Is Block’s Decision To Layoff 40% of Its Workforce A Bellwether Or Not?

Last week Jack Dorsey, CEO of Block Inc. (makers of Square and CashApp) loudly announced plans to layoff 40% of its 10,000+ employees (4,000) and replace them with AI. The announcement was simple and there were few specifics about how this is going to take place.

LinkedIn shows around 13,000 employees with most in engineering, IT, operations, and sales.

I had one of my friends (who works at Anthropic) immediately tell me what a big story this is and why more and more jobs will be lost to AI.

Since I spend my days with senior HR leaders and have never seen anything like this before, I felt I should dig in a bit.

It’s Not So Simple

As I was looking for information about what Block Inc. is doing (I figured there must be a Superagent story here) I found an article by Business Insider, who interviewed seven employees and none of them was clear precisely how this business transformation was going to take place.

Then I read this. Jack Dorsey “loves AI.”

Is AI An Excuse for Performance Management?

Before I discuss the realities of AI in the workplace (it has not eliminated nearly as many real jobs as people think, yet it is eliminating many tasks), it’s clear in many companies that AI is an excuse for tightening performance management.

I don’t know Dorsey but I do remember the layoffs at Twitter after Elon Musk found enormous amounts of bureaucracy when he took over. It’s possible that in this case AI is a great excuse to clean up the org chart.

We actually did a huge study of org redesign around AI last year and through 70+ companies found that most companies who looked at AI as a tool to increase individual productivity did not find much job reduction. It takes re-engineering to see anything close to what Block is talking about. (We detail this in The Rise of the Superworker and our study on Dynamic Work Design.)

So in the case of Block, perhaps there is a big strategy here but maybe not.

I took some time to compare Block’s financials compared to Visa and Mastercard and Shopify (different businesses but somewhat comparable). What you see is that Block is far less profitable and has less than half the gross margin of the others, indicating that the company is not operating at “scale.”

So this AI move may really be a way to push for operating efficiency, especially since investors have not been happy.

What We’ve Learned About AI Transformation

We have been working with dozens of companies on these issues so let me share what we’ve learned.

1/AI Transformation to Eliminate Jobs Is A Short-Term Approach

If you’re “hoping” that AI will eliminate jobs you’re likely going to be disappointed. Not only is this only a one-time benefit, but it may not be the right goal.

Why? Because building, implementing, and scaling enterprise AI takes staff. Not only do you have to build the agents and superagents you need, the company will need staff to manage the systems. And that means training, hiring, and a new set of operations.

These operations include like model-building, training, verification, and checking that the code and interactions of your AI are correct. Remember this is a non-deterministic technology – it may do things you didn’t expect. Do you want CashApp to “sometimes” be correct and other times drift into errors?

(Read Jack Clark’s article about this for more.)

2/The cost of AI is significant.

Second, even if you could eliminate your online support center (I’m guessing many of these 4,000 jobs are in support) and build or buy agents to handle the workload, the compute costs can be high.

In a call center the typical savings, to replace a $35,000 outsourced agent, are about half to 60% of operating costs. So the AI “agent” costs $15,000-20,000 per year, assuming the AI agent can use voice. These numbers will go down, but you can see the service-delivery cost alone is significant. (details here.)

Add to this the cost of training the agent, maintaining the agent ecosystem, and governance of data and you see the “layoffs” are not a 100% replacement of human labor for AI labor.

Now if you consider this cost for coding, where the engineer may earn $100,000 a year but the AI generation is higher, the operating cost of the AI can be higher than that of a human engineer.

This is a fascinating pivot because, unlike the call center scenario, software engineering is a high-context, non-linear task. In 2026, the cost isn’t just “talk time”; it’s about the “reasoning tokens” required to understand a complex codebase.

If we assume 100 engineers earning $100,000 each, your annual payroll is $10,000,000. If AI can eliminate 80% of their manual coding time, you are effectively buying back $8,000,000 of labor capacity.

Here is Gemini’s analysis:

You can see what’s going on.

So even if Block (or another company) can find AI to outsource customer support and much of engineering, the capital and operating costs may continue to be 1/3 to 1/2 or higher than the cost of staff. (If you want to get into the guts of why data centers are expensive, pour through this.)

And that leads to….

3/The Real ROI of AI is Re-Engineering Process, Not Job Displacement

If you look at the 10-100X transformations from AI, none of them start with job displacement. They all start with business re-engineering.

Allianz has built digital twins to improve claims processing and share knowledge. Travelers is using AI to scale employee enablement and speed up time to market. Hubspot is building AI to offer new services to clients, not reduce labor. Our AI Galileo lets us scale our business by 100X or more to reach clients we simply could not talk with on the phone.

In the case of Block, it may start with a layoff but it won’t end there. If and when the company re-engineers its business around AI investors should see results.

And that leads me to this question. If AI is going to be used to re-engineer the business, why isn’t that happening already? Great companies are always looking at ways to increase talent density and avoid the “hire to grow” model which reduces employee productivity.

Let’s hope there’s a major re-engineering project at Block behind the scenes.

Lesson to You

My only point here is this: don’t take this announcement at face value. AI is not a “job eliminating” strategy – it’s an opportunity to re-engineer what you do. Let’s see if Block Inc. pulls that off.

Additional Information

Enterprise AI Architecture: Imperatives for 2026

New Research: How AI Transforms $400 Billion Of Corporate Learning

Webinar recording: Watch a replay of Josh’s walkthrough of the 11 essential imperatives HR & business leaders need to know for success and progress in 2026.

Josh Bersin Podcast: Listen in as Josh provides much-needed guidance for understanding the biggest HR transformation in decades.

Galileo Learn program: Complete The Superworker Organization: AI Goes Enterprise learning program, and discover the hands-on skills required to navigate the redefinition of work, HR teams, and organizations in the era of superworkers and superagents.