KPMG Bets Big On Generative AI-Fueled Productivity

KPMG has made the most explicit statement yet on how it expects generative AI (genAI) to improve its business. It joins Deloitte, EY, and PwC in stating its intent of over $1 billion investments in AI. KPMG expects that its $2 billion investment “through an expanded partnership with Microsoft” will generate $12 billion over that same period. That means KPMG expects AI to improve productivity dramatically.

This is a prime example of AI running into KPMG from the top-down door (one of four doors that genAI walks through into enterprises). KPMG will automate what the US CEO Paul Knopp calls “mundane tasks” in tax, audit, and consulting and empower employees with better tools (copilots) that make them more productive. What does this investment in AI and the Microsoft partnership mean for KPMG’s clients?

• You should expect steady improvements in work turnarounds. If a machine can ingest a file cabinet of tax information and present an accurate (and reviewed, of course) financial summary in hours rather than weeks or months, that should deliver real benefits for you.
• You should expect steady improvements in quality. If KPMG and Microsoft do this well, they will adopt a “layers, gates, and pipes” approach to building applications that incorporates the best AIs (layers of intelligence) while protecting the inputs and outputs (control gates) and creating new applications by stitching the AIs together (application pipes). That means they will be able to build AI-powered applications that empower employees and steadily improve them over time.
• You should expect KPMG to ask you to value the outcomes, not the labor. If someone is twice as productive, should you pay them twice as much? The old approach of time and materials pricing won’t handle this radical jump in productivity very well. Expect to learn how to price outcomes.
• You should expect a deeper relationship between KPMG and Microsoft. That means more Microsoft infrastructure and applications rather than Amazon or Google.