AT&T confirms ongoing layoffs and copper network retirement as it expands AI use in network operations and customer service
AT&T confirms continued workforce reductions (following 8,000 cuts last year and 2,100 in H1 2026, with workforce near 131,000 as of June and a reported internal target of ~85,000 by 2030, which AT&T disputes as inaccurate) alongside decommissioning its copper landline/DSL network (85%+ of footprint copper-free by end of 2026) and deploying AI/automation internally: AI for customer service, cell tower site selection and maintenance issue detection, a generative AI system called GeoModeler for real-time network adjustments during extreme weather, AI-assisted software code generation, and a planned 2027 rollout of cloud software from Israeli startup DriveNets to replace physical hardware at thousands of central hubs (enabling remote technician adjustments).
Entities: AT&T, Jeremy Legg, John Stankey, Pascal Desroches, DriveNets, GeoModeler
0 primary
What happened
AT&T's CTO confirmed to WIRED that job cuts are continuing (8,000 last year, 2,100 in the first half of 2026, headcount near 131,000 as of June) alongside retirement of its copper landline/DSL network, targeted at over 85% copper-free by end of 2026. The company is also deploying AI internally for customer service, tower site selection, maintenance detection and a weather-response tool called GeoModeler, and plans a 2027 rollout of cloud software from DriveNets to replace physical hardware at thousands of hubs. A reported internal target of cutting headcount to around 85,000 by 2030 is sourced anonymously and AT&T disputes it as inaccurate.
Why it matters
This confirms a large, regulator-gated infrastructure shift (copper decommissioning) and a structural move from hardware to cloud-managed network operations, both of which are durable and checkable against future filings. It matters most to AT&T workers and unions facing continued reductions, to telecom vendors and competitors watching the DriveNets hardware-replacement model, and to investors tracking cost structure. The AI framing adds less than it appears: much of this is the continuation of a telecom employment decline that predates generative AI by two decades.
What is noise
The 85,000-by-2030 headcount target, the number doing the most work in the "AI is gutting jobs" narrative, is anonymously sourced and explicitly disputed by AT&T, so it should not be treated as confirmed. The article leans on an AI-causes-layoffs frame when copper retirement and telecom headcount decline are long-running trends with multiple causes (automation broadly, cord-cutting, cost pressure from Starlink and cable competitors), not a new AI-driven rupture. There is no primary source document (filing, transcript, press release) underlying the extraction, only secondary reporting with on-record quotes.
Watch next
- 01AT&T's Q4 2026 and Q1 2027 headcount disclosures (10-K/10-Q) to check whether reductions are tracking toward 85,000 or AT&T's denial holds up
- 02Confirmation or denial of the 85,000-by-2030 figure from AT&T in an investor call or SEC filing
- 03Whether the DriveNets cloud rollout actually begins in 2027 as planned and whether it triggers further hub-level headcount cuts
- 04Copper network retirement progress reports (FCC filings or AT&T disclosures) against the 85%-by-end-2026 milestone
Coverage
1 storyMore labour signals
Full feed →- NYT-led publishers file summary judgment brief citing internal OpenAI/Microsoft messages calling AI training "astonishing theft" and admitting chatbots substitute for journalism18 Sept 202686
- Unsealed filings in NYT v. OpenAI/Microsoft copyright suit reveal internal admissions calling AI training data scraping "theft" and an "existential threat" to publishers17 Sept 202674
- Anthropic exposes network of ~28 AI-powered dating apps using Claude to run automated catfishing scams at scale16 Sept 202673
- Microsoft leads takedown of "EvilTokens" AI-assisted phishing/fraud platform, disrupting scheme that compromised 12,000 Microsoft accounts22 Sept 202672