PwC's Global AI Head Explains Why Waiting For The Perfect AI Foundation Is A Strategic Error

In the enterprise AI race, the divide is widening between companies that are accelerating and those stuck in endless cycles of experimentation. Success now turns on a willingness to modernize legacy infrastructure and anchor initiatives to clear business outcomes, rather than chasing the newest tools. While many leaders believe they have a strategy, their organizations are operationally unprepared. This has created a market where innovation is outpacing adoption, often trapping teams in debate and analysis paralysis as they move from pilots to operational deployments.

Matt Hobbs is the US and Global Head of Cloud, Engineering, Data, and AI at PwC, where he has spent more than two decades helping enterprises translate technology investment into measurable business results. He led the creation of PwC's Microsoft business unit and has held leadership roles across digital transformation, cloud strategy, and AI alliance development. He described a market where even the most advanced organizations feel behind and explained why that pressure, if misread, becomes its own trap.

"Innovation is crushing adoption right now. If you froze innovation today, most companies would still be trying to figure out how to adopt what already exists. That paralysis is what’s holding them back," said Hobbs. The implication, he explained, is that waiting for the right moment is itself a strategic error. The clean data environment, resolved ERP, and finalized governance framework may never fully materialize. The companies separating from the field have stopped treating foundational gaps as preconditions and started treating them as parallel workstreams.

Hobbs’ point is that this paralysis often stems from focusing on the wrong problem. Instead of getting locked in evaluation cycles, he suggested leaders turn their focus to orchestration. This approach encourages leaders to unify fragmented workflows and modernize existing environments without letting them dictate the pace, overcoming the attribution challenges that can stall progress.

• The tool trap: "Companies are in this debate around what's the right tool, rather than just acknowledging that it's going to continue to change over any time horizon. Whatever tool choice they make is the wrong tool choice. You are more likely to be wrong on a tool choice than right because of the pace of innovation,” said Hobbs. “That seems to be the struggle.” And this isn't just theory. Hobbs pointed to concrete examples where this approach is already delivering value. To generate meaningful ROI, he advised anchoring AI initiatives to explicit business outcomes, noting that results are appearing in revenue acceleration through pricing personalization, margin expansion from taking cost out of core service delivery, and market expansion by using agentic infrastructure to profitably serve down-market customers.

• A choice, not a chain: This mindset, Hobbs noted, creates a clear dividing line. While some leaders pause their AI journey to fix foundational issues first, a path that leaves them unanchored from the ultimate business value, the accelerating companies push forward. Their success often comes from treating AI less like a series of tech projects and more like a single, top-down business initiative, simultaneously architecting an AI-native tech organization with a modular data architecture built for speed. "It's not like they don't have technical debt. They've just chosen not to make it a blocker. And they're addressing it as they go. They're driving it for a business reason that may happen to address the technical debt in the process," said Hobbs.

Hobbs reframed the goal: it’s not about being "AI-ready," but becoming "AI-native," where companies build true AI-native workflows into their operational DNA. In his view, this state is often characterized by a clear ambition tied to specific business outcomes, a modern and interoperable data architecture, and a commitment to embedding Responsible AI from the start. But within that framework, Hobbs stressed that two components are the most challenging to get right.

• The owner's manual: Hobbs was direct that AI transformation fails when it gets handed to a technical team without a business owner driving it. The architecture can be perfect and the tools in place, but if the functional leader isn't accountable for the outcome, the initiative drifts. "If you want to transform a finance function, the CFO has got to own that. That's not a technical problem. It requires a nonlinear rethinking of the workflow itself, prioritizing the machine's role first and the human's second. It’s about focusing on the reason for the process and ignoring all the steps in between," said Hobbs.

• Permission to self-destruct: "The hardest problem is workforce alignment and incentives. It comes down to this: if an employee sees a way to do something better, do they feel safe enough to actually do it? Or do they feel like they're creating problems for themselves? The idea that you have to be willing to disrupt yourself is true for any company that wants to succeed with AI," said Hobbs. Even organizations with strong technical foundations hit this wall, he noted. Culture is the variable that no architecture decision resolves.

His final advice for executive teams is twofold. First, he urged them to stop the cycle of “persistent, tech-driven POCs” by defining the specific business problem they intend to solve. Second, he told them to “step back from the steps within the process to the overall reason for the process.”

It’s a shift in thinking that, in his view, ultimately separates the companies accelerating with AI from those left spinning their wheels. “You are in a world of perpetual strategic ambiguity,” he said. “The thing that you just generated value on, you could redo and generate more value because innovation has already moved on before you released it.”