Federal Reserve Governor Lisa Cook is flagging an unusual source of inflation risk: artificial intelligence itself. Speaking alongside New York Fed President John Williams, Cook said the massive build-out of AI infrastructure — data centers, advanced chips and the power plants needed to run them — could keep US price pressures elevated into 2027, even though the technology is widely expected to lower costs over the long run.
Her comments land at an awkward moment for policymakers. The Fed's preferred inflation gauge was still running at 3.4% in August, well above the central bank's 2% target. That gap is why officials have been reluctant to declare victory over inflation, and why Cook pointed to the committee's recent unanimous quarter-point rate hike as evidence the Fed is not ready to ease off.
Why AI could push prices up before it pulls them down
The logic is straightforward once you separate the short term from the long term. Building AI capacity requires enormous amounts of physical stuff: land, construction labor, electricity, cooling systems and specialized semiconductors. When demand for those inputs surges faster than supply can respond, prices for them rise. Those higher costs can ripple through the broader economy, at least temporarily.
Cook described those pressures as something that "may not resolve very quickly." In other words, the inflation bump from the AI boom is not necessarily a one-quarter blip. It could persist while the build-out runs its course.
Only later, in theory, does the payoff arrive. Once AI tools are deployed across industries, they can automate tasks, speed up research and reduce operating costs — the productivity dividend that has fueled so much optimism on Wall Street. The tension Cook is highlighting is that the cost side may show up first, and the benefit side later.
A shift in how the Fed thinks about supply shocks
The more consequential part of Cook's message was about supply shocks — sudden disruptions that reduce the availability of goods or inputs and push prices higher. Historically, the Fed has often "looked through" these episodes, treating them as temporary and avoiding big policy responses.
Cook suggested that approach may need to change. If supply disruptions become more frequent and concentrated in a handful of industries — energy, semiconductors, shipping, for example — the Fed may have less confidence that any given spike will fade on its own. In that world, policymakers could respond more forcefully, with the size of the response depending on what is being squeezed.
That is a meaningful shift in framework. It implies the Fed's reaction function could become more aggressive in the face of future shocks, rather than defaulting to patience.
What it means for investors
For markets, Cook's 2027 warning makes "higher for longer" feel less like a slogan and more like a working framework. If investors come to believe the Fed will treat more inflation bursts as something to fight rather than ignore, they can start pricing a policy-rate path that stays restrictive for longer.
That repricing usually shows up first in the 2-to-5-year part of the Treasury curve, where expectations for the next few Fed moves matter most. When those yields sit higher, it tends to weigh on long-duration assets — particularly growth stocks, whose valuations depend heavily on profits expected far in the future. Higher discount rates shrink the present value of those distant cash flows.
It also cuts the other way for some parts of the market. Companies tied directly to the AI build-out — chipmakers, power producers, data center operators and the industrial firms supplying them — could see continued demand even if the broader rate environment stays tight. The same infrastructure spending that Cook flags as an inflation risk is, for those businesses, a revenue driver.
Investors watching this theme may want to keep an eye on a few things: the trajectory of the Fed's preferred inflation gauge, the tone of upcoming Fed speeches, and whether supply-chain pressures in energy and semiconductors intensify. Related cost pressures are already visible elsewhere — for instance, memory costs are pushing phone prices higher as the AI chip boom strains supply.
Bond investors, meanwhile, have been sensitive to any hint that inflation will stay sticky. The recent move in long-dated Treasury yields reflects how quickly rate expectations can shift when inflation proves stubborn.
The bottom line
Cook is not predicting an inflation crisis. She is flagging that the AI investment cycle introduces a new, less-understood source of price pressure at a time when inflation is already above target. Combined with a possible change in how the Fed handles supply shocks, that suggests the central bank may stay cautious about cutting rates — and quicker to tighten — than markets might otherwise assume.
For everyday investors, the practical takeaway is that the interest-rate backdrop matters as much as the AI story itself. A world of persistently higher rates changes which assets tend to do well, and it rewards investors who pay attention to both sides of the AI trade: the companies building it, and the inflation it may create along the way.
