On May 14, Anthropic and PwC announced an expanded partnership. PwC is deploying Claude Code and Claude Cowork across its U.S. workforce, scaling toward hundreds of thousands of professionals globally, certifying 30,000 of its people on Claude, and launching a new business unit called the Office of the CFO, built on Anthropic's stack.
The business results they're citing are real: insurance underwriting that took ten weeks now takes ten days. Security work that took hours now takes minutes. Delivery improvements up to 70% across production deployments.
The surface read is "Big Four goes all-in on Claude." But that misses what PwC just told you: exactly which roles survive the next five years, and which ones don't.
The Hourglass Nobody's Talking About
Alongside the announcement, PwC published a workforce redesign piece — "No more pyramids: Rethinking your workforce for the agentic AI era" — that contains a piece of workforce architecture every CTO and hiring manager should screenshot and tape to their monitor:
As AI agents absorb more "midlevel" work, differentiation shifts to senior professionals who excel at strategy and innovation and to junior generalists who are AI-forward enough to act as agent orchestrators. The middle, the layer of people who execute on well-defined processes, compresses.
PwC calls it an hourglass — wide at the top where strategy lives, wide at the bottom where agent orchestration is becoming the new entry-level skill, narrow in the middle where most of the execution used to happen. That's a staffing thesis. And they're proving it in production.
"With a strong, capable base and a forward-looking leadership tier, connected by a lean, high-performing middle, your org chart looks like an hourglass."
— PwC, No more pyramids: Rethinking your workforce for the agentic AI era (January 29, 2026)
Think about what that means for a consulting firm. The associate who spent three years building Excel models for due diligence (the structured, learnable, repeatable work) is increasingly being replaced by an agent that runs the same analysis in thirty minutes. The senior partner who knows which question to ask the CFO before the deal closes is not going anywhere. The sharp junior who can write the agent prompt, quality-check the output, and escalate the right exceptions? That person is suddenly incredibly valuable.
The middle hollows out. The skill premium moves up and, counterintuitively, down.
What This Actually Is
The 2020s version of this story would have been: PwC is cutting headcount because AI is doing the work. That's not what's happening. PwC is certifying 30,000 people and standing up a joint Center of Excellence, not in one office but woven into a global bench across its U.S. practice and scaling from there. They're adding the Anthropic stack, the deal teams, the enterprise function redesign. This is not a cost-reduction story. This is a market capture story.
"The conversation around AI has shifted from possibility to execution. Clients are looking for ways to apply AI that are secure, responsible, and capable of delivering measurable outcomes in complex business environments. Our collaboration with Anthropic brings together advanced AI capabilities and PwC's industry experience to help organizations move from exploration to enterprise-wide impact with greater confidence."
— Paul Griggs, US Senior Partner and CEO, PwC
They are betting that in the next five years, enterprise clients are going to need someone who can design and operate AI-native workflows across finance, HR, supply chain, and security — and they're training their entire bench to do exactly that before their competitors figure out the playbook.
That's not defensive. That's predatory.
The Staffing Signal
Here's what this means for tech leaders building teams right now.
The same hourglass logic applies to your engineering org. The mid-tier engineers running defined ticket queues, the ones doing the kind of work Claude Code can pick up with a good spec and thirty minutes, are facing real compression. Not elimination. Compression. Their leverage changes. Their comp trajectory changes.
What you actually need, and what is genuinely hard to find, is a different profile on both ends.
On the senior side: architects and principals who can scope what the agents should be building in the first place — people who know when the AI's output is wrong in ways that aren't obvious, who have seen enough failure modes to catch the 3am mistake before it ships. Those people are more valuable than ever and harder to find than ever, and everyone knows it.
"PwC has been leading AI's expansion into the parts of the economy where accuracy and reliability are non-negotiable—financial services, healthcare, life sciences, cybersecurity—and the results are clear. Insurance underwriting that took 10 weeks now takes 10 days. Security work that took hours now takes minutes."
— Dario Amodei, Co-founder and CEO, Anthropic
On the junior side: AI-fluent generalists who can orchestrate agents, QA AI-generated work, write the prompts that actually produce useful output, and escalate with judgment. This is not a standard CS new grad. This is a specific mindset and a specific skill set, and most companies have no idea how to screen for it.
The firms that figure out how to identify and develop AI-fluent generalists while recruiting the principals who can direct them are going to run circles around the ones still debating whether to deploy Copilot.
PwC just told you the playbook. Run it or get lapped.