Everything About AI
NextFin News - The Federal Reserve is staring at a rare policy riddle: does the artificial-intelligence boom heat up the economy enough to demand higher rates, or will it eventually expand supply so much that inflation fades on its own? Richard Clarida, the former vice chairman of the Fed, said on Wednesday that the central bank is weighing exactly that question - whether AI-driven growth is a demand shock that argues for tighter policy, or a productivity wave that will ultimately ease price pressure. His comments landed hours after the Fed held its benchmark rate steady at 3.75% to 4% in a split decision, leaving one of the most consequential calls in modern monetary policy unresolved.
The stakes are unusually high because the two answers point in opposite directions. If AI is primarily a demand shock, the Fed's job is to lean against it with higher rates before inflation becomes entrenched. If AI is primarily a supply shock, the correct policy is patience - possibly even lower rates later - because faster productivity growth lets the economy expand without overheating. Getting the diagnosis wrong in either direction carries a real cost: over-tightening could choke off the very productivity gains that would solve the inflation problem, while under-tightening could let the AI-fueled capex boom push prices higher before the supply side ever arrives.
The Decision That Left the Question Open
The Federal Open Market Committee voted on Wednesday to keep the federal funds target range at 3.75% to 4%, following September's quarter-point increase - the first rate hike in more than three years. The hold was not unanimous. Two of the committee's 12 voters, Cleveland Fed President Beth Hammack and Fed Governor Lisa Cook, dissented in favor of a 25-basis-point hike. It was Hammack's second dissent in favor of tightening since July, and the first time Cook has ever broken ranks with the committee.
The split matters because it maps directly onto the AI debate. Cook, who dissented on the hawkish side this time, has also been one of the Fed's most explicit voices on the AI question. Speaking at the National Association for Business Economics in February, she said artificial intelligence could have "profound implications for monetary policy" and warned that if AI raises productivity, growth could stay strong even as job-market churn lifts unemployment - forcing policymakers to choose between keeping rates elevated to fight inflation or cutting them to support employment.
"The accumulation of more information will help the Committee determine the policy stance appropriate to ensure a timely return of inflation to the committee's 2% goal."
The policy statement struck a deliberately patient tone with that line, even as it acknowledged that inflation "remains elevated relative to the Committee's 2 percent goal, in part reflecting supply shocks that have driven price increases in certain sectors, including energy." Fed Chair Kevin Warsh, at his post-meeting press conference, kept the hawkish edge he has carried since taking the gavel, insisting there is "no soft inflation target" and noting that tightening in financial markets is already doing some of the Fed's work for it.
Markets took the hold largely in stride. The decision had been widely expected, with CME FedWatch data pricing roughly a three-in-four to better-than-four-in-five chance of a hike at the December 8-9 meeting. Bitcoin briefly touched $81,185 after the release before easing to about $80,950; gold slipped 0.1% to roughly $4,036 an ounce; the dollar index held above 102 and the 10-year Treasury yield sat near 4.9%. The muted reaction underscores that the real story is not this meeting but the one to come - and the diagnosis the Fed settles on in between.
Why AI Breaks the Fed's Normal Playbook
Central banks have a clean rule of thumb for most shocks. A demand shock - consumers spending more, firms investing more - pushes output and prices up together, and the textbook response is to raise rates. A supply shock that lifts productivity pushes output up and unit costs down, and the textbook response is to look through it. The problem with the AI boom is that it is both, separated by time.
The demand side arrives first and shows up in the data immediately. Building AI capacity means spending on chips, data centers, power infrastructure, and engineering talent today. That investment counts toward GDP right now, bids up the inputs it needs, and draws in financing. The supply side - the productivity payoff from deploying that technology across the economy - arrives later, and only if the technology delivers on its promise. Warsh acknowledged this sequencing at an International Monetary Fund forum in Bangkok on October 16, saying AI was "initially contributing to economic demand, while its longer-term effects on productive capacity had yet to fully emerge."
"Business investment and household spending (can) both move ahead of realised productivity gains, as many argue is happening now with investment in AI infrastructure."
The IMF's Silvana Tenreyro put the dilemma precisely in research published by Bank of England staff in August. Her conclusion was that even if AI lifts productivity, the inflation effect is ambiguous - productivity gains in services are more likely to lower domestic inflation, while gains concentrated in exports tend to push up domestic wages. That distinction matters: if the AI payoff lands mainly in traded goods and export capacity, the United States could see stronger output without the disinflation that the patience camp is counting on.
There is also a financing question hanging over the whole boom. Reporting in August noted that policymakers do not have a clear picture of who is financing the roughly $3 trillion in AI-related investment, which matters for financial-stability risk if the payoff disappoints. The comparison that hangs over the room is 2000, when the Fed kept raising rates as internet-era entrepreneurs insisted it did not matter - and the dot-com boom imploded anyway. The difference today is scale: the AI buildout touches power grids, real estate, semiconductors, and the labor market for high-end engineering talent all at once, making it harder for a single rate lever to isolate.
The Transmission Channel the Market Is Watching
The second-order question is how the AI-versus-supply diagnosis travels from the FOMC statement into the real economy. The channel runs through financial conditions, and it runs both ways.
If the Fed decides the AI boom is a demand shock, it signals more hikes ahead. Longer-dated Treasury yields rise, the dollar strengthens, and equity valuations - especially for long-duration growth stocks that dominate the AI trade - compress on a higher discount rate. That tightening in financial conditions then does the Fed's work for it: borrowing costs for households and firms climb, capex plans get deferred, and the AI investment cycle slows. In that world, the Fed does not need to raise rates as much as it signals, because the market raises them instead.
If the Fed decides the boom is supply-led, it signals patience. Yields stabilize or fall, risk assets rally, and the capex cycle gets a second wind - which, in the near term, adds more demand before the productivity arrives. That is the self-defeating loop the dissenters worry about: a patient Fed validates the investment boom, the boom adds to demand, and inflation stays elevated until the supply side finally materializes, if it ever does.
This is why the December meeting carries more weight than this one. It will bring fresh economic projections and a new dot plot, and it will force the committee to put a number on how many hikes it thinks it needs and over what horizon. A hold in October with a December hike telegraphed is the committee's way of buying time without surrendering its inflation credibility. It is also the most honest expression of the diagnosis: the Fed does not yet know which side of the AI equation is winning, so it is holding policy steady while it waits for the data to pick a side.
The Fed Is Already Split on the Answer
Clarida's framing is not an academic abstraction; it tracks a live division inside and around the central bank. On one side stand those who see the AI boom as inflationary and rates as needing to stay high. Barr, a Fed official who has disputed the idea of AI as a productivity accelerator, pointed to electricity-supply constraints on the power grid colliding with surging energy demand from data centers as a concrete inflationary mechanism. "For all of these reasons, I expect that the AI boom is unlikely to be a reason for lowering policy rates," Barr said. Minneapolis Fed President Neel Kashkari has similarly surmised that AI is likely to keep rates elevated.
"For all of these reasons, I expect that the AI boom is unlikely to be a reason for lowering policy rates."
On the other side stand those who see a once-in-a-generation supply expansion. Warsh has called AI "the most productivity-enhancing wave of our lifetimes" and argued it will be "structurally disinflationary," paving the way for lower rates. San Francisco Fed President Mary Daly recently invoked Alan Greenspan in the 1990s, noting that he looked past official data that had not yet captured higher productivity from technological advances and held rates steady rather than raising them. The Greenspan analogy is the strongest card the patience camp holds: in the late 1990s, productivity growth ran well above its historical trend, inflation stayed contained, and the Fed's willingness to look through it helped deliver a long expansion without overheating.
Between the two camps sit the data watchers. Stephen Brown of Capital Economics has argued that so far the evidence shows little inflation coming from stronger AI-driven demand, and that the Fed will not need to hike because of it. That is the empirical hinge the committee will be tracking: not what AI could do, but what the incoming prints on productivity, unit labor costs, and core inflation actually show.
The Cyclical Call Versus the Structural Call
The cleanest way to frame the policy question is to separate what is cyclical from what is structural - and to recognize that the AI boom contains both, operating on different clocks.
The investment surge is cyclical. Capex booms build, peak, and revert. Financing conditions tighten, returns disappoint at the margin, and spending growth slows. The demand pressure from AI - the bids for chips, power, and talent - is real today but mean-reverting by nature. If this were the whole story, the Fed's job would be straightforward: lean against the cycle while it is hot, and the inflation impulse would fade as the cycle turns. Three historical episodes support the mean-reversion view. The late-1990s tech capex boom peaked and rolled over even as the Fed held steady. The 2004-2008 commodity and housing capex cycle reversed sharply once financing tightened. And the 2020-2021 pandemic goods boom unwound as supply chains normalized and demand rotated back to services. Each followed the same arc: investment front-loaded, capacity caught up, and the demand impulse faded.
The productivity payoff, if it arrives, is structural. A genuine step-up in the economy's ability to produce more with the same inputs changes the trend growth rate and the neutral interest rate for years, not quarters. That is the regime-shift case, and it is why the patience camp is willing to tolerate a hot investment cycle: they are betting that the supply expansion on the other side will do more to stabilize prices than a rate hike would. But a structural call requires structural evidence - sustained acceleration in output per hour, broad diffusion of the technology beyond a handful of hyperscalers, and unit labor costs that stop climbing. That evidence is not yet in hand.
The asymmetry cuts against patience in the near term. Demand arrives on a quarterly clock; supply arrives on a multi-year clock. Over the next 12 to 18 months, the demand side of the equation is the side the Fed can actually observe. That is the most likely reason the committee held in October rather than cutting, and why December - with fresh projections - matters more than this meeting did.
What Would Change the Call
The falsifying signal is concrete and it is already on the calendar. The Bureau of Labor Statistics' productivity and unit-labor-cost reports are the scoreboard for the supply-side thesis. If nonfarm productivity fails to accelerate above roughly 2.5% annualized over the next two quarters while AI capital expenditure keeps rising, the supply-disinflation argument loses its empirical footing and the demand-shock view - higher for longer - becomes the default. Conversely, if productivity jumps while unit labor costs flatten, the Greenspan-1990s analogy moves from rhetoric to data, and the case for patience strengthens materially.
Thursday's Personal Consumption Expenditures price index, the Fed's preferred inflation gauge, is the near-term test. A hot print would validate the dissenters' call to hike and push the December probability higher; a benign print would give Warsh room to keep the committee on hold while it waits for more information. Either way, the AI question does not get answered by this meeting - it gets deferred, and the market will price that deferral as a wider range of possible outcomes for 2027.
Three scenarios frame the path from here. In the base case, the Fed holds through the rest of 2026 and delivers one more 25-basis-point hike in December, keeping policy restrictive while it waits for two more productivity prints. In the upside case for growth assets, productivity accelerates above 2.5% and unit labor costs flatten - the supply story wins, the Fed stays patient, and the AI trade rallies on a lower discount rate. In the downside case, inflation re-accelerates while productivity disappoints - the demand-shock view wins, the December hike becomes two, and the market reprices both bonds and equities lower. Each scenario has a trigger, and each is observable in the data rather than a matter of conviction.
For investors, the practical implication is a barbell of risks rather than a single line. Beneficiaries of higher-for-longer rates - parts of the financial sector, the dollar - sit on one side; beneficiaries of a productivity-led expansion with a patient Fed - technology, growth equities - sit on the other. The Fed's next move will be less about where inflation is today than about which side of the AI equation the committee believes is winning.
The bottom line: the Fed is not choosing between higher rates and lower rates. It is choosing between two stories about the economy, and the AI boom has made both stories plausible at the same time. Until the productivity data picks a winner, policy stays data-dependent, December stays live, and the market prices the uncertainty rather than the resolution.
Explore more exclusive insights at nextfin.ai.
