There is a growing risk of economic overheating in the US as the artificial intelligence boom expands beyond semiconductors and spills into the broader economy — never mind the tame wage growth and house prices that would typically point in the opposite direction. If AI spending continues apace, accompanied by a rising stock market, there is likely no way to avoid a widening pickup in inflation.
This marks a shift from last year when AI optimism was roughly offset by weakness in other parts of the economy, leading to a K-shaped dynamic that Americans have generally disliked.
More than US$700 billion in capital spending by four technology behemoths alone this year is creating new winners and new bottlenecks as it spreads through the economy. Add to that an energy supply shock that businesses did not have to worry about even three months ago, and it is hardly surprising that bond markets are taking notice, driving up borrowing costs for households and companies.
The broadening of AI spending has been a three-year process. Nvidia Corp was the main immediate beneficiary of the launch of OpenAI’s ChatGPT in late 2022, which boosted demand for the company’s high-powered chips and helped it unseat Microsoft Corp at the top of the S&P 500 Index. Next came data centers as companies realized the enormity of what AI represented. The insatiable demand for computing power has since hit electrical infrastructure, grid equipment and power generation equipment.
This AI infrastructure spending was, until recently, twinned with weakness in other parts of the industrial economy.
Residential investment has been in the doldrums since 2022, while inflation-adjusted spending on office buildings is at the lowest level in 15 years. Construction spending in manufacturing dropped 20 percent over the past 18 months as US President Donald Trump ended much of the Biden administration’s support for green industries. Crude oil prices below US$60 a barrel as recently as the start of this year cooled off the oil and gas sector.
It is also important to note that conviction in the scale of the AI investment cycle has wavered at times. The DeepSeek moment in early last year briefly raised the possibility that more compute-efficient models would allow for a cheaper buildout. “Liberation Day” tariffs, proposed last April, were another short shock to the system, as was the first month of the war with Iran.
However, the industry was in some ways caught off guard by how rapidly Anthropic PBC’s Claude products were adopted this year, driving a renewed intensity in capex plans.
The outlook for the industrial economy now appears more solid than it has been since the Fed began raising interest rates in 2022, thanks to Alphabet Inc, Amazon.com Inc, Microsoft and Meta Platforms Inc. Their spending spree is boosting not just semiconductor stocks such as Advanced Micro Devices Inc and memory companies including Micron Technology Inc, but also the energy equipment company GE Vernova Inc, where orders grew 71 percent last quarter from a year earlier, and Caterpillar Inc, where power generation sales grew 48 percent. In a sign of the times, Ford Motor Co is getting into the grid-battery business to tap into the AI boom.
Manufacturing surveys point to a pickup as well. The new orders component of the Institute for Supply Management’s manufacturing report has come in stronger in each of the past four months than it has been at any point since the first half of 2022.
An argument could be made in the recent past that none of this was finding its way into core inflation, which excludes volatile energy and food prices. Somewhat elevated readings have largely reflected one-time effects from tariffs and lags in things like shelter inflation. Wage growth for workers has continued to cool — at less than 4 percent now; it is similar to the late 2010s when inflation was not a concern — and real-time measures of apartment rents and home prices have shown very little inflation at a national level.
That is what makes some of the more recent data worrisome. Overall job growth has picked up in 2026, with important cyclical industries such as manufacturing and construction showing headcount growth. Housing market activity also appears to be accelerating in May despite higher mortgage rates, with stock-market wealth one potential source of the increase in home sales.
Computer software and accessories have made a meaningful contribution to the core personal consumption expenditures deflator — a favored Fed inflation measure — in the five months through April, according to Inflation Insights president Omair Sharif. A category worth a little more than 1 percent of the index weight has accounted for about 18 percent of the average monthly increase in core inflation, he wrote in a report Tuesday, adding that further increases are likely since memory- and storage-related categories “show no sign of slowing down.”
The billions of dollars flowing into these components, data-center construction, electrical equipment and power generation alongside trillions of dollars of value creation in the stock market would eventually overwhelm the cyclical weakness we have had in the labor market and industrial sector over the past few years. That time might be now.
We still need a few pieces to fall into place for the full picture to emerge. Job market sentiment is still in the doldrums, and a sharp pullback in tech stocks could quickly cool investment plans. Failing that, the assumption should be that a continued capex and stock market boom would put the K-shaped economy to bed and awaken an inflation impulse that policymakers and politicians could no longer wish away.
Conor Sen is a Bloomberg Opinion columnist. He is founder of Peachtree Creek Investments.This column reflects the personal views of the author and does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
The closure of the Strait of Hormuz has sent the vast Asian chemicals industry into a tailspin. Deprived of the likes of Qatari natural gas and Saudi Arabian oil, the region’s fertilizer and plastics plants are slowing production or even shutting down. Everywhere except China, that is. In petrochemicals, China is unique. As well as a traditional industry that uses oil and gas as feedstock, it has parallel output that relies on its abundant domestic coal. Unsurprisingly, India and other regional powers want to copy and paste the Chinese method. This would not be easy — or climate friendly. The
KMT Chairwoman Cheng Li-wun’s (鄭麗文) recent visit to Beijing and her upcoming visit to Washington will serve as a high-level test of her diplomatic mettle. In Beijing, Cheng was received with symbolic gestures, a warm reception, and high-level access. In Washington, she will receive far less pomp and far sharper questions about the KMT’s vision for the future of Taiwan. Her challenge will be to persuade Washington that the KMT’s engagement with China can coexist with strong deterrence. Cheng’s April 7-12 visit to mainland China coincided with an intense period of conflict in Iran. Despite the strategic significance of Cheng’s trip,
History might remember 2026, not 2022, as the year artificial intelligence (AI) truly changed everything. ChatGPT’s launch was a product moment. What is happening now is an anthropological moment: AI is no longer merely answering questions. It is now taking initiative and learning from others to get things done, behaving less like software and more like a colleague. The economic consequence is the rise of the one-person company — a structure anticipated in the 2024 book The Choices Amid Great Changes, which I coauthored. The real target of AI is not labor. It is hierarchy. When AI sharply reduces the cost
US President Donald Trump recently repeated his claim that “Taiwan stole America’s chip industry,” reigniting public debate on the issue. As a former Taiwanese minister of economic affairs and an entrepreneur deeply involved in semiconductor supply chain development, I feel a responsibility to clarify this misunderstanding. From the perspective of global industrial evolution and the economic principle of comparative advantage, such a statement appears overly simplistic and risks obscuring the essence of the issue. The rise of Taiwan’s semiconductor industry was not built on “replacing America,” but rather emerged as a result of countries pursuing different development paths within the