Technology has been a deflationary force, and AI may push this to the extreme. The cost of many services could decline rapidly, but at the same time, this progress will become a unique driver for increased commodity demand. The broader societal impacts of AI will be massive and deserve their own article, but today, we will examine these inflationary and deflationary countertrends.
AI has a unique ability to reduce costs in service-related industries. Services make up a significant portion of consumer spending; services account for about 70% of consumption in advanced economies such as the U.S. This includes everything from healthcare and education to banking and insurance. The labor-intensive nature of these sectors, including the required skills and credentials for employees, has historically made them costly. AI, however, is changing this dynamic.
The trend is occurring across a range of simple to complex services. For example, the rise of AI-powered chatbots and voice assistants already slashes the need for human labor in almost all digital customer service forums. Call centers are crucial for many industries and can account for between 2% and 3% of a company’s operating expenses. This number is often higher for complex businesses like telecom. AI systems can now handle routine customer inquiries, complaints, and services at a fraction of the cost of human agents. This leads to substantial savings, lowering consumer costs in a competitive market.
Klarna, for example, recently achieved significant cost reductions through AI initiatives, particularly by deploying an AI assistant powered by OpenAI. This tool handled the equivalent workload of 700 full-time employees, handling two-thirds of Klarna’s customer service conversations. Every situation is unique, but even a fraction of the potential savings from the above has profound implications for costs across industries.
The ability of AI to reduce service costs is also applicable to the most complex services. Historically, complex professions had high costs because people needed compensation for the decades of research required. A key example would be healthcare which also happens to be the number one cost leading to consumer bankruptcy in many countries and one of the most significant line items for multiple governments worldwide. It’s simply too expensive, and healthcare inflation hasn’t slowed, and the training required has only increased. AI-enabled systems can now quickly analyze medical data, identify patterns, and assist doctors in making more accurate diagnoses. This increases efficiency and significantly lowers healthcare costs by reducing diagnostic errors and unnecessary treatments. This trend is also occurring in multiple other industries, such as legal, as they incorporate tools to lower service costs.
AI technology, though incredibly efficient in the above applications, still requires significant amounts of hardware, energy, and materials, driving demand for commodities. To begin with, AI models require vast computational power to operate, and training these models demands even more. The Electric Power Research Institute recently increased its data center power consumption estimates after incorporating more AI growth, saying data centers could consume over 9% of US power. Third-party forecasts such as this are steadily climbing as many aren’t used to a demand source growing at this rate.
Moreover, AI is increasing the demand for specific metals critical to building the hardware infrastructure to run AI systems. Copper, for example, is essential for creating the wires and components in data centers and electrical systems. Countries that have historically taken a long-term view on commodity trends, such as China, have already let their copper stockpiles reach multi-year highs. Recently, as shown by Figure and Tesla, robots have proven capable of making AI decisions and even delivering services. They will require all the same critical materials at scale, as if we were conceptualizing a new auto industry. While AI reduces the labor cost that operates and maintains these systems, it cannot eliminate the need for the commodities that make them function.
This dynamic is likely to result in a bifurcation in inflationary trends. On the one hand, consumers will see rapidly lower costs in services like customer support, healthcare, education, and financial services, thanks to AI-driven efficiency. On the other hand, the prices of commodities like metals, natural gas, uranium, and industrial materials may increase due to the needs of AI infrastructure and the associated power demands that continue to eclipse estimates from just a few months ago.