Will Artificial Intelligence Be Inflationary or Deflationary?
Insights from Victor H. Marin, VP, Senior Investment Analyst for Tompkins Financial Advisors
Artificial intelligence is often described as a potentially powerful disinflationary force. The basic argument is simple: AI reduces the cost of performing cognitive tasks, increases worker productivity, automates routine processes and allows businesses to produce more with fewer resources. If these potential productivity gains translate to greater supply and lower unit costs, AI could place persistent downward pressure on inflation.
Productivity is a central argument to this disinflationary thesis. AI can accelerate research, software development, customer service, logistics, inventory management and administrative work. Employees supported by AI may produce more in the same amount of time, while businesses may require fewer labor hours to deliver the same output. In the most competitive industries, some of these savings should ultimately be passed on to consumers through lower prices, or at the very least slower price increases.
AI may be particularly disruptive in digital services, where the marginal cost of producing an additional unit can approach zero. Software, analysis, marketing content and certain professional services that once required substantial human effort can increasingly be replicated at very low incremental cost. Thinking through the logic here, human personnel are often the most expensive cost of doing business, which means these savings could have material effects on the bottom line. This is the long-term disinflationary thesis. The near-term reality, however, may not conform to these lofty, optimistic expectations.
AI is not appearing in the economy without cost. Businesses are spending heavily on semiconductors, data centers, electricity, software, engineering talent and the restructuring required to incorporate AI into existing operations. The Federal Reserve Bank of New York has coined the term “productivity J-curve” to describe this phenomenon. In short, measured productivity may initially disappoint because investment and implementation costs arrive before the full benefits of the technology. What ultimately matters most in determining the impact on inflation is not just whether AI raises productivity, but whether it raises productivity faster than it raises the cost of adoption.
Evidence of near-term price pressures is already visible. The Richmond Fed has identified unusually strong price increases in electrical equipment, software publishing and power-related engineering services associated with the AI and data-center buildout. AI-related investment is therefore expanding demand for scarce physical and technical resources before it has meaningfully expanded the economy’s productive capacity or led to increased corporate profitability.
The memory chip market provides a useful example of the unintended consequences, or “unknown unknowns,” that can accompany this transition. The rapid growth of AI infrastructure has increased demand for high-bandwidth memory. Manufacturers have consequently directed more production capacity toward these higher-margin products, tightening the supply of conventional memory units that are commonly used in personal computers and other consumer electronics. Rising memory costs have placed upward pressure on laptop and device prices in the short-term. Look no further than the refreshed MacBook prices on Apple’s website. This dynamic demonstrates how a new technology, expected to reduce costs over time, can initially make unrelated products more expensive.
At Tompkins Financial Advisors, we believe the inflation debate surrounding AI is best viewed through a time horizon lens. Our conclusion is that AI is likely to be inflationary in selected areas during the investment and adoption phase, but increasingly disinflationary as productivity gains diffuse and ripple across the economy. It goes without saying that this transition will not occur uniformly. Some costs will collapse, while bottlenecks in electricity, semiconductors, infrastructure and specialized labor may drive other prices sharply higher.
AI may ultimately reduce the marginal cost of intelligence, expand productive capacity and lower the economy’s underlying inflation rate. But the journey to that destination is unlikely to be linear. The decisive question is not whether AI will increase productivity, it almost certainly will, but whether those gains will be broad, competitive and large enough to exceed the considerable costs and disruptions required to achieve them.
https://www.tompkinsfinancialadvisors.com/about-us/contact-us
Investments and insurance products are not insured by the FDIC, not deposits of, obligations of, or guaranteed by the bank or its affiliates, and are subject to investment risk including possible loss of principal. The opinions voiced in this material are for general information only and are not intended to provide specific advice or recommendations for any individual.
Work Cited
https://www.cnbc.com/2026/06/26/ai-memory-chip-shortage-consumer-electronics-prices.html