State of the Mobile Union Report

A major element in AT&T’s 5G story involves the midband 3.45 GHz spectrum licenses the operator is using to supplement its existing C-band (3.7 GHz – 3.98 GHz) spectrum holdings. The operator is clearly keen on this band, considering the $23 billion worth of such spectrum AT&T agreed to purchase from EchoStar in August. Midband spectrum, like 3.45 GHz and C-band, generally creates the foundation for speedy 5G networks.

In their tests in the second half of 2025, RootMetrics’ technicians began seeing additional 3.45 GHz holdings popping up in AT&T’s network starting in September, in locations stretching from California to Indiana. Testing showed AT&T generally deploying 5G on 60 MHz chunks in the band, up from 40 MHz chunks previously. A few areas, including in Oklahoma and Maine, showed AT&T services across 100 MHz blocks of spectrum in the 3.45 GHz band. This expansion—touted by AT&T itself in November—will undoubtedly help improve the operator’s overall 5G performance. After all, more spectrum typically results in better, faster networks.

AT&T’s shift from Nokia to Ericsson gear is also clear in RootMetrics data. Ericsson now accounts for more than 80% of AT&T’s radio access network (RAN) in RootMetrics testing samples. Further, AT&T’s switch to Ericsson seems to be bearing fruit: AT&T markets with Ericsson equipment ranked second in RootMetrics performance metrics, trailing only those operated by Verizon and its vendors (Samsung and Ericsson). In terms of RootMetrics’ reliability testing, AT&T is also seeing higher task success rates, and higher median task speeds, in its Ericsson markets when compared with its legacy Nokia markets.

Finally, AT&T tied with Verizon on the RootScore Award for call performance—that’s noteworthy considering the vast majority of AT&T’s calls in RootMetrics testing traveled over its VoLTE service. Voice over LTE (VoLTE) technology sends voice calls over a 4G LTE connection, generally producing clearer audio and faster call connections than legacy voice protocols. In comparison, most of T-Mobile’s calls traveled over the newer VoNR standard (basically, voice calling over 5G), while Verizon used a mix of VoLTE, VoNR, and EPS Fallback, a 5G technology that automatically switches a call to a 4G LTE network if necessary.Overall, less than 1% of the calls traveling over AT&T and Verizon’s networks failed in the U.S. cities covered by RootMetrics’ metro-area testing. In more rural RootMetrics state-area testing (which covered locations in large and small towns, as well as the highways between them), that call failure rate hovered around 3% (or three out of every 100 calls). T-Mobile’s call-failure rates were a few points higher in both metro and state testing, which may reflect the teething pains of implementing the relatively new VoNR standard.

T-Mobile continues to leverage the latest in 5G technologies to juice the speeds of its connections. That strategy is clear in the operator’s ongoing embrace of 5G SA technology as well as its effort to stack together multiple aggregated carriers. (Carrier aggregation technology essentially combines transmissions across multiple spectrum bands into a single data pipe, thus supporting faster speeds and more stable connections.)

In metro areas, fully 93.2% of T-Mobile’s testing samples leveraged 5G SA technology. That’s important considering SA connections generally support snappier connections as well as advanced 5G technologies like network slicing (which can carve out network resources for specific applications). In comparison, AT&T showed no SA samples in RootMetrics testing, while Verizon’s SA samples in the fourth quarter of 2025 totalled 59.7%, with the remainder of its connections spread across 5G NSA and LTE.

T-Mobile also showed a lead in carrier aggregation: Just over two-thirds of T-Mobile’s 5G samples in urban, metro areas used 4-carrier aggregation. That level of aggregation created a connection spanning 216.9 MHz worth of spectrum. That’s a big reason why T-Mobile’s national median download speed reached 374.5 Mbps and was faster than those of both AT&T and Verizon.

In comparison, almost a half of AT&T’s samples in metro areas used 2-carrier aggregation (for a total of 128.9 MHz worth of spectrum), and a majority of Verizon’s connections used 3-carrier aggregation (for 167 MHz).

The correlation between spectrum and speed is also visible in RootMetrics’ calculation of the total average 5G and LTE spectrum deployed by each of the operators in urban, metro areas. For this specific metric, T-Mobile registered 191.3 MHz in the fourth quarter of 2025. That allowed the operator to post median download speeds of 508.2 Mbps (again, this is specific to metro areas). Verizon’s 382.5 Mbps relied on the operator’s 174.1 MHz worth of spectrum. Meanwhile, AT&T’s 383.6 Mbps speeds used just 134.7 MHz of bandwidth, indicating the operator is able to eke out clear efficiencies from its spectrum holdings.

So why didn’t T-Mobile win the Fastest 5G Award? Basically, it’s because Verizon’s performance among its slowest testing samples (the bottom 5%) was better. RootMetrics’ scoring system doesn’t just measure the fastest single connection an operator provides (although it does do that too—Verizon’s peak speeds in urban areas, powered by its millimeter-wave spectrum, reached 5.49 Gbps in the fourth quarter of 2025!). Instead, RootMetrics’ scoring system looks at all of an operator’s testing samples, all over the country, and uses a scoring curve to determine the overall fastest provider across all of those samples.

A headline for Verizon’s 5G efforts in 2025 would center on its shift to SA technology. Verizon expanded its use of 5G SA from 24.5% of samples in the first half of 2025 to fully 59.7% in the second half of 2025, in RootMetrics’ metro testing areas. Verizon’s embrace of 5G SA varies across the country, ranging from 19.4% in Jackson, Mississippi, to 62.3% in Reno, Nevada, to 98.4% in Omaha, Nebraska.

But the share of Verizon’s 5G SA samples were lower in the more rural, state-area locations in RootMetrics’ testing. Specifically, 28.7% of Verizon’s samples in these locations registered 5G SA technology in the second half of 2025, up from 10.2% in the first half. That’s not a surprise, though, considering wireless operators typically deploy advanced technologies first in big cities and then, later, into more rural locations.

Also during 2025, Verizon increased the use of its midband C-band spectrum to 81.3% of all samples in metro areas by the fourth quarter of 2025, up from 74.4% in the first quarter of 2025. And it shifted more of its systems to higher orders of QAM (which packs more data into transmissions) and higher orders of MIMO (which transmits and receives signals across multiple antennas). These efforts generally helped Verizon expand the reach and performance of its 5G network (though it’s not yet clear how Verizon’s new CEO might address this work into 2026).

Interestingly, Verizon also leveraged SRS (Sounding Reference Signal) antenna switching technology to a greater degree than either AT&T or T-Mobile. Close to half of Verizon’s samples in metro areas leveraged this technology. SRS antenna switching allows a user’s phone to quickly cycle its transmitter across all its antennas, helping an operator’s base station “see” which one is best for sending back data. This too can support better connections.

But it was Verizon’s performance in more rural areas—those along RootMetrics’ state-area testing locations—that did the most to raise the operator’s Overall RootScores. There were a number of factors at play here.

First, Verizon made more use of its legacy 4G LTE operations than AT&T and T-Mobile. Almost half of Verizon’s samples along RootMetrics’ state-testing routes connected to its 4G LTE network, above the one-third of AT&T’s samples, and far ahead T-Mobile’s 4G LTE usage, which was in the single digits. This is important because Verizon, like all U.S. operators spent, much of the 2010s deploying and reinforcing its 4G LTE network all over the country. That network is now mature and robust. 5G technology, meanwhile, is just a few years old.

Further, Verizon’s 4G LTE play leveraged its 700 MHz Band 13 and 850 MHz Band 5 spectrum holdings in these more rural areas, a move likely intended to harness the coverage capabilities of this lowband spectrum.

This strategy helped Verizon provide suitable connections even among RootMetrics testing samples with the slowest speeds.

Note: All testing in this report is based on data collected from off-the-shelf smartphones, with measurements pulled from the Android operating system and device modem logs. The standard results and awards sections focus on application-layer measurements, which are designed to reflect real-world user experience, while the Analyst Deep Dive focuses on physical-layer measurements, which are better suited for detailed technical analysis. Because application-layer and physical-layer measurements capture performance differently, metrics in the Analyst Deep Dive may differ from those reported elsewhere in this report.

Mike Dano

Mike Dano is a Lead Industry Analyst in Ookla’s research and content team. He covers the North and South American markets, and global technology trends. Previously, Mike was a journalist covering the global telecom industry for 25 years at publications including RCR Wireless News, Fierce Network and Light Reading.