OpenAIとPwCが提携、CFO業務の再構築へ OpenAI and PwC collaborate to reimagine the office of the CFO

OpenAI and PwC have announced an expanded collaboration aimed at reshaping the office of the CFO with generative AI, applying tools such as ChatGPT and Codex to core finance and accounting workflows. The move is notable because it targets one of the most conservative and tightly regulated functions inside large enterprises, where AI adoption has so far lagged behind sales, marketing and software engineering.

Under the partnership, the two companies plan to embed OpenAI's models into processes overseen by chief financial officers, including financial reporting, budgeting and forecasting, expense analysis, contract review and audit preparation. The goal, according to the companies, is to move beyond pilot deployments of chatbots and toward AI-native operating models for finance teams. PwC will contribute industry knowledge, controls expertise and implementation capacity, while OpenAI provides the underlying models and developer tooling.

The collaboration builds on PwC's earlier large-scale rollout of ChatGPT Enterprise to its own workforce, which made the firm one of OpenAI's most visible enterprise customers. PwC has been using the technology internally for consulting, tax and assurance work, and the new agreement appears designed to extend that experience outward to client finance organizations. Codex, OpenAI's coding-focused agent, is positioned as a way to automate the scripting, data transformation and system-integration tasks that typically sit between ERP platforms, spreadsheets and reporting tools.

The timing reflects mounting pressure on CFO organizations. Finance departments are contending with persistent talent shortages, growing data volumes and rising compliance costs tied to disclosure regimes such as ESG reporting and evolving tax rules. Tasks like drafting journal entries, preparing disclosure narratives, reviewing contracts and running scenario analyses have long been flagged as well suited to large language models, since they combine structured data with substantial natural-language work.

Competition in this segment is intensifying. Microsoft is pushing Copilot into Dynamics 365 Finance and Excel-based workflows, Workday has introduced a series of AI agents aimed at finance and HR processes, and SAP is embedding its Joule assistant across S/4HANA and analytics products. Specialist vendors and Big Four rivals, including Deloitte and EY, have announced their own generative-AI alliances. Against that backdrop, CFO technology is shaping up as one of the more contested enterprise AI battlegrounds heading into 2026.

For OpenAI, the deal reinforces a broader strategy of positioning Codex and the Assistants and Agents APIs as a foundation for automating back-office systems, not just as developer productivity tools. Enterprise revenue has become an increasingly important pillar for the company alongside its consumer ChatGPT business, and high-visibility partnerships with global professional-services firms help validate that pitch to risk-averse buyers. PwC, in turn, gains a differentiated story to take to CFO clients that have been cautious about deploying generative AI in financial close and reporting processes.

The harder questions are likely to surface during implementation. Finance workflows demand traceability, segregation of duties and auditability, and regulators in the United States, European Union and Asia are still defining how AI-assisted outputs should be documented and controlled. Hallucinations, model drift and data-leakage risks all carry sharper consequences when the output feeds into statutory filings or audit evidence. PwC's governance and controls heritage may help here, but the partnership will ultimately be judged on whether it can deliver workflows that external auditors and regulators are willing to accept.

Neither company has disclosed detailed financial terms, customer commitments or a timeline for specific product releases. Early deployments are expected to focus on areas where the risk profile is more contained, such as internal management reporting, analyst productivity and contract analytics, before extending into closer-to-the-ledger tasks. If the collaboration succeeds in producing reference architectures that combine OpenAI models with PwC's control frameworks, it could accelerate a broader template for how regulated functions adopt generative AI.