Summary: On May 14, 2026, PwC and Anthropic announced an expansion of their strategic alliance. PwC will deploy Claude Code and Cowork to U.S. teams first, then scale to its global workforce. A joint Center of Excellence will train and certify 30,000 PwC professionals on Claude. Production deployments already show 70% delivery time reductions. PwC also maintains an existing OpenAI partnership — one of the first firms in professional services to run both AI relationships at enterprise scale. Part of our AI Industry Analysis coverage.
What Was Announced
On May 14, 2026, PwC and Anthropic announced an expansion of their strategic alliance, deepening how PwC uses Claude to build technology, execute deals, and reinvent enterprise functions for clients across its service lines.
The core commitments:
- Claude Code and Cowork rolling out to U.S. teams, expanding toward PwC’s global workforce of hundreds of thousands of professionals
- A joint Center of Excellence with a training and certification program for 30,000 PwC U.S. professionals
- Three strategic pillars: agentic technology build, AI-native deal-making, and reinvention of the enterprise function
This is an expansion of an existing relationship, not a new one. PwC has been using Claude in production. The announcement formalizes the institutional structure and scales the deployment.
What “Expansion” Means Here
The distinction between a procurement relationship and a strategic alliance matters in professional services.
Most enterprise AI deals are licensing agreements: a firm buys seats, runs training, and professionals use the tool when they want to. The joint Center of Excellence structure here is different. PwC and Anthropic are building a shared roadmap, co-developing delivery methodology, and certifying professionals through a structured program — not just distributing software access.
The 30,000 certification target extends earlier work: PwC had already run hands-on sessions with more than 5,000 partners and senior leaders. The new program scales that institutional capability down through the professional organization, not just at the leadership level.
Claude Code and Cowork are the two products PwC is centering the rollout around:
- Claude Code — the AI-native software development environment that writes, reviews, and modifies code alongside developers and non-developers. For a professional services firm, this means accelerating the technology build portion of client engagements without requiring full software team deployments.
- Claude Cowork — the collaborative agentic workspace that lets teams work alongside Claude on complex multi-step tasks. For consulting engagements, this compresses the document-heavy, multi-iteration work that typically drives consultant hours.
Production Results Already Documented
The announcement was not speculative. PwC cited production deployments across five verticals:
Insurance underwriting — Underwriting that took ten weeks now takes ten days. This is the most concrete ROI number in the announcement and covers one of professional services’ most document-intensive workflows.
Cybersecurity — Security work that took hours now takes minutes. The specific tasks cited involve the kind of log analysis, vulnerability triage, and incident documentation that historically required senior analyst time.
Mainframe modernization — Claude Code is accelerating the translation and rewriting of legacy mainframe code, a category of work that has historically been expensive and slow because of the scarcity of COBOL and RPG expertise.
HR transformation — AI-assisted restructuring of HR processes, including policy documentation, workflow automation, and organizational design support.
Professional sports operations — Claude is running in production in unnamed professional sports clients, likely for contract analysis, player performance analytics, or operations workflows. This is an unusual vertical for a Big Four firm, and the reference suggests PwC’s early deployments extended across client industries, not just into professional services’ own operations.
Delivery times across these deployments are down up to 70%.
The Finance Business Group
One of the specific new structures announced is a Claude-native finance business group within PwC.
The group combines PwC domain knowledge with the full Anthropic product surface: Claude in the productivity suite, Cowork, and Claude Code. The finance focus makes sense given PwC’s concentration in financial services, where it works with banks, insurers, asset managers, and fintechs on audit, tax, and advisory mandates.
The “Claude-native” framing suggests this is not a retrofitted version of an existing team. It is a new organizational construct built around Claude as the primary tooling layer — a sign that PwC is betting the alliance is durable enough to organize people around, not just software to buy.
PwC Runs Two AI Partnerships
PwC entered a multi-year strategic partnership with OpenAI in 2024. That deal included significant seat deployments and integration with PwC client tools, and positioned PwC as one of OpenAI’s largest enterprise partners.
The Anthropic expansion announced in May 2026 does not replace that relationship. PwC is running both.
This dual-AI-partnership structure is becoming a pattern in large enterprises. Intuit announced separate deals with both Anthropic ($100M+) and OpenAI for different parts of its product strategy. PwC’s approach appears similar — using OpenAI and Anthropic for different purposes within the same organization, rather than consolidating to one vendor.
For enterprise buyers evaluating professional services firms on AI capability, PwC’s position is that it is model-agnostic at the partnership level and deploying whatever tool performs best for the task. Whether that flexibility holds as both partnerships deepen is a practical question: the joint Center of Excellence and certification program are structured around Claude specifically, which implies a growing institutional investment that will create switching costs over time.
Where PwC Fits in the Big Four AI Landscape
Five days after the PwC announcement, KPMG announced a global alliance with Anthropic giving all 276,000+ KPMG employees Claude access and embedding Claude Cowork and Managed Agents directly into Digital Gateway, KPMG’s flagship client delivery platform.
That sequential timing — PwC on May 14, KPMG on May 19 — illustrates the competitive pressure in professional services AI. Each announcement reinforces the other. KPMG claimed to be the first Big Four firm to embed a frontier model into its primary client delivery platform. PwC’s announcement preceded it by five days.
The Big Four AI landscape as of May 2026:
- PwC — Anthropic alliance (expanded May 14, 2026) running in parallel with OpenAI partnership (2024). Claude Code and Cowork central. 30,000 professionals to be certified.
- KPMG — Global Anthropic alliance (announced May 19, 2026). Claude Cowork and Managed Agents embedded in Digital Gateway. Named Anthropic’s preferred PE partner.
- Deloitte — AI partnerships with AWS, Microsoft, and Salesforce. Various integrations across service lines.
- EY — Microsoft Copilot integrations, proprietary EY.ai platform.
Two of the four are now running formal Anthropic alliances, both announced within five days of each other. The timing is not coincidence — it reflects both firms securing a position before the other could claim it, in a market where AI partnership announcements affect client perception of which firm is most capable.
Limitations and Open Questions
The PwC announcement has real substance — production results, a defined certification target, named products. But several things are not yet visible:
Global rollout timeline. The announcement says U.S. teams first, then “global workforce.” No dates are given for global expansion. PwC’s global headcount is in the hundreds of thousands; full deployment would be a multi-year effort.
The 30,000 certification scope. Thirty thousand professionals is a significant number, but it is a fraction of PwC’s total workforce. What the certification covers — basic Claude usage, Claude Code development, agentic workflow design — and how it differs from the 5,000-leader program that preceded it are not detailed.
The OpenAI relationship. How the two AI partnerships coexist in practice — which tools PwC recommends to which clients and for which tasks, and whether there is internal competition between the programs — is not addressed in the announcement.
Client visibility. The five production verticals are real but unnamed. Unlike KPMG Blaze, which is a named commercial product aimed at PE portfolio companies, PwC’s production deployments are internal case studies, not publicly branded products.
What This Signals for Enterprise AI
PwC’s expansion does two things that are worth tracking.
First, it confirms that the Big Four are treating AI partnerships as institutional commitments rather than software procurement. The Center of Excellence model — shared roadmap, joint certification, co-developed methodology — is a structural relationship that takes years to build and creates real switching costs. That is a different signal than buying seats.
Second, the dual-partnership model (Anthropic + OpenAI at the same firm) suggests the enterprise AI market is not converging on a single winner in professional services. The largest firms are hedging, running multiple AI relationships simultaneously. If that pattern holds, the long-term competitive question for AI companies is not who wins the Big Four client — it is who becomes the standard for the specific workflows where performance matters most.
For PwC, the 10-week-to-10-day insurance underwriting result is the kind of number that becomes a case study clients ask about. If that result generalizes across the insurance underwriting engagements PwC runs globally, the economic case for the Anthropic alliance pays for itself many times over.
ChatForest covers AI tools, platforms, and enterprise AI deployments. For related coverage, see our analysis of the KPMG–Anthropic global alliance and our review of Intuit’s five financial apps via MCP.