Cook, An Update on AI and the Economy

September 28, 2026 An Update on AI and the Economy Governor Lisa D. Cook At the Oakland Tech Week Opening Keynote, cohosted by the Kapor Center, Oakland, California Share --> --> --> --> --> --> Thank you, Freada and Mitch, for that kind introduction. I appreciate the invitation to speak here at Oakland Tech Week and for the opportunity to return to the East Bay, where I lived and spent several formative years when I attended graduate school at Berkeley. 1 Oakland was a vibrant, exciting place then and is an even more vibrant and exciting place now, and I am happy to have the opportunity to engage with you on a critical topic: artificial intelligence (AI) and its effects on our economy. In fact, my interest in the economics of innovation and artificial intelligence began at Cal. One of the most intriguing, impactful courses I took was on growth theory taught by Paul Romer who posited that investing in science and ideas could produce continuous, unbounded growth and later received the Nobel Prize for his seminal contributions advancing our thinking related to economic growth. I believe we are living in an era that, coupled with the historic, post-World War II investment in basic science, is seeing these and similar ideas coming to fruition and being tested in the global economy today. AI is poised to become the most significant technological shift of our lifetime. I view it as a general-purpose technology, on par with or exceeding breakthroughs, such as the steam engine, electricity, and the internet. Those innovations spread throughout the economy, sparked downstream innovation, and improved over time. 2 AI is doing the same. As a long-time researcher of the economics of innovation and current monetary policymaker, I have observed these developments with keen interest. AI affects nearly every aspect of my role at the Federal Reserve, including monetary policy, financial stability, bank supervision, financial market infrastructure, and our own operational preparedness. 3 While AI introduces an infinite set of exciting possibilities, it also offers us much to contemplate in a sober way. From the data on which AI models are trained to AI safety, I have long been an advocate of responsible AI development. 4 Whether all the beneficial possibilities of AI are realized—and how—will depend on how researchers, consumers, businesses, and policymakers across the country rise to meet its opportunities and its challenges. To fully recognize the benefits of AI, we must pair optimism with caution and be cognizant of valid concerns AI may pose for privacy, bias, workers, fraud, cybersecurity, and intellectual property rights. For our part at the Fed, as a supervisor of banking organizations, we have encouraged, and continue to encourage, responsible use of AI within the financial sector in a manner that is consistent with safe and sound practices and in compliance with applicable law. Beyond that, any specific role for the government in the path of AI's trajectory is for elected officials to decide. I will continue to monitor closely the speed, direction, and magnitude of AI development as I assess the economic outlook, the appropriate path of monetary policy, and financial stability. What are AI's implications for monetary policy? Today, I will offer an approach for thinking about AI's economic implications related to our dual mandate of promoting price stability and maximum employment. I am studying these effects closely, because it is my job to set monetary policy in a way that will navigate the multiple sources of near-term pressure while fostering conditions for long-run gains that benefit all Americans. AI and Inflation First, I would like to consider inflation implications in the short run and longer run. A broad range of factors have caused inflation to remain above the Fed's 2 percent target over the past five years. Over the past year, one of these factors has been AI-driven investment. Prices for AI-related goods—such as chips, computers, and software—have surged. I believe that some of these steep price increases reflect a shift in demand toward AI-related sectors rather than an increase in economy-wide demand. When a surge in demand is concentrated in one sector, goods and services in that sector can get pushed onto a steep part of its supply curve. Had that demand been spread evenly across the economy, the overall price index would not climb as much. As supply chains adjust and efficiency gains accrue, this kind of pressure should resolve on its own without policy intervention. In fact, attempting to fight sector-specific inflation with monetary policy could be a mistake. Our tools are too blunt to target narrow sectors, and addressing relative price shifts is not our role. Nonetheless, I see some economy-wide pressure from AI-fueled demand. Data-center investment relies on inputs, like construction labor and energy, that are broadly used in many sectors in the economy. As a result, increased AI investment could introduce price pressure to those other sectors. And even more investment is in the pipeline, and companies have only spent a small fraction of the $2 trillion in announced plans. 5 Further, a large portion of the rise in equity prices over the past few years can be attributed to enthusiasm about AI, and that added wealth appears to be feeding through to household spending. You can see signs in the inflation data that the pressure may be broadening: Electricity and water costs are each up around 5 percent over the past year, potentially attributable in part to AI, and core goods prices, which were drifting down before the pandemic, are running over a 3 percent annual pace so far this year. This introduces the risk that, even as inflation in the narrow AI sector moderates, new and more broadly based price pressures may take its place. A well-timed productivity boom could counter broadening price pressure, if it were to increase the supply capacity of the economy more than it increases demand. To understand this mechanism, consider an economy where, because of lower input costs, it becomes possible to produce more goods and services at the same cost. If demand does not expand to meet additional supply, you would expect prices to fall. This reduction in price pressure can be described in a classic aggregate supply—aggregate demand framework—a first and major toolkit of macroeconomists. Currently, I anticipate that productivity gains will provide modest disinflation within the next few years. However, I do not expect those effects to arrive in time to offset the broadening inflationary pressure later this year. Moreover, uncertainty surrounds any estimates related to how and when this mechanism may operate, and it warrants further research and discussion. The key is determining the conditions under which a productivity boom would provide inflation relief relative to today by increasing the economy's supply potential more than demand. In a scenario where the productivity gains are spread evenly across the economy, I expect the relief to be limited but real. It is limited, because higher productivity not only raises the economy's potential supply but also raises demand through the expectation of higher future wages, better returns on investment, and the accompanying gains in wealth. In fact, some of that extra demand is likely already baked into today's economy through AI data center investment and wealth effects. Therefore, going forward, the new supply generated by rising productivity may outweigh the new demand, bringing supply and demand into better alignment and reducing the upward pressure on prices felt today. On balance, a broad-based increase in productivity, when and if it comes, could lead to a modest easing in price pressure. To be clear, I am also attuned to other scenarios, including those in which a productivity boom generates less disinflation than in my baseline, as well as those in which it leads to even more downward pressure on inflation. The scenarios where we get more inflation relief imply that demand is adversely affected and falls well short of the economy's capacity. This could happen. if the productivity gains were highly concentrated among higher-income consumers who tend to spend less of each additional dollar of income or wealth; if productivity gains are not passed on to wages because of low worker bargaining power; or if a painful stretch of job reallocation leaves unemployment higher and confidence lower, increasing households' precautionary savings. Scenarios where we might get less inflation relief are ones in which productivity gains are not passed on to prices—say, if market concentration leads to less competition and higher markups. Ultimately, the timing of any disinflationary payoff will depend on how quickly and broadly businesses adopt AI tools, how changes in business practices translate to higher productivity, and how fast any productivity gains pass through to the labor market. I am very uncertain as to the breadth and timing of these channels and will adjust my view depending on what I see in the data. AI and the Labor Market While AI's implications for inflation are increasingly notable, the technology's potential effects on the labor market have long captured the public's attention and deserve policymakers' close study. Looking at the history of technologies that changed the way humans work—such as the mechanical loom and the PC—we see widespread anxiety about those advancements at the time of their introduction. In the past, the benefits and costs of technological change did not necessarily arrive simultaneously, and that could well be the case with AI. This technology could bring the most significant reorganization of work in generations. 6 AI technology and its adoption are still in their early stages. At the moment, there is limited evidence that AI is yielding significant changes to the structure of the labor market. The data underscore this: Both the unemp