How Block’s CFO became convinced the company needed only 60% of its staff

Good morning. Why would a profitable, fast-growing fintech cut nearly half its staff?

That’s the question swirling around Block after CEO Jack Dorsey announced plans to lay off 4,000 employees, about 40% of the company, just as the parent of Square and Cash App reported Q4 gross profit of $2.9 billion, up 24% year over year. I sat down with Block CFO and COO Amrita Ahuja, who said the move isn’t about trouble or “bloat,” but about how far the company has gone in reshaping its workforce around artificial intelligence.

“This is a two-year journey for us,” she says. “This was not an overnight decision.” Ahuja walks through how Block built and deployed its own AI agent—code-named goose—to sit on top of large language models, automate workflows, and accelerate software development. She explains why Block raised its 2026 guidance even as it cut thousands of roles, how she thinks about employee morale and “reskilling” the remaining workforce for an AI-driven future.

Ahuja also responds to skeptics who say AI is just a convenient label for old-fashioned cost-cutting. You can read more about my conversation with Ahuja here.

In the age of AI, companies like Block are navigating a new tension: the promise of massive productivity gains alongside the need to rethink how work is organized. Research from McKinsey & Company on generative AI and the future of work finds the technology’s largest potential in knowledge-heavy fields such as financial services and software engineering. Many core tasks in those jobs—coding, document synthesis, customer communication, and data analysis—can be automated or dramatically accelerated, creating large productivity gains if companies redesign workflows and retrain workers.

At the same time, research connected to the Massachusetts Institute of Technology and highlighted by MIT Sloan School of Management finds that when highly skilled workers use generative AI within its “sweet spot,” performance can improve by roughly 40%. But when the tools are misapplied, performance can actually decline—underscoring that AI tends to work best when paired with human judgment.

Adopting AI may be the easy part. Reorganizing work around it is the real challenge.
 
Have a good weekend.

Sheryl Estrada
sheryl.estrada@fortune.com

This story was originally featured on Fortune.com