AT&T Plans a Smaller Workforce. Verizon Is Paying for AI Upskilling.

In the same week, AT&T said it expects a smaller workforce five years from now and Verizon committed $70 million to free AI upskilling. Both companies can buy the same AI. The advantage that is hardest to copy will belong to whoever builds the people who know what to do with it, and these two telecom rivals just placed their bets in public.

Both moves deserve a fair reading. Jeremy Legg, AT&T's technology chief, was candid about headcount in an interview with WIRED and just as candid that AT&T will keep hiring. Verizon CEO Dan Schulman put real money on the table. The useful question is what each bet protects.

AT&T is planning for fewer people and new roles

Legg told WIRED that AT&T expects a smaller workforce in five years, and that the company benchmarks its staffing against Verizon and T-Mobile. AT&T is automating customer service, network operations, tower site location, maintenance issue detection, software code writing and service problem management. WIRED's reporting framed his statement as a direction for headcount, with no agreed endpoint.

AT&T had 131,000 employees as of June 2026. It cut 8,000 jobs in 2025 and another 2,100 in the first half of 2026. You may have seen a bigger figure repeated online: 85,000 job cuts by 2030. AT&T disputes it as an inaccurate, unofficial extrapolation: WIRED calculated it from 2025's cuts, and an anonymous source called it a target. AT&T also has not said how many of its cuts came from automation and how many came from everything else.

Legg deserves credit for the second half of his answer. AT&T will still hire, specifically people to "develop and govern AI agents" and people to maintain fiber lines. A company that plans to put software agents into its network and names the governance of those agents as a hiring priority is planning for who checks the machine's work.

Any of those 131,000 employees who read the interview this week faced the same kitchen-table question: is my work on the automation list or the hiring list? Every employer owes its people a plain answer to that question before the next round of headlines about AI job cuts answers it for them.

Verizon is paying for AI upskilling beyond its own payroll

On September 23, Verizon announced Verizon AI Skills for America, a $70 million AI upskilling initiative built from $20 million in its existing Reskilling and Career Transition Fund plus $50 million in new money. The training is free, and Verizon puts the price of comparable premium training at $700 per person per year. IBM, Google, Microsoft, Anthropic, Coursera and OpenAI supply the course content. Community partners, including Goodwill Industries International, the largest nonprofit provider of job training and career placement services in North America, carry it to the people who need it.

The program is aimed at job seekers, early-career professionals, displaced workers, educators and small businesses. Schulman made the case for it in one sentence:

"Companies, working closely together and with the public sector, have a responsibility to invest in people with the same urgency they invest in technology."

Here are the two bets side by side, using only what each company has put on the record.

Both companies will soon have the same AI

Any telecom, AT&T and Verizon included, can buy AI from the same short list of providers, and most of the companies writing Verizon's courses sell their models to any business that pays. Stanford HAI's AI Index has tracked the distance between the leading AI models and the fast followers shrinking year after year. Follow that trend line far enough and everyone will have the same AI. Not everyone will have the same people.

What each company owns, once you subtract the identical tools, is judgment: the network engineer who knows which alarm matters at 2 a.m., the service rep who can tell when the model's answer will lose a customer. AI raises every competitor's floor at the same moment. People raise the ceiling.

We have watched a version of this before. In March 2007, Circuit City fired its 3,400 highest-paid, most experienced salespeople to cut costs. By November 2008 it was in Chapter 11, and Best Buy hired a number of the people Circuit City had let go. Many forces drove that bankruptcy. The cost cut failed to stop it, and the expertise Circuit City shed went to the competitor that survived.

I spent 20 years at Cisco and shaped its $1.1 billion innovation portfolio. The lesson I carried out of that work is simple: a competitor can buy your technology within a year or two, and it takes far longer to find people who know what to do with it.

Training counts when it turns into jobs

Verizon's bet has its own open question, and Schulman's standard of urgency is the right yardstick for it. Free courses are generous, and $70 million is a serious commitment, yet the measure that matters is how many displaced workers and early-career professionals land work that uses what they learned. The announcement names partners and audiences but no enrollment or placement numbers. Goodwill's placement work gives the program a path from course to paycheck; the results will show how many people walk it.

AT&T has its own version of the question. Legg's plan to hire people who govern AI agents is a smart move. The harder test is whether the current workforce gets a fair shot at those roles. A customer service veteran who has handled years of hard calls knows where an automated system will fail a customer, and that kind of knowledge never ships with a software license.

In my forthcoming book, The Great Relearning, I call this moment the relevance cliff: the point where a person's skills stop matching the work their employer plans to do next. Whether that person reaches the next role, or spends their career wondering will AI take my job, depends on whether anyone paid for the climb.

If you lead a team, take one question into your next leadership meeting: when we buy the next AI tool, what are we buying for the people who will judge its output? If your own work sits on one of those automation lists, ask your manager which list it is on and what training comes with it. Companies worth staying at can answer.

AT&T and Verizon will both have strong AI five years from now. The people who know what to do with it will decide which one uses it better. If you are weighing the same choice, I'm easy to find.

Here is what makes Alex a credible voice on this topic: Alex Goryachev shaped Cisco's $1.1B innovation portfolio and now advises the California State University system on AI governance, so he knows from the inside the choice AT&T and Verizon made this week: what to spend on technology, and what to spend on the people who use it.

Bring Alex in to help your leadership team plan AI and workforce decisions together →