AT&T's 600 MHz Ericsson radios: why the real story is uplink, not coverage

Two blocks of spectrum arrived at AT&T in the same transaction. One is already carrying traffic across nearly 23,000 cell sites. The other is carrying nothing at all — and will not carry anything for a long while yet.

That asymmetry is the most instructive thing in this week's announcement, and it is the part most coverage will skate past. For anyone who buys, integrates or plans radio networks, the AT&T–Ericsson 600 MHz agreement is not a coverage story. It is a case study in three things that decide whether a spectrum asset becomes revenue or becomes a stranded balance-sheet item: radio hardware lead times, tower-top physics, and the fact that uplink has become the binding constraint on network design.

According to AT&T's announcement, Ericsson was selected through what the operator described as an open and competitive process to supply 600 MHz dual-band radios for spectrum acquired from EchoStar, enabling AT&T to deploy uplink-enhancing 8RX technology across low bands for the first time. Network CTO Yigal Elbaz framed it as continuity rather than a new programme:

We're more than halfway through our wireless network modernization. — Yigal Elbaz, SVP & Network CTO, AT&T

Now let's talk about what the engineering actually does.

Uplink is the constraint — and the US starts from behind

Start with the number that should be on every RF planner's wall. Light Reading, citing Ookla's research, reports that US operators dedicate just 5.1% of network throughput to the uplink — the lowest share among the 22 markets Ookla studied, and down from 8.0% in 2023.

That is the starting position for an industry now betting on agentic workloads, real-time video and fixed wireless access.

Against that backdrop, the receive-chain upgrade matters more than the spectrum. Light Reading notes that low-band FDD spectrum such as 600 MHz is typically served by radios with 4RX; AT&T is moving to eight receive branches.

The engineering logic is first-principles and worth stating plainly: doubling the number of receive branches doubles the signal energy the base station captures, which in ideal maximum-ratio-combining conditions corresponds to about 3 dB of array gain, with further diversity gain where the additional branches see decorrelated fading.

Omdia senior principal analyst Gabriel Brown, quoted by Light Reading, placed the move in context: 8RX is more often deployed in mid-band FDD and remains rare below 1 GHz, though he says technology gains are making it viable there — and he called it brilliant for uplink and coverage. Brown frames the wider trend as an "FDD renaissance": after years in which 5G attention went to TDD massive MIMO in mid-band for downlink headline speeds, the FDD workhorse bands are getting engineering attention again, with what he expects to be a positive effect on cell-edge reliability. He also notes strong US interest in FDD massive MIMO, with Verizon leading in that technology.

There is a second advantage that deserves more attention than it gets. 8RX is a base-station-side change. Receive combining happens in the radio; it asks nothing of the handset. Unlike uplink MIMO, which depends on two-transmit-chain devices and a refresh cycle no operator controls, receive-side gain accrues across the entire installed base the moment a sector is turned up. For anyone modelling RAN capex payback, that is a materially different curve — and it is NextGComm's view that this, not the AI framing, is the strongest commercial argument for the upgrade.

AT&T CEO John Stankey has been signposting the rationale. On the company's Q2 earnings call, as reported by Light Reading, he tied the acquisition to managing robust upstream traffic in an agentic environment, arguing the way to do that is a really strong lowband position and that 600 MHz would make AT&T's existing low-band advantage more substantial. Strip out the AI framing and the logic still holds: uplink capacity already matters today for video calling and FWA, and T-Mobile raised the same subject on its own Q2 investor call.

The dual-band decision is the commercially clever part

The dual-band designation refers to support for Band 14 and Band 71 — FirstNet's 700 MHz public safety spectrum alongside the newly acquired 600 MHz.

This is not a spec-sheet footnote. It is the deployment economics.

A single-band 600 MHz radio would mean an additional unit at every site: a new climb, incremental wind loading, structural analysis against TIA-222, a lease amendment with the tower owner, and possibly a new passive antenna. Multiply across a nationwide footprint and the programme stops penciling. A dual-band unit lets AT&T address Band 14 and light Band 71 in the same visit, in the same footprint — and, as a consequence of the shared radio, brings 8RX receive gain to FirstNet's public safety uplink as well.

For integrators and tower services firms, that bundling is the pipeline signal in this announcement.

The open question to put to vendors is antenna ports. Eight receive branches on a band require eight receive paths into the antenna system. Neither AT&T nor Ericsson has identified which radio model is being supplied, and neither has said what the configuration requires at the antenna. Whether existing passives can carry it across AT&T's actual estate is simply unknown from public information — and if a meaningful share of sites need antenna work, cost and schedule change materially. Ask it in every vendor conversation this quarter.

How this compares with the alternatives

AT&T's approach is one of several ways to attack uplink, each with a different dependency profile. The table below is a general engineering comparison of 3GPP-defined options, not a description of any operator's disclosed plans.

The competitive read: AT&T is not the first US operator into 600 MHz — T-Mobile has built its nationwide low-band 5G layer on the band — but AT&T is arriving with a receive architecture that Omdia describes as rare below 1 GHz. Whether that becomes a measurable field advantage depends entirely on deployment pace.

The open RAN question worth asking out loud

AT&T ran what it called an open and competitive process and selected Ericsson — already its prime wireless supplier under a $14 billion five-year agreement announced in December 2023, which Light Reading describes as covering Nokia radio replacement, open RAN and cloud RAN adoption. Fierce Network asked how many bidders were considered; AT&T declined to share details of its bid process, and also declined to say how many radios it is buying.

This deserves scrutiny rather than cynicism. Open RAN's core promise was that standardised interfaces would let operators mix radio vendors. Here, an operator with one of the world's most advanced O-RAN programmes ran a competitive process for a radio unit — architecturally the most substitutable element — and selected the incumbent.

There are defensible reasons: integration risk, RIC and management-plane compatibility, spares and field-services commonality, and the genuinely narrow supplier field for dual-band B14/B71 4T8R units. But the lesson for anyone building an open RAN business case — including enterprises sold the multi-vendor pitch — is that open interfaces reduce switching costs; they do not eliminate them.

The clock is regulatory, and it is specific

AT&T has not given deployment timelines or commercial availability dates, and has previously said the rollout would take a number of years.

The contrast with 3.45 GHz makes the hardware point better than any analyst could — and the FCC's own order spells out the mechanism. In its filings, AT&T explained that its existing 3.45 GHz deployments run on software-defined radios already installed across its network, meaning EchoStar's 3.45 GHz could be added by remote reconfiguration rather than a site visit. Combined with a short-term spectrum manager lease that let AT&T start before the deal closed, that is how the mid-band went live across nearly 23,000 cell sites covering more than 5,300 cities, with AT&T reporting up to an 80% increase in mobile download speeds and 55% for fixed wireless customers. AT&T also told the Commission it does not currently use 600 MHz spectrum in its network at all. Same transaction, same operator — one band was a software change, the other is a hardware programme.

On buildout, the FCC set requirements tougher than AT&T proposed. Per Light Reading's reading of order DA 26-470, AT&T must cover at least 40% of the population under its 600 MHz licences within three years of transaction close, and 75% within five years.

Those are hard dates against a narrow supplier base and a nationwide site programme.

For context on what the modernisation has delivered so far, AT&T says that in areas with upgraded Ericsson equipment it is seeing up to two times improvement in average speeds, 10% fewer dropped calls, 70% fewer occurrences of slower data speeds and reduced uplink interference. These are operator-reported figures for the existing programme, not projections for 600 MHz.