Meta, PwC and Jack Henry show how to decide where to bet on AI when the board demands returns

For years, it was enough to prove that the company was doing something with generative AI to justify budget. That phase, according to a report published by The Information on November 13, 2026, has come to an end. In a conversation bringing together leaders from PwC, Jack Henry & Associates and Meta Platforms, the question stopped being who is using AI and became which use pays for itself.

For those who found or lead technology teams, the shift is not semantic. It changes the criterion by which AI budget is approved, what product teams need to measure before asking for more funding, and, above all, what to do with the time AI actually frees up within the operation.

From freedom to experiment to the demand for ROI

Within Meta, Mark Zuckerberg treated 2023 as the year of mandatory generative AI adoption across the entire company. The internal guidance, according to Heidi Rhodes, global director of corporate GTM and sales enablement at the company, was simple: test everything and accept failure as part of the process.

He basically said, 'This is going to be the year that everyone is going to adopt AI.' We told our employees, 'Figure out what works and go fail.'

Heidi Rhodes, global director of corporate GTM and sales enablement, Meta Platforms

That unrestrained period of token consumption helped Meta map which workflows actually benefited from generative AI, but it came at a price that the executive herself acknowledges is not replicable on every budget.

In 2023, it was: Any single problem that you have, rub some AI on it. Then we got our token bills. Now organizations are saying, 'Unless you can show me real ROI, we can end the conversation there.'

Heidi Rhodes, global director of corporate GTM and sales enablement, Meta Platforms

Jack Henry & Associates, a technology provider for banks and credit unions, never had that loose experimentation phase. Because it operates in a regulated industry, the company started at the opposite end: it first defined which AI tools would be approved and which data each employee could access, according to CFO and treasurer Mimi Carsley.

At PwC, the bottleneck was never a lack of ideas, according to Karen Young, chair of the firm's U.S. partner board and sales leader: employees already arrive with AI use proposals, and the approval filter remains the same one that existed before AI, which is the client's real need.

The per-head price the CFO won't let slide

At Jack Henry, giving a Microsoft Copilot license to each of the company's 7,300 employees would cost around $3 million a year, according to Carsley. It's this kind of number, multiplied by tool and by head, that has made AI budget stop being treated as a fixed expense and turn into a line item under constant audit.

The company's response was to regularly reassess individual license usage, not just distribute and forget.

If someone's not using their license, we'll go talk to them and coach them up. If they're still not using it, we actually seize it back and redeploy it elsewhere in the organization.

Mimi Carsley, CFO and treasurer, Jack Henry & Associates

At PwC, spending control took a different path: instead of reducing the number of licenses, the firm invested in training employees to choose the right tool and model for each task, according to Young. The logic for a startup founder is the same, just on a smaller scale: the cost of a generative AI license sitting idle on someone's account who doesn't use it is waste just as real as any idle cloud computing line item, and deserves the same periodic audit process.

What to do with the time AI gives back

The most debated part of the conversation, and the one most relevant to those deciding where to invest, was what happens after generative AI cuts the time a task takes. Young cited the case of a PwC client that used AI to transform 14 financial processes, reducing the budget forecasting and scenario planning cycle from weeks to minutes.

The question the executive asks to decide whether this kind of gain justifies more investment isn't about speed: it's about what the finance team now does with the freed-up time.

The real question is: What are those employees in finance now doing to be more strategic with their time?

Karen Young, chair of the U.S. partner board and sales leader, PwC

Carsley takes this reasoning to the opposite extreme, that of the small everyday task, to show that the principle doesn't change with scale.

If someone's gonna use AI to draft an email and you save five minutes, like, did you go get a cup of coffee or did you do something with the five minutes? If you're not redeploying that savings to another productivity or efficiency, did you really gain anything by the spend?

Mimi Carsley, CFO and treasurer, Jack Henry & Associates

For a founder, this question applies just as much to a three-person team as to a 300-person one: time freed up by AI that isn't redirected to another deliverable just becomes individual comfort, not the return on investment the board will demand at the next budget meeting.

Cutting headcount or multiplying results

The most delicate point of the conversation was about what to do when AI actually makes a team more productive. Rhodes cited the case of a sales client that became 20% to 30% more productive after AI took over research hours that used to be done manually. The market's most common reaction, according to her, is to interpret that gain as justification for reducing headcount.

Rhodes advocates for the opposite path: using the productivity gain to expand the operation, not to shrink the team.

Do you make a reduction in humans? Or now that you have one of the world's most efficient salespeople, do you go out and hire more?

Heidi Rhodes, global director of corporate GTM and sales enablement, Meta Platforms

For early-stage founders, this decision has an extra component that doesn't appear in companies the size of Meta, PwC or Jack Henry: limited cash. The right question isn't moral, it's financial. If the productivity gain opens room to grow revenue with the same headcount, hiring more makes sense. If the gain only reduces cost without opening a new revenue front, the decision to cut isn't wrong, it just needs to be made with clarity about what is being traded off.

The question that decides where to bet

The most useful conclusion of the report for those who approve AI budgets isn't about tool, model or vendor. It's about pre-existing competitive advantage. Carsley sums up the criterion she uses to decide where Jack Henry places its bets:

Ask yourself, what do you have an advantage in? Then apply AI to that as an accelerant. Do the thing you do best as an organization, then turn it up 10% to 20%.

Mimi Carsley, CFO and treasurer, Jack Henry & Associates

This reverses the logic that dominated the last three years, of applying generative AI wherever it seemed technically most impressive (automated customer service, content generation, code copilots) and hoping the ROI would show up later. The criterion that PwC, Meta and Jack Henry describe is the opposite: first identify where the company already beats the competition, and only then ask whether AI accelerates that advantage or just adds token cost without changing the outcome.

For those deciding the next AI budget cycle, the questions left by the report are more useful than any tool checklist: which license is sitting idle and unused, which internal process got faster and where the freed-up time went, and whether this quarter's productivity gain is being reinvested in growth or just absorbed as slack. Whoever can't answer all three is still in the adoption phase, not the return phase.