Internet Service Providers

Most U.S. residents are served by one or two large cable or telecom incumbents — Charter (Spectrum), Comcast (Xfinity), AT&T, Verizon, T-Mobile, Cox, CenturyLink/Brightspeed, Frontier, Spectrum, and Optimum together cover the majority of populated ZIP codes — supplemented by a long tail of regional fiber overbuilders, electric cooperatives, fixed wireless ISPs, and municipal networks. The FCC National Broadband Map is the federal source of record; it ingests semi-annual filings from every facilities-based provider with at least one residential or business connection and standardizes them into a unified, address-level dataset.

Of the 140 providers tracked here, 37 qualify as “national” (1,000+ ZIP codes), 90 are regional (200-999 ZIP codes), and 13 are local or specialty operators. 22 of the providers offer at least one fiber-to-the-home product. Federal funding through the NTIA Broadband, Equity, Access, and Deployment (BEAD) Program is currently financing roughly $42.5 billion in fiber expansion through state subgrants, which will reshape this list materially over the next three to five years.

For independent throughput verification — useful when an ISP's advertised speed and a customer's lived experience diverge — the open-source measurement project M-Lab (Measurement Lab) publishes anonymized speed-test microdata going back over a decade. Real-world median throughput typically runs 60-80% of advertised peak speeds during evening hours.

The U.S. residential broadband market splits into five technology buckets, and the technology mix a provider reports in the FCC BDC dataset is one of the strongest predictors of its grade distribution. Cable (DOCSIS 3.0 and 3.1) is the largest single bucket by population served — Comcast, Charter, Cox, Optimum, and Mediacom together cover most U.S. metros. Cable advertises high peak download (1 Gbps to 2 Gbps with DOCSIS 3.1 and emerging 6 Gbps tiers under DOCSIS 4.0) but is structurally asymmetric, capping upload at 35-50 Mbps in most current deployments.

Fiber-to-the-home is the highest-quality technology — symmetric multi-gig, low latency, decade-plus service life — and is the architecture targeted by the bulk of NTIA BEAD funding. AT&T Fiber, Verizon FiOS, Frontier Fiber, Google Fiber, and a long tail of regional fiber overbuilders (Ziply, Lumos, Brightspeed, Quantum Fiber, Metronet) currently dominate this segment. Legacy DSL (ADSL2+, VDSL2) is rapidly retiring as copper plants are decommissioned; it remains the only option in some rural ZIPs but typically delivers under 25 Mbps and earns a D or F grade. Fixed wireless access (FWA) — both licensed (T-Mobile 5G Home, Verizon 5G Home) and unlicensed (rural WISPs) — has become the fastest-growing access technology since 2022, often delivering 100-300 Mbps download in markets without cable competition. Low-Earth-orbit satellite (Starlink) is now widely deployed in rural and remote areas, with median measured throughput in the 50-200 Mbps range and latency low enough for most real-time applications.

The provider technology mix shapes both the headline grade and the practical experience. A cable-only provider in a single-provider ZIP will earn a B or C even at gigabit download because the competition factor is unmet; a fiber overbuilder entering the same ZIP can pull the area to an A within a single BDC release cycle. The Federal Communications Commission tracks the long-run shift in technology share through its annual Broadband Progress Reports.

A provider's headline grade is the average of the Broadband Grades across every ZIP code in the provider's reported coverage footprint. The ZIP-level grade itself combines four weighted factors: download speed (forty percent), provider competition (thirty percent), fiber availability (twenty percent), and upload speed (ten percent). A provider with a wide but speed-poor footprint such as legacy digital subscriber line plant across rural counties will earn a lower aggregate score than a provider with a narrow but high-quality footprint such as a fiber overbuilder concentrated in a single metropolitan area. Read the full grading methodology for the precise factor definitions, normalization rules, and rollup math.

Our coverage data refreshes whenever the Federal Communications Commission publishes a new Broadband Data Collection release, which currently arrives roughly every six months. On each release we recompute every ZIP-level grade and roll the new averages back up to the provider, county, state, and national leaderboards. Coverage in this dataset is what providers themselves report to the FCC under the BDC mandate — it does not validate that every individual address in a reported ZIP can actually subscribe at the advertised maximum speeds. The FCC's public data download portal accepts location-specific challenges from consumers, local governments, and tribal entities, which feed back into the next release and progressively improve the dataset's ground-truth fidelity.

Provider pages also surface technology mix, advertised maximum download and upload speeds, fiber-zip counts, state-by-state distribution, and the full list of ZIP codes where the provider reports at least one residential service tier. We treat each provider equivalently. A regional fiber cooperative serving thirty counties is graded by the same rubric as a national cable incumbent serving thousands of ZIP codes. The intent is to make small fiber overbuilders directly comparable to the largest national operators on quality dimensions that matter to households, rather than letting scale of footprint dominate the comparison. A small but excellent fiber operator can earn a solid A grade with two hundred ZIPs of coverage; a national cable operator can land a B or C if its footprint averages older plant with limited fiber and minimal competition. Both signals are meaningful for prospective subscribers and for policy researchers tracking the long-run shift in the U.S. broadband market.

Several caveats are worth flagging directly. First, the FCC dataset reports advertised maximums, not measured throughput, and real-world experience can diverge meaningfully from the headline tier. Second, ZIP-level rollups can mask address-level variation. A single ZIP can contain census blocks with full fiber coverage and adjacent blocks with only legacy DSL. Third, fixed wireless and satellite providers like T-Mobile, Verizon, and Starlink often appear in many ZIP codes simultaneously, which inflates simple provider-count metrics relative to wired alternatives. The competition factor in the BroadbandGrade rubric requires each counted provider to meet the FCC 100/20 Mbps benchmark, which screens out the lowest tiers and produces a more meaningful competition signal than a raw provider count.