Nvidia Plans to Convert About 130 Brightspeed Telecom Buildings into Small AI Data Centers

• Dispatched: Nov 7, 2026 · 19:38 ET (Nov 8, 00:38 UTC)

Magnitude: Medium news (Magnificent Seven / data centers and AI compute infrastructure: categories 2 and 7)

Originally reported: November 5, 2026 ET; relayed by StockTwits at 16:31 ET on November 5)

In one sentence: The Information reported on November 5 that Nvidia has submitted a written proposal to Apollo Global Management to convert more than 130 telecom buildings owned by its broadband provider Brightspeed into a network of small AI data centers spanning up to 20 US states, with Nvidia estimating the sites hold roughly 1 gigawatt of available power and some buildings able to begin installing Vera Rubin server racks as early as the first quarter of 2027.

Source links:

• The Information report (via StockTwits)

• The Information report (via TradingKey)

• Cross-check Verified content:

• Nvidia has approached Apollo Global Management with a plan to use more than 130 telecom buildings owned by Brightspeed (formerly used to handle voice and internet traffic) as a distributed network of small AI data centers spanning up to 20 US states; Nvidia estimates the buildings together hold roughly 1 gigawatt of available power, and many sites already have electrical connections capable of supporting Nvidia servers.

• Nvidia formally submitted a written proposal to Brightspeed on Wednesday, November 4, and has requested additional facility data and site visits; Nvidia staff have discussed with suppliers beginning retrofits with Vera Rubin racks as early as the first quarter of 2027.

• The proposed sites are described as an "inference network" for running already-trained models, rather than large-scale training clusters.

• Brightspeed is the local phone and broadband business Apollo bought from Lumen in 2021 for about $7.5 billion; it is reportedly under substantial debt pressure, seeking financing and possibly filing for bankruptcy protection in December.

• Background: after the midterm elections, Democratic winners in states including Texas proposed moratoriums on new data center approvals, making retrofits of existing buildings a way to sidestep new-construction approval resistance and get chips running faster; Apollo separately reached an approximately $40 billion AI chip and server lease financing arrangement with SpaceX.

Why it matters: This is Nvidia's direct business response to the core driver of this week's sell-off across AI and semiconductor equities (concerns that data center construction will be restricted): instead of building new sites, convert existing buildings with power already in place to bring Vera Rubin chips online faster. If it goes ahead, a roughly 1-gigawatt distributed deployment channel would become a new shipment path for Nvidia, directly relevant to the supply-chain setup ahead of Nvidia's November 17 earnings report.

Sourcing note ⚠ single source, unverified:

This report rests on two people with direct knowledge cited by The Information; neither Nvidia nor Apollo has commented in the reports available, and no agreement has been signed. The proposal stage is all that has been reported: the exact number of buildings to be converted, who would purchase or lease the chips, and customer arrangements are all undetermined, and neither company has confirmed the plan. No further reliable details are available.