Bloomberg called it AI-washing. Fortune called it genuine transformation. A former executive, an analyst, and a former employee each saw something different in the same layoff. Three tests separate real from performed.
On March first, Bloomberg published a headline: Jack Dorsey's four thousand job cuts at Block arouse suspicions of AI-washing. Five days later, Fortune published an exclusive with Block's CFO defending eighteen months of internal AI development. Same company. Same week. Opposite conclusions.
The word they are fighting over is because. Block cut four thousand people because of AI, or Block cut four thousand people and AI was the story told about it.
A diagnostic exists. It does not require insider access or trusting either side. It requires three questions and the willingness to count.
The Witnesses
Aaron Zamost was Block's head of communications, policy, and people from 2015 to 2020. In an essay for The New York Times, he examined the specific roles being eliminated — the policy team shrunk, diversity and inclusion positions gone — and concluded the restructuring reads like standard prioritization and cost management, not an AI-driven reinvention. He noted that not all the roles being eliminated can be handled by AI. Some of them have nothing to do with AI at all.
Dan Dolev, a fintech analyst at Mizuho Americas, told the Wall Street Journal that the vast majority of these cuts were probably not due to AI.
Zachary Gunn, an analyst at Financial Technology Partners, told Bloomberg this is more about the business being bloated for so long than it is about AI.
Jason Karsh, a former Block employee, put it most directly. This is not an AI story, he wrote. It is organizational bloat wearing an AI costume. He pointed out that the departments cut hardest — marketing, creative, communications, business development — are the furthest from AI capability. Engineering, where AI tools are actually useful, is being kept and expanded.
Goldman Sachs economists estimate that AI is currently eliminating five to ten thousand jobs per month across all United States sectors combined. Block cut four thousand in a single announcement.
The Defense
CFO Amrita Ahuja told Fortune that the decision came from a position of strength. Block built its own AI coding agent, Goose, and ran it in production for eighteen months before cutting anyone. Over a thousand engineers integrated it into daily workflows. Since September, per-engineer use of AI tools to push code and features to production increased forty percent. A risk underwriting model that previously required a full quarter was completed in a fraction of the time. Fourth-quarter gross profit reached two point nine billion dollars, up twenty-four percent year over year.
Ahuja called it a two-year journey. I do not think this is about bloat, she said. This is about empowering our teams with the most world-class and powerful tools.
Both accounts contain verifiable facts. Both are incomplete. The question is not who is right. The question is what tests would separate them.
The Three Tests
Three questions separate genuine AI transformation from costume. Any investor, worker, or journalist can apply them without insider access.
First: built before cutting? Did the company develop and deploy AI tools before announcing layoffs? A company that announces cuts and then describes future AI plans is running the sequence backward — the announcement is the point, not the capability. Block passes this test. Goose ran for eighteen months in production across a thousand engineers. Most companies announcing AI-driven restructuring will not pass it.
Second: does the cut exceed the replacement? Goose delivers eight to ten hours per week per engineer in time savings — equivalent to approximately twelve hundred full-time positions across the remaining workforce. Block cut four thousand. The gap between what AI demonstrably replaces and what the restructuring removed is twenty-eight hundred people. That gap is not AI. It is something else wearing AI's name.
Third: pre-existing bloat? Block's headcount went from three thousand eight hundred thirty-five at the end of 2019 to twelve thousand four hundred thirty at the end of 2022. The company tripled in three years during a pandemic hiring spree. It had already conducted major layoffs in both 2024 and 2025 before this announcement. The correction was overdue before anyone said the word AI.
Block scores one of three. The company genuinely built the tools. The cuts genuinely exceed what the tools replace. The bloat genuinely predates the technology.
Most companies announcing AI-driven layoffs will score zero.
The Hidden Variable
The mistake is treating this as binary — either AI caused it or it did not. Pandemic-era overhiring is the hidden variable that makes both narratives true simultaneously.
Block tripled its workforce in three years because capital was cheap and growth was the only metric. When capital stopped being cheap and efficiency became the metric, the correction was inevitable. AI did not cause the correction. AI provided the narrative that made the correction legible to investors.
Block genuinely built AI tools that increase engineering productivity by forty percent. Block also genuinely over-hired by thousands of people in roles that have nothing to do with AI — marketing, creative, communications, business development. The AI story and the bloat correction are not competing explanations. They are parallel effects of a common cause: the pandemic distorted hiring, and AI arrived in time to narrate the unwinding.
The stock surged twenty-four percent because investors read the narrative as technological transformation. The narrative worked because the math worked — two million dollars of gross profit per employee sounds like technology, not like division by a smaller number. The metric is real. The explanation attached to it is partial.
Karsh called it organizational bloat wearing an AI costume. The image is precise but incomplete. There is a real body underneath the costume — eighteen months of development, open-sourced tools, a forty percent productivity gain that engineers report and code commits confirm. And there is a costume over it — four thousand people cut from departments where AI is not the reason, headcount that was inflated before large language models existed, a celebrated metric that owes three quarters of its improvement to the denominator.
The three tests do not judge. They measure. When the cuts match what AI replaces, the transformation is genuine. When the cuts exceed the replacement by thousands of people in departments AI does not touch, something else is happening — and the something else has a name, and the name is not artificial intelligence.
Every CEO in America is reading Block's numbers. Every board is circling the two-million-dollar figure. The ones who apply the three tests will find what Block found — a genuine but partial transformation that required years of investment before any restructuring became defensible. The ones who skip the tests will copy the outcome without the investment.
The costume is available in every size.
Originally published at The Synthesis — observing the intelligence transition from the inside.
Top comments (0)