(co-authored with Joanna Doven)

A version of this article first appeared in Governing Magazine on October 31, 2024

The accelerating pace of generative AI is driving a more physical economic revolution, one that goes deeper than digital chatbots enhancing worker productivity. This opens the door for transformation in cities that embrace its potential.

We wrote recently about why Pittsburgh has the ingredients to be a major part of this mix. Its strengths were on full display at the AI Horizons Summit held on October 14th. Headlined by Pennsylvania Democratic Gov. Josh Shapiro, a standing-room-only crowd of 700 attendees lent the event an atmosphere of near-religious fervor, with Carnegie Robotics’ autonomous dog playing fetch at the exhibition hall.

The summit’s pinnacle came when Nvidia announced Pittsburgh as its first “AI Tech Community city.” A signing ceremony involving Shapiro, Nvidia’s Anthony Robbins, Carnegie Mellon University President Farnam Jahanian and University of Pittsburgh Chancellor Joan Gabel marked the beginning of a game-changing partnership.

The summit showcased a rapidly growing network of firms that are using AI to solve hard challenges in the physical world. This next stage in AI evolution, called physical AI, is where generative AI, robotics and simulation converge. Think of your Roomba, but exponentially more capable — machines able to tackle complex, real-world tasks, reprogramming themselves to real-time environmental changes. Instead of robots performing pre-programmed tasks with static code, digital-twin models enable robots to learn as they go, performing intricate tasks in challenging environments, from monitoring bridge safety in rugged terrain to assisting in complex surgeries and, of course, autonomous driving.

The Nvidia partnership exemplifies how Pittsburgh has rapidly become a veritable innovation playground for physical AI and market leaders. Nvidia’s commitment will embed solutions architects within the university ecosystem — namely in robotics — while empowering startups and scaleups with its in-demand chips.

Nvidia is not alone in Pittsburgh. Another firm, Gecko Robotics, integrates AI and robotics to transform how the world builds, operates and maintains its most critical infrastructure. Its distinctive-looking robots go where humans cannot, inspecting complex defense, oil and gas, power, steel and mining systems. SkildAI’s humanoids optimize computer infrastructure and streamline industrial processes, enabling companies to reduce costs and increase efficiency through intelligent task automation. And Hellbender, a pioneer in AI-enabled computer vision systems, manufactures on-edge technologies for defense and industrial applications with a focus on secure, American-made solutions, ensuring robust performance in high-stakes environments. Twenty-nine percent of its employees performing AI-based manufacturing are military veterans.

While Pittsburgh is poised to lead the new AI economy, with a projected compound annual growth rate of nearly 40 percent from 2024 to 2030, its lack of venture capital threatens to limit the city’s ability to fully capitalize on this historic opportunity. Without the influx of venture funding, other efforts such as strategic placemaking marked by infrastructure investments, creative policies and workforce upskilling must happen swiftly to ensure that Pittsburgh’s physical AI companies and their related tech and blue-collar jobs not only take root, but sprout. Unlike traditional tech, with workers mostly gazing at computer screens, the new AI economy is multidimensional and requires big thinking and uncharted collaborations with “place” at its center.

Innovation districts, compact yet powerful hubs of creativity and invention, will remain the epicenters of progress. But they will look different and include innovation teams from all sectors. Pittsburgh’s Bakery Square, along with the emerging AI Innovation Corridor along Penn Avenue, has already become a vibrant blend of advanced government units (including the Army’s Artificial Intelligence Integration Center), established tech giants (Google and Duolingo), promising startups (Lovelace AI, Strategy Robot, Hellbender), academic institutions (Carnegie Mellon University’s Tech Transfer Office) and investors (Magarac Venture Partners, UPMC Enterprises). This district combines repurposed manufacturing spaces with high-quality housing, retail and shared experiences.

But given national and global competition — with, for instance, Saudi Arabia’s government investing $40 billion in AI — U.S. cities including Pittsburgh need both dense and larger innovation footprints, enabling innovation and expansion near university talent and skilled laborers.

Take Hellbender, for example. The company is relocating to a former industrial site. As it scales up operations with the acquisition of co-robotic equipment, faster production and growing sales will lead to a workforce of 600 assembling AI-enabled products. In contrast, large industrial players such as defense primes need more expansive tracts of land, which will likely be found at a site known as Hazelwood Green, a revitalizing former steel mill site that also has the capacity to accommodate AI factories.

Another critical geography of physical AI lies within existing urban and suburban spaces where innovations come to life — schools, hospitals, universities and government buildings. These places will be testing grounds for AI deployment, sending crucial feedback to an interdisciplinary network of inventors, researchers, technologists, educators, physicians and policymakers. Through these real-world experiments, AI prototypes will be refined and scaled, driving widespread adoption of digital twins. These traditional education, finance and health-care organizations are equally in an economic race — eager to harness the power of AI through talent recruitment, some are establishing AI hubs along the stretch of Penn Avenue known as “AI Avenue.”

Another demand of physical AI is energy generation. AI consumes enormous amounts of power. The International Energy Agency predicts that by 2026, data centers alone could consume 1,000 terawatt-hours of electricity — double the amount in 2022 and comparable to the entire energy consumption of Japan. As the market evolves from data centers to AI factories, the demand for power will soar, driven by the supercomputing needs of self-driving cars, autonomous machines and industrial robots.

Here again, Pittsburgh offers a significant advantage. The region sits atop one of the world’s largest natural gas reserves, has access to abundant water supplies and boasts a skilled labor force. Former vacant steel sites in suburban counties, positioned along existing fiber Internet lines, are ready for reinvention as forward-thinking initiatives by utilities, companies and public entities continue to push the innovation agenda.

Physical AI will not only reshape our cities but will also demand that cities become active designers and participants in this transformation. Ultimately, it is humans who will bridge the gap between digital intelligence and real-world action.

Bruce Katz is the founding director of the Nowak Metro Finance Lab at Drexel University. Joanna Doven advises on economic growth and positioning strategies for developers and institutions as CEO of Premo Consultants.