In a Wall Street Journal interview published April 19, Verizon CEO Dan Schulman said two things that tested how blunt a Fortune 50 chief executive is allowed to be in 2026.
The first: AI will push US unemployment to 20% to 30% within the next two to five years. The second: in a recent all-hands meeting, he asked Verizon employees to use AI tools to write their own obituaries — to see, he said, how the technology framed their lives.
The interview hit on a Sunday. By Friday April 24, Bloomberg’s editorial board had published a counter-piece titled “The AI Job Apocalypse Is Being Delayed”. It did not name Schulman. It did not need to.
What Schulman actually said
Schulman became Verizon’s CEO in October 2025. One month into the job, the company announced 13,000 layoffs — the largest single workforce reduction in Verizon’s history, roughly 13% of the company. He simultaneously stood up a $20 million career-transition and retraining fund for affected employees and began publicly pushing other Fortune 100 CEOs to match it.
Six months later, in the WSJ interview, he laid out what comes next:
- AI achieving artificial general intelligence by the end of 2027, in his estimation.
- Quantum computing arriving as the next layer of disruption shortly after.
- Humanoid robots displacing manual-labor jobs that have been considered safe.
- Total unemployment 20–30% in 2–5 years as a base case, not a tail risk.
- A blanket charge to other CEOs: stop pretending. “Being authentic, being realistic, telling the truth, as best you can” is, per Schulman, the actual job description for a 2026 CEO.
The internal-meeting detail — write your own obituary using AI — landed differently. It is the kind of thing that reads either as deliberate provocation, sincere reflection exercise, or unforced HR error, depending on which of the 99,600 Verizon employees you ask. The framing in the WSJ piece is the second; the framing on Verizon-employee-adjacent message boards is closer to the third.
What Bloomberg ran on Friday
Five days later, Bloomberg’s opinion side ran an unsigned-by-headline piece arguing that the AI job apocalypse is being delayed, with three specific structural reasons:
- Disruption arrives in recessionary bursts, not boom-time linearity. The piece points to travel-agent employment as the canonical example: travel-agent headcount fell rapidly during and after the 2001 recession, despite the internet-booking infrastructure existing through the late 1990s. The technology was ready years before the labor market reflected it.
- Corporate profits are too high to force the issue. Companies cutting AI capex out of operating expense are not, on aggregate, companies whose revenues are falling. Until top-line growth rolls over, the political and HR cost of mass involuntary cuts exceeds the savings.
- Compute supply is the actual rate limiter. AI agent deployment at the scale that would justify a Schulman-level unemployment forecast requires GPU capacity that does not currently exist. Half of US data-center buildout planned for 2026 has been cancelled or delayed.
The piece’s punchline: workers should not fear the AI boom; they should fear the next recession. Whatever AI displacement is coming will arrive inside the next downturn, not before it.
Who’s right
Both. The disagreement is timing and trigger, not direction.
Schulman’s frame: a Fortune 50 CEO running an AI-first restructuring right now, with 13,000 cuts already executed and a $20M retraining fund as evidence the next wave is coming. The forecast is grounded in his specific operating reality. Verizon’s customer-care headcount is down materially, AI is doing the routing and triage, and the headcount line never goes back up.
Bloomberg’s frame: a macro-level read on the labor market that has, so far, refused to crack. Q1 2026 unemployment was 4.0%, and tech-sector layoffs of 92,000 in four months are a rounding error inside a 161-million-job labor market. The displacement Schulman is describing has not shown up in BLS data. Yet.
Both views are reconcilable inside a single sentence: AI displacement is real, layoffs are accelerating in the firms most exposed to language-model-replaceable work, and the macro labor market is being held up by sectors that AI hasn’t reached yet — health care, construction, hospitality, state and local government. When the macro tide goes out — recession, demand shock, fiscal contraction — the AI-exposed sectors will not have the cover to absorb their own restructuring quietly. That is the moment Bloomberg is describing and Schulman is pricing in.
The obituary homework, taken seriously
LostJobs has covered enough corporate-rationalized AI displacement to know the script: framing memo, retraining fund, sympathetic press tour, layoffs in tranches over four to six quarters. The Schulman variant is not a different script. It is the same script delivered with unusual candor at the front, a level of candor that other Fortune 100 CEOs are explicitly not willing to match.
The obituary exercise is the part that tells you the candor is intentional. A CEO does not give 99,600 employees that homework assignment by accident. The exercise either:
- Forces individuals to internalize the speed of change so they self-select into reskilling. Schulman’s stated intent.
- Creates a paper trail of the company is being honest about AI risk, useful when the next 10,000 cuts arrive and the WARN notices need a defensible narrative.
- Both.
What LostJobs is watching
Three datapoints will determine whether Schulman is the leading or trailing indicator:
- Q2 Verizon earnings, mid-July. If Verizon is meaningfully ahead of AT&T and T-Mobile on operating-margin improvement, the rest of the telecom sector follows the playbook within ninety days. If Verizon is in line, the AI-first thesis takes a year longer to mature.
- Whether other Fortune 100 CEOs match the $20M retraining fund. Schulman is asking them to do this on the record. None has, as of this writing. If a peer commits in Q2 — most likely candidates are Microsoft, Salesforce, or one of the Big Banks — the public narrative shifts.
- Whether the next BLS jobs report shows tech-sector unemployment ticking up materially. Challenger, Gray & Christmas data shows 37,638 AI-attributed cuts in Q1; the labor-force-participation rate has not moved. When it does, both Schulman and Bloomberg will be partially right at the same time.
A dry coda: the Verizon employees who actually completed the homework — wrote AI obituaries of themselves and read them aloud — reportedly described the experience as “clarifying” and “a lot.” Of those two reactions, the one that ends up in next quarter’s resignation letters is the second.