The AI Era: Will Artificial Intelligence Bring Inflation or Deflation?
For decades, when economists discussed inflation, they typically looked for a single number that could represent the overall price environment. When the Consumer Price Index rose, it meant inflation; when it fell, it meant deflation. Prices of different goods and services have never moved in perfect sync, of course, but in a traditional economy, a broad price index was at least a reasonably effective way to describe the overall direction of prices.
Artificial intelligence may be changing that.
If we look only at the most direct economic impact of AI, it appears inherently deflationary. Generative AI can perform information processing, analysis, writing, coding, and many forms of administrative work in a fraction of the time previously required. If companies can produce the same output with fewer employees, unit labour costs should fall, and competition should eventually force at least part of those cost savings through to consumers. In this sense, AI is similar to many technological revolutions that came before it: fundamentally, it is a technology that increases productivity and lowers the cost of production.
But AI is also an extraordinarily expensive industry. Training and running large models requires enormous amounts of semiconductors, data centres, electricity, cooling systems, communications networks, land, and construction. Meeting the growing demand for AI also requires massive investment in power generation, transmission, and other infrastructure. These investments are not abstract numbers on a balance sheet. They represent competition for finite real-world resources. When large amounts of capital flow into these areas at the same time, demand for chips, energy, construction, land, and certain raw materials inevitably rises.
AI therefore creates two opposing forces from the very beginning. It reduces the cost of knowledge and information processing while increasing demand for energy, computing capacity, and infrastructure. Suppose a company previously needed 100 employees to perform a particular business function. After adopting AI, it may need only 50, but it may also have to spend significantly more on computing and electricity. If the prices of raw materials and infrastructure are rising at the same time, the ultimate change in the company’s total cost will depend on the difference between the savings in labour and the increase in other input costs.
This suggests that AI may not simply push the entire economy toward either inflation or deflation. Instead, it may change the relative prices of different factors of production. It is entirely possible that knowledge-intensive labour and software services become progressively cheaper while electricity, computing capacity, land, and infrastructure remain relatively expensive.
AI May Change the Economics of Scale
These changes are unlikely to affect companies of different sizes in the same way.
Large companies employ large numbers of people, but a significant portion of that workforce is not directly involved in producing the final product. Employees spend considerable time in meetings, reporting, approvals, project management, finance, human resources, and coordination between different departments. By contrast, in a company with only two or three employees, a much larger share of employees' time is often spent directly on sales, product development, and customer service. If AI can automate large amounts of information processing and internal coordination, its impact on large organizations may go beyond improving individual productivity. It could reduce the number of people required to operate the organization itself.
Historically, one of the major advantages of large companies has been economies of scale. The fixed costs of R&D, IT, finance, management, and sales can be spread across a large revenue base, while smaller companies often cannot afford to build those capabilities. If AI can automate a significant portion of these functions, that advantage may gradually weaken. A company with only a handful of employees could potentially use AI to perform work that previously required an entire department, lowering barriers to entry and increasing competition in certain industries.
Large companies will not lose their advantages overnight. Capital, brands, customer relationships, supply chains, and proprietary data will continue to matter, while organizational change is usually much slower than technological change. As a result, the probability of widespread organizational restructuring over the next one or two years remains relatively limited. But if AI eventually moves from helping employees perform individual tasks to independently handling entire workflows, the possibility of major organizational redesign over the next five to ten years becomes much more significant. At that point, the deflationary effect of AI may come not only from higher productivity, but also from changes in the scale and competitive structure of businesses.
Inflation in the AI Era May Become More Structural
If these changes continue, we may need to rethink whether a single inflation measure can fully describe the price environment of the AI economy.
Suppose that over the next several years, the cost of data-centre construction, electricity, semiconductors, and infrastructure continues to rise, while the unit cost of software development, translation, basic consulting, and many forms of information processing declines. If the CPI ultimately rises by only 1% or 2%, we could reasonably say that overall inflation is relatively low. But that number would conceal significant changes taking place within the economy.
As a result, producer prices, unit labour costs, productivity, energy prices, and capital-goods prices may become increasingly important. The relationship between wages and productivity will be particularly important. If an employee's wage rises by 10% but productivity rises by 15%, the labour cost required to produce one unit of output is actually falling. It is therefore entirely possible for wages to continue rising while unit labour costs and pressure on final goods prices do not increase at the same pace.
At the same time, higher productivity can itself create new sources of demand. Higher corporate profits generated by AI can stimulate additional investment. Higher labour productivity can eventually translate into higher incomes and consumption. AI will also create entirely new products and services. If this additional demand grows faster than productivity and the economy's ability to expand supply, AI could become a source of inflation as well.
The key question, therefore, is not simply whether AI increases productivity. What matters is whether productivity can improve faster than the new demand for resources and economic activity that AI itself creates.
Time May Matter More Than Direction
The inflationary and deflationary effects of AI may also emerge on very different timelines.
During the infrastructure-building phase, data centres, semiconductors, electricity generation, and communications networks require enormous amounts of capital, while the supply of these resources often takes years to expand. AI investment may therefore have a distinctly inflationary effect in its early stages. But once that infrastructure has been built, it becomes productive capital. Additional computing capacity can then be used by thousands of businesses, gradually turning what was initially a massive fixed investment into lower unit costs of production.
The AI economy of the next few years may therefore look very different from the AI economy of the next decade. In the short term, investment and resource constraints may be more important. As AI becomes embedded in everyday business processes, changes in productivity and organizational structure may gradually become the dominant forces.
Over a five- to ten-year horizon, we believe that a scenario in which inflationary and deflationary forces coexist for an extended period deserves serious consideration. Productivity-driven deflation could eventually become the dominant force, particularly if AI productivity continues to improve rapidly. But if energy, infrastructure, and other resources remain structurally constrained, inflationary pressures could persist for much longer. Sustained high inflation, however, would require AI investment and the new demand it generates to remain significantly greater than the economy's ability to expand supply.
Conclusion
AI could bring both inflation and deflation. On one side, it lowers the cost of labour and knowledge production; on the other, it increases demand for energy, semiconductors, infrastructure, and other scarce resources. More importantly, these two forces may emerge at different points in time. The process of building the AI economy may push up the prices of certain resources, while the widespread adoption of AI in production may gradually reduce the unit cost of producing goods and services.
The more important question in the AI era may therefore no longer be whether inflation is 1%, 2%, or 3%. It may be how the costs of different inputs are changing relative to one another, and whether the pace of productivity growth can exceed the new demand for resources and economic activity created by AI itself. In our view, whether AI ultimately makes the economy more inflationary or more deflationary will depend largely on the relative speed of these two forces over the next several years and the next decade.
Disclosure: This material is for informational purposes only and does not constitute investment advice, a recommendation, or an offer to purchase or sell any securities. This commentary is only a synthesis which does not provide the full picture. Reliance on the information provided herein is at the sole discretion of the reader.
Investing involves risks, and you should always seek the help of a qualified financial professional for personalized advice tailored to your individual circumstances and risk tolerance. The opinions expressed are subject to change without notice. This information is not intended to be complete or exhaustive, and no representations or warranties, either express or implied, are made regarding its accuracy or completeness. This material may contain estimates and forward-looking statements that are not a guarantee of future performance.
This material has not been reviewed or approved by any Canadian securities regulator.
免责声明：本材料仅供参考，不构成投资建议、推荐，亦不构成买卖任何证券的要约。本评论仅为综述，并不提供完整信息。读者对本文所载信息的依赖，由其自行判断并自行承担责任。
投资涉及风险，您应始终寻求合格金融专业人士的协助，以获得针对个人情况与风险承受能力的个性化建议。本文所述观点可能在不另行通知的情况下发生变化。本信息不以完整或详尽为目标，对其准确性或完整性不作任何明示或默示的陈述或保证。本材料可能包含估计与前瞻性陈述，并不构成对未来业绩的保证。
本材料未经任何加拿大证券监管机构审阅或批准。