Token economics, physical AI, and beyond: McKinsey’s Chris Smith previews what to expect at NVIDIA GTC
Every year, NVIDIA GTC brings together leaders across industries to talk about AI. But this year’s gathering signals something bigger: a shift in the conversation itself.
Themes such as token economics, agentic AI, and physical AI signal where the industry is heading next. For organizations making long-term bets on AI, understanding that direction is critical.
As GTC gets underway in San Jose, California, we spoke with Chris Smith, a McKinsey partner who coleads our NVIDIA alliance work with clients. He talks about what this moment means for leaders—and shares insights from ongoing conversations with NVIDIA CEO Jensen Huang and the team through McKinsey’s NVIDIA alliance.
McKinsey: GTC has evolved significantly over the past few years. From your perspective on the ground, what makes this event unique compared with other major tech conferences?
Chris Smith: One of the things that makes GTC different in general is the breadth of the ecosystem in one place. You see CXOs, start-ups, hyperscalers, enterprises, infrastructure players, and software companies all grappling with similar questions about how to scale and operationalize AI.
But what’s changed this year is the economic framing. In recent conversations, Jensen has been very clear: This is not about isolated “AI projects” or incremental change. It’s about the emergence of what he calls “AI factories,” where data and electricity go in, and tokens and insights come out.
That reframing matters. Historically, data centers were cost centers. In this model, they are production facilities. Compute is directly tied to revenue. As Jensen has put it bluntly: Without compute, there are no tokens; without tokens, there is no revenue.
Finally, there’s a signaling effect from this annual conference. The insights to come out of NVIDIA GTC—whether token economics, physical AI, or sovereign AI—tend to show up in boardrooms very quickly.
McKinsey: You colead our NVIDIA alliance at McKinsey. What are some ways we’re working together across industries that are particularly exciting?
Chris Smith: At its core, our work with NVIDIA is about helping our clients understand how AI will transform their companies, and even more critically to help them deliver the impact with it. NVIDIA has a unique position to help them understand the direction of what is happening through not only their hardware but their software.
That has profound implications.
One of our most inspiring collaborations is in life sciences, where we’re working to accelerate drug discovery. What’s changed is not just model quality, but the economics of simulation, which requires large-scale compute in short bursts. That not only changes the economics of drug discovery, which means more R&D spend goes to drugs that will help people, but it also changes the speed at which drugs can get to market to help people.
In automotive and industrial sectors, the shift toward physical AI is particularly exciting. We’re helping clients use NVIDIA Omniverse™ to build digital twins of manufacturing facilities and optimize physical spaces.
I’m also inspired by the work we did in partnership for Toshiba Tec, which allowed our client to take the millions of data points they have and turn them into an engine for informed decision-making.
How Toshiba Tec and McKinsey are turning retail data into real-time decisions with NVIDIA
Across all of these examples, the common theme is moving from experimentation to production. Our clients are not stopping at pilots, but moving beyond to see real impact across the organization.
McKinsey: You’ve attended GTC for several years. What excites you most about this year’s event—and this phase of AI?
Chris Smith: We never quite know everything that will be revealed, but that’s part of the fun. I expect a few themes this year, particularly around agentic AI, physical AI, AI factories, and sovereign AI.
Every year that goes by, it becomes increasingly evident that we’re in the early stages of a new industrial revolution. Jensen is clear that this is not a moment for incremental ROI thinking. In fact, he openly challenges the idea that it is necessary to wait for fully provable returns before acting. He argues that in this moment the imperative is unmistakable.
That mindset is important. Many organizations are still treating AI as a pilot initiative. But if your company runs on software—as many do—then your operating model, cost structure, and organization design must evolve as that software evolves.
The continued adoption of agentic systems is also inspiring. It is changing our interaction pattern with software, from asking questions to issuing commands, and what we’re seeing is the transition from software being thought of as tools to software being thought of as skills that we integrate in our day-to-day work. That shift has a profound impact on how enterprises need to think about managing what are effectively two workforces: carbon-based and silicon-based, each with different management and improvement loops.
Importantly, this does not mean jobs—specifically engineering roles—disappear. In fact I prefer to think about this as a new age for engineers who are valued for their core skill of problem-solving and therefore value creation. This is in stark contrast to the traditional view of technology and software development teams in enterprises as a cost base.
As always, GTC is simply amazing for connecting with friends old and new. I’ve made connections here every year and we’re hosting some events for clients with panels including leaders from NVIDIA and clients who are leading real change in their industries.
