Business leaders told KPMG—and its recently anointed chair and CEO, Timothy Walsh—that they’re wrestling with uncertainty across several different areas of their work. This is well documented and is to be expected, Walsh told Fortune in an interview. “There’s this general, as you would expect, general conversation around business uncertainty,” Walsh said, adding that he was encouraged at least to see the “alignment” in terms of topics coming up in C-suite conversations.

Peeling back the survey data, Walsh revealed that an unsurprisingly sizable majority (89%) say tariffs will “significantly impact” their business’ performance and operations over the coming three years. And nearly as many, 86%, said their firm will increase prices as needed. They are working hard to get around this, with 85% saying their company will strive to shift its sourcing strategies to minimize the impact as much as possible. The landscape is so uncertain that nearly every CEO says they need to make some kind of change: 79% said they’ve adapted their growth plans.

Walsh talked to Fortune about uncertainty on tariffs and AI, and the importance of trust in a climate of such uncertainty. CEOs are concerned with another advancing technology with terrifying capabilities, Walsh said: cyber and quantum. “That honestly keeps me up at night.”

Cybersecurity’s quantum challenge

Cybersecurity risks remain elevated, especially as quantum computing approaches. As for advances in quantum computing, Walsh said it could one day soon be capable of breaking all encryption, and companies tell him that they’re doing full assessments. It’s a “massive effort” to ensure that they’re not exposed when that quantum computing capability arrives, Walsh warned.

Adding into the mix the capabilities of AI agents and, Walsh said, “in many cases, a nation-state-type investment,” he’s very concerned about malware and deepfake-type technologies escalating in danger. Over the next three years, 82% of CEOs polled said cybercrime and cyber insecurity was a top trend that could hurt their organization. Cyber risk was overall the second-highest cited pressure behind CEOs’ short-term decisions. CEOs are most concerned about fraud detection and prevention (65%) and identity theft (52%), but they also said they have plans in place to mitigate.

All that being said, Walsh said CEOs are “feeling optimistic because they see so many growth opportunities.” The economy has been surprisingly strong despite all the uncertainty, the tech sector is driving a very strong stock market, and he even noted some “large deals and transactions” are coming through when it comes to M&A. “Capital flows are starting to move and [be] a bit more liquid.”

Tariffs and the AI element

Walsh told Fortune that tariffs are obviously the number-one thing on every CEO’s mind. And it’s not only the fact of tariffs but potential changes to tariffs, and “the uncertainty around whether those tariffs will continue to change.” There’s an overwhelming need for businesses to not only consider what will change but to get agile enough to work on their supply chains to be prepared for future, still uncertain, changes to come. To that end, 34% of CEOs said in the survey that supply chain resilience is the top pressure driving short-term decisions, followed by cyber security risks (29%) and global economic uncertainty (25%).

Walsh emphasized that tariffs are introducing a multi-dimensional challenge for CEOs. “The CEOs I speak with are addressing tariff impacts in three areas: cost take-out, supply chain optimization including reshoring, onshoring considerations, and ultimately pricing.” He said KPMG is actively working with clients in all of those areas and yes, AI is part of this transformation, too. The prominence of AI is another layer of uncertainty being added to the picture, but Walsh said it’s helping a lot of CEOs: “AI is not just an efficiency play, CEOs are focused on innovating their business models and introducing new revenue streams and products.”

The AI hourglass to come?

Walsh said AI capabilities are changing quickly, and he acknowledged that companies are starting to restructure in response. The survey found that CEOs “mostly see an hourglass shape” to their organizations in next three years, Walsh said, noting that’s typical with every new technology deployment. He added that “no one knows exactly where [workforce shape] is headed … It’s a challenge to forecast as AI advances rapidly.” In the survey, 35% said they are planning for workforce reductions in some areas over the next two to five years due to AI, and 69% see an hourglass with higher numbers of senior leaders and early-career workers and fewer in the middle (another 16% said a vertical triangle, 13% a triangle and 2% an inverted pyramid).

Managers are facing new responsibilities, managing teams with integrated AI agents, for instance. Walsh said some CEOs describe teams with both people and AI agents on them, “and managers of those teams have to ensure [that] agents complete steps in the workflow process, that agents have good data inputs so that their outputs can be relied upon, and continuously review those outputs.” CEOs surveyed said 86% of them see AI agents becoming embedded team members next year, and half think managers will be primarily responsible for managing AI agents’ performance as opposed to, say, HR or IT.

Walsh agreed with Fortune‘s reporting that “human skills” still matter as AI implementation shows the necessity of reviewing AI outputs. “Human skills are critically important,” Walsh said. Even though KPMG invests in and spends time upskilling its workers on AI and providing them with tools and licenses, he said he continues to remind leaders that “human-to-human relationships are critical … both internally and externally. Trust is more important than ever. Building trust with our teams, clients and ensuring we can trust outputs of technology like AI.” Given the uncertain climate, he added, trust is at a premium. The top change that CEOs see coming is retaining and re-training high-potential talent (75%), followed by redesigning roles to reflect AI collaboration (65%) and hiring AI-capable talent (64%).