Most layoff headlines describe a company in trouble. This one describes the opposite, and that is what makes it worth a careful read for anyone deciding how to build an AI-capable team.
On July 28, 2026, Visa reported its fiscal third quarter and disclosed a workforce reduction in the same breath as record numbers. Net revenue came in at $11.6 billion, up 14% year over year. GAAP net income was $5.6 billion, or $2.97 per share. The company processed 71.7 billion transactions, up 10%. Then, on the earnings call, management explained a $563 million severance charge and a plan to eliminate roles, with the heaviest cuts falling on the teams that build Visa's own technology.
If you have felt behind on AI, this is a useful case to sit with. The pressure to reorganize around AI is no longer limited to companies trying to survive. It now reaches profitable, growing firms operating from strength.
What Visa actually said
Two executives described the change in their own words on the July 28 call.
CFO Chris Suh explained the charge plainly: "we had $563 million in severance costs related to changes to our workforce." The stated purpose was to drive efficiency across the company and reinvest the savings in the highest-potential growth areas.
CEO Ryan McInerney was direct about where the reductions fall: "we are eliminating roles, with the majority being in our technology and product teams." He tied it to positioning the company for future growth.
The severance charge Visa disclosed alongside record revenue in its fiscal Q3 2026 report.
Source: Visa Q3 Fiscal 2026 SEC earnings release, 2026The $563 million severance figure comes straight from Visa's SEC earnings release, which describes it as costs "resulting from actions taken to drive operational efficiencies and reinvest in high-growth opportunities."
A note on the number you may have seen in other coverage: several outlets reported a cut of roughly 7% of the workforce, or about 2,600 roles, based on an internal staff memo from McInerney. That percentage and headcount were not published in Visa's SEC earnings release. Against the company's fiscal 2025 employee count of about 34,100, a reduction of roughly 2,600 roles is in the same range as the reported 7%. Treat the exact figure as reported from an internal memo, not as a Visa-published statistic. What is first-party and firm is the $563 million charge and the direction of the cuts.
The part leaders should study: the team redesign
The mechanism here is concrete, and that is rare. McInerney did not wave at "AI is coming." He described a specific change to how work is organized.
He said Visa is "reforming our product development teams that used to be 10 or more into smaller and more nimble agentic squads of two to four." He then attached results to the change: "80% more code commits" and "65% plus faster feature development."
Read that again as an operating decision, not a tech decision. A ten-person product team became a two-to-four-person squad working with AI agents. The output went up, not down. The organizing unit of the work got smaller while the work itself got faster.
A ten-person product team became a two-to-four-person squad working with AI agents. The output went up, not down.
That is the destination a lot of companies are heading toward, whether they have named it or not. For a leader, the question is how small and how capable an effective team becomes when each person operates with AI, and whether you are building people who can work that way.
Why the cuts concentrated on the technical teams
It is worth pausing on where the reductions concentrated. The majority fell on technology and product staff, the people closest to the tools. A company's most technical employees are not automatically the safest when the company reorganizes the work itself.
The honest read, and the one that keeps this in the guide-not-guru spirit, is that AI was part of the reason and not the whole story. Visa also said it wants to redeploy savings into growth. AI acted as an accelerant of a broader efficiency-and-reinvestment plan, not a single cause that erased jobs on its own. Overclaiming the AI angle would misread what management described.
Where this maps on The 7 Levels of AI Proficiency
The Visa example is a clean illustration of what higher proficiency looks like in practice.
The 7 Levels of AI Proficiency describes a climb from first awareness of AI tools up to orchestrating whole systems of work. The early levels are about individual capability: learning to prompt, learning to check the output, learning to build a repeatable workflow. The upper levels are about leadership and integration, where a person designs how AI, data, and people fit together across a team.
A squad of two to four people working with AI agents is that upper end made visible. It only works if the people inside it operate at a high level: they hand real work to AI, they know how to verify what comes back, and they organize the whole flow rather than run a single tool. A two-person squad producing more than the old ten-person team is a story about more capable people supported by AI, and about a leader who redesigned the work to let that happen.
For a professional who feels behind, the invitation is straightforward. The advantage is going to the people who learn to work inside small, AI-leveraged teams, and to the leaders who know how to build and measure that capability. That skill is learnable, and it is the same skill Visa just paid a $563 million charge to move toward.
What to do with this
You do not need to cut anyone to take the lesson. A few starting questions for the team you already have:
- Where does a smaller AI-supported team already beat a larger manual one? Pick one workflow and look honestly. That is where the redesign starts, the way Visa started with product development.
- Can your best people verify AI output, or only produce it? The squad model depends on that skill. Level up the checking, not just the generating.
- Are you measuring capability, or assuming it? Visa attached numbers to its change. You can measure where your team sits and where the next level is.
A practical next step: take The 7 Levels of AI Proficiency assessment to see where your team stands today, then decide which one workflow to redesign first. The goal is to build the kind of team that a two-to-four-person squad becomes when the people in it are ready.
Related reading: Level 6: The Admiral (Systems Integrator).
Sources
- Visa Inc. Q3 Fiscal 2026 Earnings Release (SEC)
- Visa Inc. Fiscal 2025 Form 10-K (SEC)
- Visa Q3 2026 Earnings Call Transcript (third-party transcript of Visa's July 28, 2026 call)
Frequently Asked Questions
Did AI cause the Visa layoffs?
Management framed AI as part of a broader efficiency-and-reinvestment plan, not the sole cause. CEO Ryan McInerney described restructuring product teams into smaller "agentic squads," and CFO Chris Suh described a $563 million severance charge aimed at driving efficiency and reinvesting savings in growth. AI acted as an accelerant of that strategy.
How many jobs did Visa cut?
Visa disclosed a $563 million severance charge in its SEC filing but did not publish an exact headcount there. Several outlets reported roughly 7%, or about 2,600 roles, based on an internal staff memo. Treat that figure as reported from the memo rather than as an official Visa statistic.
Was Visa struggling financially?
No. In the same fiscal Q3 2026 report, Visa posted net revenue of $11.6 billion, up 14% year over year, GAAP net income of $5.6 billion, and 71.7 billion processed transactions, up 10%. The reductions came alongside record results.
What is an "agentic squad"?
It is the term Visa's CEO used for product teams that used to have 10 or more people and were reorganized into two-to-four-person groups working with AI agents. Visa reported 80% more code commits and over 65% faster feature development from the change.
Find your AI Proficiency level
The free 7 Levels assessment places you across seven stages of AI capability. Under ten minutes. Research-backed scoring.