McKinsey CFO Reveals AI Costs and Talent Shifts in New Podcast

McKinsey & Company CFO Yuval Atsmon revealed in a Reuters podcast that the consulting giant is aggressively deploying AI internally while managing soaring infrastructure costs and unpredictable token bills. Atsmon argued that AI elevates the value of generalists over specialists, challenging the narrative of mass job displacement. The firm is building proprietary AI platforms and internal chargeback models to control expenses, reflecting broader enterprise struggles with generative AI adoption.

July 14, 2026, Inside AI — McKinsey & Company is navigating a dual reality: aggressively deploying artificial intelligence internally while guiding global clients through their own AI transformations. In a recent episode of Reuters' The Big View podcast, CFO Yuval Atsmon offered a rare, candid look at the financial and strategic calculus behind the consulting giant's AI push. Atsmon, who oversees the firm's technology investments, laid out the core tension: AI's promise of productivity gains is tempered by soaring infrastructure costs, unpredictable token bills, and a shifting talent landscape. The conversation with host Peter Thal Larsen, a Reuters Breakingviews columnist, peeled back the curtain on how one of the world's most influential professional services firms is betting on generative AI. The podcast, released as part of the series that also explored OpenAI's business conundrum, highlighted McKinsey's role as both practitioner and advisor. Atsmon emphasized that the firm is using AI to enhance its own operations—from automating internal knowledge retrieval to accelerating analysis—while simultaneously building the playbook for clients across industries. "We are seeing a step change in the speed at which we can synthesize information," Atsmon said, describing how large language models are reshaping the traditional consulting workflow. But he was equally frank about the financial discipline required. Token-based pricing models for AI services can spiral, demanding rigorous governance to avoid budget overruns. The discussion turned to workforce implications, a topic that has rattled the professional services sector. Atsmon pushed back on the narrative of mass job displacement, instead arguing that AI elevates the value of generalists—professionals who can connect disparate domains and apply contextual judgment. This marks a departure from the decades-long trend toward hyper-specialization. The Hidden Costs of Enterprise AI Adoption Beneath the surface of McKinsey's AI narrative lies a complex web of infrastructure decisions. The firm has invested heavily in proprietary AI platforms, but Atsmon acknowledged that managing consumption-based costs remains a moving target. Unlike traditional software licensing, generative AI expenses fluctuate with usage intensity, creating budgeting challenges that few enterprises have fully solved. Industry analysts note that McKinsey's experience mirrors a broader corporate struggle. A 2025 survey by Gartner found that 67% of large enterprises exceeded their AI budgets in the first year, primarily due to underestimated inference costs. Atsmon's comments suggest McKinsey is tackling this head-on through internal chargeback models and strict approval workflows for high-volume AI queries. The podcast also touched on the competitive dynamics of AI consulting. While McKinsey has publicly partnered with major AI labs, Atsmon hinted at a more nuanced strategy: building internal capabilities that reduce reliance on any single vendor. This aligns with the firm's historical playbook of absorbing cutting-edge methodologies and scaling them across its global network. Generalists in the Age of Algorithms Perhaps the most provocative insight was Atsmon's take on talent. As AI automates routine analytical tasks, he argued, the premium shifts to professionals who can ask the right questions, challenge assumptions, and integrate insights across silos. "The generalist is making a comeback," he said, pointing to McKinsey's own hiring trends. This view challenges the prevailing industry wisdom that AI will hollow out middle-skill roles while rewarding deep technical expertise. Instead, Atsmon's forecast suggests a renaissance for the liberal arts mindset within corporate strategy. It also raises questions about how business schools and professional training programs must adapt. The conversation with Larsen, recorded amid ongoing market volatility and geopolitical uncertainty, underscored the urgency of AI adoption. Yet Atsmon cautioned against moving too fast without guardrails. He cited examples of clients who rushed to deploy chatbots without adequate testing, only to face reputational damage from hallucinated outputs. McKinsey's own journey reflects this balance. The firm has rolled out an internal AI assistant, codenamed Lilli, which now handles over 70% of initial research queries for engagement teams. But Atsmon stressed that every output is reviewed by human consultants, preserving the advisory rigor that commands premium billing rates. Looking ahead, Atsmon predicted that AI will compress the timeline for strategic insights from weeks to hours, but will not replace the trust-based relationship between advisor and client. The podcast, produced by Ujjaini Dutta and edited by Oliver Taslic and Aditya Srivastav, is available on Apple, Spotify, and the Reuters app.