Fed’s Williams sees AI supply squeeze lasting beyond 2 years

An AI equipment supply squeeze happens when companies want more AI hardware than manufacturers can produce or deliver. The gap affects chips, servers, power equipment, and other infrastructure needed to build AI systems. It matters because scarce equipment becomes more expensive and harder to obtain.

The article points to AI infrastructure investment as the main source of pressure. As companies expand their AI capabilities, they compete for limited equipment. That competition is pushing up chip prices and the prices of goods that use chips. Bitcoin miners moving into AI may also compete for chips and power equipment.

John Williams expects the imbalance to last longer than two years if demand continues as expected. The shortage could keep raising costs across the economy. It may also affect miners seeking AI contracts, since they face substantial financing needs and increasingly selective lenders.

Williams said AI equipment supply could take more than two years to catch up with demand. In other words, he does not expect the market to become balanced within the next two years. A longer shortage would mean continued pressure on equipment availability and prices.

His forecast depends on demand remaining strong. When asked whether supply and demand would become more balanced over the next two years, he said, “I think it’s going to take longer.” He also called continued demand a major assumption, showing that the timeline depends on how quickly AI investment keeps expanding.

The forecast matters beyond technology companies. Higher chip and equipment costs can feed into the prices of goods that use them. The pressure may also affect Bitcoin miners developing AI businesses, because they compete for chips and power equipment while facing difficult financing conditions.

Building AI infrastructure requires large amounts of specialized equipment, especially chips. When many companies expand at the same time, their combined orders can exceed current production capacity. Suppliers then face a market with more buyers than available equipment, creating upward pressure on prices.

The article gives the mechanism directly: the AI infrastructure buildout is pushing up chip prices and the prices of goods that use chips. A more expensive chip raises costs for equipment and products that include it. Limited power equipment can add another bottleneck, particularly as Bitcoin miners pursue AI projects and compete for the same resources.

Williams expects supply to take more than two years to catch up if demand stays strong. That means the price pressure may not disappear quickly. It could raise costs for AI developers, miners, and other businesses, while also contributing to broader inflationary pressure.

When chips and AI equipment become more expensive, companies using them may face higher production and infrastructure costs. Some businesses absorb those costs, while others pass them to customers through higher prices. This is how a technology supply squeeze can create wider inflationary pressure.

The article connects the AI buildout with rising chip prices and prices for goods that use chips. Williams said the supply shortage could add to inflation. If inflation remains too high, policymakers may keep interest rates higher or raise them again to reduce demand and bring price growth down.

Williams reaffirmed that one more rate hike may be appropriate this year. However, he said he had no reason at that time to believe more action would be needed in 2027. The future decision will depend on inflation, demand, and how quickly AI equipment supply improves.

The Federal Reserve is the United States’ central bank. It guides monetary policy, including the interest rates that influence borrowing and spending across the economy. The article focuses on John Williams, president of the Federal Reserve Bank of New York, and his view that another increase may be appropriate.

In general, raising interest rates makes loans and other financing more expensive. That can reduce spending and investment, easing demand for goods and services. Slower demand can reduce businesses’ ability to raise prices, helping inflation move lower. This general explanation is established economic knowledge; the article itself only states that the Fed is seeking to lower inflation.

Williams reaffirmed that one more rate hike may be needed this year. He said the question of 2027 would require waiting to see how conditions develop. His comments link persistent AI-related cost pressure with the Fed’s continuing inflation challenge.

AI companies may keep investing because they expect AI to produce very large productivity gains and transform the economy. If those expected benefits are valuable enough, higher borrowing costs may not stop companies or investors from funding infrastructure. This is why strong AI spending can continue even during tighter monetary policy.

Williams said AI investors are “relatively price insensitive.” He explained that they believe AI will bring huge productivity gains and transformative changes. Their investment decisions may therefore depend less on interest rates than investments with smaller or less certain expected benefits.

He did not say rates have no effect. Williams said AI investment is not completely independent of interest rates, but he does not expect monetary policy to have a deep effect on it. That suggests demand for AI equipment could remain strong, prolonging the supply squeeze and its inflationary pressure.

AI infrastructure usually starts with chip design. Manufacturers then fabricate chips in specialized facilities, inspect and test them, and package them for use in servers. The article does not describe these production steps, so this explanation uses established semiconductor and data-center knowledge rather than article-specific detail.

Next, companies assemble servers and networking equipment inside data centers. Builders prepare buildings, cooling systems, racks, and connections. Power equipment makers produce and install systems that deliver electricity safely and reliably. These parts must work together before AI systems can run at scale.

The article highlights the result of this long chain: demand for AI infrastructure is rising faster than supply. That pressure is pushing up chip prices and prices for goods that use chips. Supply may take more than two years to catch up, while miners and AI companies compete for chips and power equipment.