The canary is spluttering in South Korea, surrounded by the hot gas of AI hype, and it’s not looking good. Federal Reserve Chair Kevin Warsh should keep a close eye on the sickly bird as he toils in the mines of American monetary policy. <\/p>\n
On Thursday, the Bank of Korea (BOK) raised its policy rate from 2.75 percent to 3.00 percent<\/a>, its second increase in succession after a 25-basis-point hike in July.<\/a> <\/p>\n South Korean policymakers are tightening into a semiconductor-led expansion that gives Warsh a live example of AI’s awkward economic sequence. Investment arrives before productivity\u2014with inflation<\/a> pressures in between. <\/p>\n Read More on Analysis<\/p>\n This is an early-warning case rather than a template for the United States. South Korea’s boom runs heavily through semiconductor exports, while America’s AI expansion is showing up through data centers, equipment, software and electricity demand. <\/p>\n Yet the transmission mechanism now occupying the central banks of both nations is strikingly similar. A technology expected to lower costs eventually is first creating a price-inflating rush for capital, materials and skilled labor. <\/p>\n For Warsh, the timing is especially pointed. He is due to deliver keynote remarks at Jackson Hole on Friday<\/a>, his first appearance at the symposium since taking office as chairman on May 22<\/a>, with U.S. inflation still running well above the Fed’s target. <\/p>\n To hike or not to hike? As always, that is the question. The AI boom may force Warsh’s hand and his rate-setting board. A heavily indebted America\u2014and a vengeful president demanding rate cuts\u2014is waiting. <\/p>\n South Korea’s Chip Boom Reached Monetary Policy<\/p>\n The BOK’s explanation for Thursday’s decision went well beyond an isolated jump in chip prices. It said exports and investment were maintaining high growth<\/a> because the semiconductor sector had been stronger than expected, while improving incomes were supporting a faster recovery in consumption. <\/p>\n Core inflation rose to 2.6 percent in July, and the bank raised its forecast for core inflation to 2.5 percent in both 2026 and 2027. <\/p>\n The link from AI hardware to household demand has been unusually clear in South Korea. <\/p>\n A BOK study earlier this month found that first-quarter GDP grew 3.8 percent from a year earlier<\/a> while gross domestic income jumped 13.2 percent, as higher semiconductor prices improved the country’s terms of trade. <\/p>\n The bank said those income gains could strengthen household purchasing power and companies’ capacity to invest, and described the current semiconductor upturn as reflecting structural demand associated with the diffusion of AI.<\/p>\n For monetary policy, this shows productivity gains and investment spending need not arrive together. <\/p>\n Factories, data centers and chip supply chains require equipment, energy and workers before businesses have fully incorporated the resulting technology into their everyday production, delivering the efficiency gains that can soothe costs. <\/p>\n Korea offers a case in which the income generated by the boom is itself helping demand recover while the productivity dividend remains incomplete. <\/p>\n To be fair, AI was only one channel in the Korean decision. The BOK also cited accumulated cost pressures, exchange-rate and oil-price risks, rising wages, strong housing prices around Seoul and substantial household-loan growth<\/a>. <\/p>\n So any U.S. comparison that turns South Korea’s rate hike into a single-cause AI story would outrun the bank’s own account. But it remains a valuable lesson for policymakers in America. <\/p>\n The Fed Is Already Debating the Same Sequence<\/p>\n The American version of all this is developing through a different industrial structure. <\/p>\n Minutes from the Fed’s July meeting<\/a> said staff attributed part of elevated inflation to the surge in demand from the AI buildout, alongside tariffs and Middle East-related energy and input costs, while AI investment continued to support business spending. <\/p>\n Fed officials also recorded large price increases for data-center materials such as chips and steel<\/a>, with pressure appearing in computer equipment, software and electricity. <\/p>\n Several policymakers thought the effects remained confined to selected categories; others believed AI investment was already pushing more broadly on prices through aggregate demand, or soon would. <\/p>\n Meanwhile, the engine of spending remains powerful in the AI sector. <\/p>\n Just look at Nvidia, which reported $96.2 billion of quarterly revenue<\/a> for the three months ending July 26, more than double the figure a year earlier. Its data-center business generated $89 billion. <\/p>\n As CEO Jensen Huang put it, “demand is accelerating.” That is a lukewarm way of expressing a hot number. Nvidia is forecasting 70 percent revenue growth next year. Yes, you read that correctly.<\/p>\n Warsh told Congress in July that rapid business-investment growth was being driven in large part by data centers<\/a> and AI-related equipment and software, with high-tech equipment spending up nearly 25 percent over four quarters. <\/p>\n During the Senate hearing, Warsh went further on the timing question. <\/p>\n Asked about the AI boom’s price effect, he said, “Will it increase measured prices over the course of the next 12 months? I suspect it will<\/a>,” while arguing that a one-time increase need not turn into persistent inflation if supply responds. <\/p>\n It is a distinction likely to matter at Jackson Hole as Warsh delivers his keynote. <\/p>\n July PCE inflation was 3.7 percent from a year earlier<\/a> and core PCE inflation\u2014the Fed’s preferred measure\u2014was 3.3 percent. <\/p>\n The Fed held its target range at 3.50 percent to 3.75 percent at its July meeting. But three policymakers preferred a quarter-point increase. The number of those on the hike side of the argument may grow quite soon. <\/p>\n Lisa Cook’s Productivity Case<\/p>\n Fed Governor Lisa Cook has made a strong case for looking through the buildout phase of AI. <\/p>\n She said in May that she expected firms’ adoption of AI to boost productivity and that those gains could “put downward pressure on inflation<\/a>.” <\/p>\n Her baseline forecast was for disinflation to resume without another rate increase, although she said she was prepared to tighten if that improvement failed to arrive. <\/p>\n As the July Fed minutes show, it’s a view that enjoys support beyond Cook. Some policymakers expected AI-driven productivity to reduce production costs and expand aggregate supply, while acknowledging substantial uncertainty about how quickly those effects would appear.<\/p>\n This is, then, the test that could make the Korean warning less relevant to America. <\/p>\n If U.S. productivity accelerates quickly enough, and today’s higher prices for chips, power, construction and specialist labor remain concentrated rather than spreading through wages and consumer prices, the investment surge could coexist with renewed disinflation. <\/p>\n But if those price pressures broaden before the supply response arrives, Korea’s sequence becomes more instructive for America. <\/p>\n Warsh’s Problem Is Also Trump’s<\/p>\n President Donald Trump has repeatedly pressed for cheaper money, saying in July that the U.S. should have the lowest interest rate in the world<\/a>. This came after years of tormenting Warsh’s predecessor, Jay Powell, on the issue. <\/p>\n At the same time, Trump’s administration is encouraging rapid AI infrastructure growth, even as it openly recognizes how that can turn into higher household costs.<\/p>\n The White House’s Ratepayer Protection Pledge<\/a> asks AI companies and data-center operators to finance the additional power generation and grid infrastructure their projects require rather than pass those costs to ordinary ratepayers. <\/p>\n The politics of that balancing act are acutely sensitive before the midterms, in which the Republican majorities in the House and Senate are on the line. <\/p>\n A Reuters\/Ipsos poll conducted from July 29 through August 3<\/a> found 48 percent of Americans called the cost of living the most important factor in deciding their 2026 vote; 70 percent disapproved of Trump’s handling of inflation and the cost of living. <\/p>\n Higher rates also operate in an economy carrying large stocks of debt<\/a>, a concern for all of us who remember the 2008 financial crisis. <\/p>\n U.S. household debt totaled $18.77 trillion<\/a> in the second quarter, and nonfinancial corporate debt securities and loans stood at about $14.45 trillion in the first quarter<\/a>. <\/p>\n Of course, those totals do not by themselves imply a debt crisis. Fed staff in July characterized household and nonfinancial-business debt vulnerabilities as moderate and said household balance sheets remained strong. <\/p>\n But they do mean that another tightening cycle would eventually feed through to borrowers. It would suck more consumer money out of the real economy, diverted instead to higher debt repayments. A double-whammy with higher inflation too. <\/p>\n Too Late, Too<\/p>\n Once Warsh finishes speaking at Jackson Hole, attention will shift to the next Federal Open Market Committee meeting on September 15-16<\/a>. <\/p>\n Evidence that AI-linked price pressure is spreading beyond a few inputs, alongside persistent core inflation and stronger wages in construction and technical trades, would make the Korean warning stronger and more pertinent. <\/p>\n Faster productivity growth, easing input prices, and renewed core disinflation would weaken its value as the Korean canary in the American coal mine. <\/p>\n