover most helps. No gate is critical for any carrier: 0 of 309 gate-carrier pairs measure below −110 dBm.

Strongest indoor signal; the DAS incumbent. Healthy 5G RSRP at 95 of 103 gates, venue median −90.5 dBm, best 5G NR at 72 of 103 gates with all-carrier data. The current measurement is the AT&T baseline the incoming neutral-host system must hold at parity.

Cleanest quality; the LTE network of record. Median SINR +12.5 dB (103 of 103 gates healthy — the most of any carrier), best-serving LTE at 76 of 103 gates, healthy LTE median −90.0 dBm, and healthy 5G at 87 of 103 gates. New this window: best-NR honors across all of Concourse B and indoor-stronger tenant medians with no recorded in-building system to explain them (section 8).

Weakest indoors; the largest neutral-host lift. Healthy 5G at only 17 of 103 gates, venue median −97.9 dBm, and the worst carrier at 93 of 103 carrier-asymmetry gates. The pattern is consistent with no in-building path: T-Mobile rides macro spillover where AT&T rides the DAS. A hosted T-Mobile path is the single largest measurable lift the Boldyn system can deliver.

SEA is operated by the Port of Seattle (~25.4M annual enplanements; CIO Matt Breed Imagine knowledge graph). The measured facts in this report sit on top of one structural fact: the in-building system is changing hands.

Carriers are ordered by descending venue 5G median throughout this report. The table summarizes the gate suite; the figures show the distributions behind the medians.

Verizon's LTE median is −90.0 dBm. Dashes: the report publishes only the winning carrier's count per best-network label. Full per-gate tables ship in the PDF appendices.

Fig. 1 · Venue-wide carrier medians · 5G RSRPSource · Ookla

Locked ladder: healthy ≥ −95 · at-risk −95 to −110 · critical < −110 dBm. Window 2025-08-01 to 2026-07-31.

Fig. 2 · SINR distribution per carrierSource · Ookla

Verizon's quality lead is venue-wide, not localized. SINR healthy threshold ≥ 10 dB.

Concourse-level healthy shares and medians locate each carrier's gaps. The at-risk band dominates T-Mobile across most concourses; AT&T holds majority-healthy distributions in the Alaska hub. Critical-band gates (< −110 dBm): zero, for every carrier.

Fig. 3 · Healthy-gate share by concourseSource · Ookla

Fig. 4 · Median 5G RSRP by concourseSource · Ookla

Fig. 5 · Coverage-class distribution by concourseSource · Ookla

A = AT&T, V = Verizon, T = T-Mobile. n per concourse on the axis. Window 2025-08-01 to 2026-07-31.

This is the split that decides who pays. A gate weak for all three carriers is a building, geometry, or propagation problem — the neutral-host system's job. A gate weak for exactly one carrier is that carrier's path problem.

At SEA the dominant pattern is one-or-two-carriers-weak with AT&T healthy — a carrier-path problem concentrated in T-Mobile (and situationally Verizon), which is precisely the gap a hosted neutral-host path closes. Zones showing all-three-weak gates price as propagation problems and belong in the Boldyn design review.

Fig. 6 · Carriers below the healthy line per gate, by concourseSource · Ookla

New section this revision (VNIR §4.3). Window 2025-08-01 to 2026-07-31.

The Alaska hub is SEA's dominant single-airline cluster (38 gates; C expanded June 2026, N is Alaska-exclusive). AT&T is best-NR at 38 of 38 hub gates with NR data and healthy at 32 of 38. T-Mobile and Verizon coverage in C and N should be named acceptance criteria in the Boldyn test plan — those are the carriers riding macro spillover in the hub today.

Fig. 7 · Per-gate 5G RSRP · 38 Alaska-hub gates (C + N)Source · Ookla

The South Satellite hosts international premium-cabin traffic — the passengers with the highest expectations and the least tolerance for a dead zone at the gate. Best-NR split across its 14 gates: AT&T 12 · Verizon 2. Carrier coverage in S belongs in the post-cutover validation set.

Fig. 8 · Per-gate 5G RSRP · 14 South Satellite gatesSource · Ookla

Foreign-flag carriers plus Delta international. Window 2025-08-01 to 2026-07-31.

A large positive outdoor-minus-indoor delta is the macro-spillover signature; a flat or negative delta indicates an in-building source. The three carriers split exactly along the deployment story:

Fig. 9 · Median RSRP · indoor vs outdoor tenant categoriesSource · Ookla

PROXY: category-based masking, not Ookla's native building-polygon indoor mask (queued as a next layer). Window 2025-08-01 to 2026-07-31.

Verizon's rev-2 profile — best-NR at 17 of 17 Concourse B gates, indoor-stronger tenant medians (−6.0 dB), the cleanest SINR in the venue — has no recorded in-building system to explain it. The airport graph attests an AT&T DAS; it records nothing for Verizon.

This report deliberately does not explain the anomaly from tile data. It converts it into an attestation question: locate, attest, and document whatever carries Verizon indoors — before the Boldyn cutover decommissions legacy paths unexamined. An unrecorded independent path decommissioned by accident is the largest transition risk in the record.

The percent-low-band statistic is an indirect indicator of in-building dependence: a carrier whose tiles disproportionately attach to sub-1 GHz spectrum (600/700/850 MHz) is leaning on coverage-grade bands because higher-capacity mid-band is not reaching the tile reliably.

Fig. 10 · LTE percent-low-band distribution per carrierSource · Ookla

The Site Finder layer resolves the macro grid each carrier rides where it has no in-building path. Around the aerodrome: Verizon 112 estimated sites (nearest 550 m — the densest grid), T-Mobile 71 (nearest 733 m), AT&T 67 (nearest 414 m). Centroids are Ookla-estimated technology-layer positions derived from crowdsourced RF — estimates, not carrier registry data; LTE and NR layers on one structure appear as two entries.

Fig. 11 · Estimated macro-site positions vs the KSEA aerodrome · site-distance distributionSource · Ookla Site Finder

Distances computed to a disclosed main-terminal reference point. 250 centroids, big-3 technology layers.

Venue-wide, 58.7% of observed data volume moves over Wi-Fi (Σ Wi-Fi MB ÷ Σ all MB, summed across the three carrier sampling panels). Concourse-proximate tiles blend to 58.6%; the remaining on-airport tiles (landside, garages, periphery) run 70.9%.

Fig. 12 · Observed data volume by zoneSource · Ookla

Fig. 13 · Wi-Fi share of observed data volume by concourseSource · Ookla

Per gate, the atlas names the weakest carrier and the spread to the strongest. T-Mobile is the worst carrier at 93 of 103 asymmetry gates. Spread compression at the named gates is one of the acceptance criteria recommended for the Boldyn cutover plan (section 12).

Fig. 14 · Carrier-asymmetry atlas · worst carrier and spread per gateSource · Ookla

Window 2025-08-01 to 2026-07-31. Every published pair n ≥ 30.

1. Write the acceptance criteria into the Boldyn cutover plan now

   Asymmetry-spread compression at the named gates, T-Mobile NR validation in the Alaska hub (C+N), and the no-critical-gates floor. T-Mobile's hosted path is the largest single measurable lift (17% of gates healthy today).

   Owner · Port of Seattle + Boldyn

2. Baseline the AT&T legacy DAS before decommissioning

   As-builts, PCI plan, re-tuning history. The incumbent profile is the parity bar for the new system, and its documentation disappears at cutover.

   Owner · Port of Seattle + AT&T

3. Attest whatever explains Verizon's indoor-stronger profile — before cutover

   Locate, attest, and document the path behind the Concourse-B dominance and indoor-stronger medians (section 8). An unrecorded independent path decommissioned unexamined is the largest transition risk in the record.

   Owner · Port of Seattle + Verizon

4. Re-measure with the identical instrument 90 days post-cutover

   Same window length, same thresholds; Ookla's native date-range comparison renders the before/after directly.

   Owner · Imagine Wireless

5. Treat Wi-Fi as a coexisting capacity layer in the new design

   Venue offload is material (58.7%). Obtain controller telemetry to size it — no Wi-Fi node is even recorded in the graph today.

   Owner · Port of Seattle IT

6. Decide the CBRS / private-wireless question alongside the DAS build

   Not after it — retrofitting operational spectrum into a commissioned neutral-host design costs more than designing it in.

   Owner · Port of Seattle