AT&T’s Fiber Joint Venture: 3.3 Million Addresses in 76 Metros and the Operators Who Share Them

The venture combines two networks. Forged Fiber 37 holds the consumer fiber business, construction operation and customers AT&T acquired from Lumen in February 2026, in eleven states from Florida to Washington. Gigapower is the open-access network AT&T formed with an infrastructure investor in 2023. AT&T will own half the combined company. The deal is expected to close in the first half of 2027 with nearly 5 million fiber locations, more than 1 million AT&T subscribers and 16 states.

The FCC filings for December 2025 locate 3.3 million of those addresses: 3.1 million on Lumen’s fiber and 181,000 on AT&T fiber built outside AT&T’s own telephone territory, which is where Gigapower operates. The remaining gap of about 1.7 million is explained by business addresses, which this series excludes, by ten months of construction since the filing date, and by Gigapower markets inside AT&T’s own states that the filings cannot separate. Treat 3.3 million as a floor.

Cite these figures: G2M Insights, "AT&T’s Fiber Joint Venture: 3.3 Million Addresses in 76 Metros and the Operators Who Share Them," G2M Communications Research, edition open-access-2025-12-v1, g2m.ai/research/communications/att-fiber-joint-venture.

Plan against the network that exists, not the closing date. The fiber, the customers and the construction crews behind these addresses have belonged to AT&T since February 2026; the 2027 closing changes who funds the next phase, not who you compete with today.

Sort the venture’s addresses into the nine market types this series uses and almost all of them fall into two. Cable against fiber covers 70% of venture addresses, against 40% nationally. Two or more fiber networks cover 28.2%, against 10.9%. The types that define rural and underserved America, a single cable operator, cable against fixed wireless only, or no qualifying service at all, are effectively absent.

Cite this chart: G2M Insights, "AT&T’s Fiber Joint Venture: 3.3 Million Addresses in 76 Metros and the Operators Who Share Them," G2M Communications Research, edition open-access-2025-12-v1, g2m.ai/research/communications/att-fiber-joint-venture.

Counted another way: at 60% of venture addresses at least two other operators already offer 100/20 Mbps, and only 1.1% have none. A cable network offering 100/20 Mbps is present at 92% of them.

Read the venture as a change in the quality of an existing competitor, not as a new entrant. Your churn and win-back models already contain this fiber network; what they do not contain is the operator now running it.

The venture’s addresses sit almost entirely in 76 metros, and the twelve largest hold most of them. The share of each metro it reaches is what decides how far a local response has to go. In Cape Coral, St. George, Denver, Salt Lake City, Omaha, Minneapolis, Colorado Springs and The Villages the venture reaches a third or more of every residential address.

Cite this chart: G2M Insights, "AT&T’s Fiber Joint Venture: 3.3 Million Addresses in 76 Metros and the Operators Who Share Them," G2M Communications Research, edition open-access-2025-12-v1, g2m.ai/research/communications/att-fiber-joint-venture.

The metros differ in what sits alongside the venture. In Salt Lake City, Des Moines, Ogden and Fort Collins an independent fiber network already serves more than three quarters of venture addresses, so the venture is the third wireline option there, not the second. In Seattle, Portland, Las Vegas and Tucson there is almost no independent fiber at all, and the contest is the venture against cable alone.

Rank your metros by the share of addresses the venture reaches, not by its address count. Where it reaches a third of the metro or more, a metro-wide response is justified; where it reaches a fifth, a response priced for the venture’s blocks alone costs less and protects the same customers.

AT&T has said Lumen’s fiber was about 25 percent penetrated when it agreed to buy it, against about 40 percent on AT&T’s own fiber. Applied to the 3.1 million Lumen fiber addresses we can locate, closing that gap would mean roughly 470,000 more fiber households. That is an illustration from AT&T’s own figures, not a forecast and not a company target.

Those households would not be new to broadband. Across this series, the share of households buying wireline broadband in a neighborhood is set by its demographics, and the competitive structure around it does not move that number. In markets where cable faces fiber, a fiber operator raising its penetration takes customers from the other operators at the address. At venture addresses, that is cable at almost every one and a national fixed wireless operator at most.

Cite this chart: G2M Insights, "AT&T’s Fiber Joint Venture: 3.3 Million Addresses in 76 Metros and the Operators Who Share Them," G2M Communications Research, edition open-access-2025-12-v1, g2m.ai/research/communications/att-fiber-joint-venture.

The second change is the bundle. Lumen sold broadband on its own. AT&T sells it with mobile service, and reports that more than 40 percent of its home internet households also buy its wireless. National fixed wireless operators report service at 91% of venture addresses, at 100/20 Mbps or better at 46%. At those addresses the venture puts fiber and mobile in one offer against a rival whose home broadband is itself a mobile network product.

Assume the households at risk are your current customers, not new ones. Set a retention budget for the venture’s addresses now, and if you sell mobile, put it in the offer at those addresses first: it is the part of the bundle the previous fiber operator never had.

In all, 308 operators report residential service somewhere in the venture’s footprint. Measured in addresses, the overlap is dominated by the largest operators: the two national fixed wireless operators and the largest cable operators each share hundreds of thousands to more than two million addresses with it. Measured as a share of their own footprint, the same operators are the least exposed, because the venture covers a small part of a national network.

Cite this chart: G2M Insights, "AT&T’s Fiber Joint Venture: 3.3 Million Addresses in 76 Metros and the Operators Who Share Them," G2M Communications Research, edition open-access-2025-12-v1, g2m.ai/research/communications/att-fiber-joint-venture.

Below the named brands sit the operators the filings do not identify by brand. Among operators with fewer than a million addresses, 222 have a quarter or more of their footprint in the venture’s footprint or the copper ring beside it, and 158 have half or more. Almost all are local fixed wireless operators and independent fiber builders.

Cite this chart: G2M Insights, "AT&T’s Fiber Joint Venture: 3.3 Million Addresses in 76 Metros and the Operators Who Share Them," G2M Communications Research, edition open-access-2025-12-v1, g2m.ai/research/communications/att-fiber-joint-venture.

Measure your exposure as a share of your own footprint, not in addresses. A national operator can absorb a loss of share across a few percent of its network; an operator with half its customers inside these counties cannot, and needs a plan for the venture before it closes.

Lumen built fiber by upgrading its own copper telephone network, and its construction operation comes with the venture. In the eleven states of the Lumen footprint, 8.3 million residential addresses are still served by Lumen copper with no Lumen fiber, and 7.5 million of them sit inside counties where the venture already has fiber. AT&T has said the venture is its main vehicle for building beyond its own telephone territory and that it expects to pass more than 60 million fiber locations by 2030.

Where the venture builds next has not been announced, and this ring is our inference from where the network and its construction capacity already are, not a plan. It is still the most useful map an operator has: an upgrade here turns a copper telephone line into a gigabit fiber rival at an address you already serve.

Addresses are counted once per operator and summed within a type, so a type’s total can exceed the number of distinct addresses.

The ring also changes the value of what other operators have built. The venture is a wholesale company with infrastructure investors behind it, and a network already lit inside its counties is the fastest way to add addresses. For a fiber operator in these counties, the venture is a likely competitor and a possible buyer at the same time.

Map your customers against the ring as well as the footprint. Where you serve copper-only addresses today, the case for upgrading or locking in long-term customers is strongest before a fiber build is announced, not after.

Six findings, read from the operator’s side of the table.

• If you sell cable in these metros, the venture’s growth comes from your customers. Treat its addresses as a defined retention market with its own budget, and lead with mobile where you sell it. Roughly 470,000 households is the scale of the gap between Lumen’s penetration and AT&T’s.
• If you sell fixed wireless nationally, you are present at 91% of venture addresses, and they now face a fiber offer bundled with a rival’s mobile service. Your case there is price and simplicity; make it before the venture makes its own.
• If you build fiber in the same counties, you are already the second fiber network at many venture addresses, and you are the network an open-access wholesaler would most want to add. Decide which you are: a competitor accelerating in the copper ring, a tenant or partner, or an asset. Each needs a different plan and a different set of records.
• If you run a local fixed wireless network in these states, check your share of footprint in the ring first. 127 smaller fixed wireless operators have a quarter or more of their addresses in the footprint or the ring, and a copper upgrade there removes the reason many of your customers chose you.
• If you are a cooperative or a rural telco, the venture barely overlaps you today: incumbent telco service is present at 2.6% of its addresses. Its relevance to you is as a model and a buyer. If you consider open access, the terms that matter are who sets engineering standards, who holds the records and who owns the customer.
• Whoever you are, rank your markets by the venture’s share of the metro, not its address count, and put the copper ring on the same map. The closing date in 2027 is not the start of this contest.

These figures are reported availability, not subscribers. They mark where the venture competes and with whom, not who will win.