Cook's warning: oil and AI are the two horsemen of America's next inflation

The Federal Reserve spent three years cutting a path down from 4.75%. On Monday, one of its governors suggested the path may already need reversing. Speaking in Oakland, Lisa Cook named the two forces that could undo the entire easing cycle: oil, and artificial intelligence.

Reversals are rare in central banking, and that is the point. The Federal Reserve moves in cycles that last years: it cuts until the economy needs restraint, then hikes until something breaks, then cuts again. To cut on September 17 and face a 70–75% market-implied chance of hiking by October is not a cycle — it is a U-turn inside a fortnight, the monetary equivalent of slamming the brakes and the accelerator in the same intersection. Cook's speech is best understood as the Fed's way of telling the market the intersection is coming.

The language was the message. Central bankers do not shout; they calibrate, and every calibrated word is parsed like scripture. 'More salient' — Cook's description of the oil and AI risks — belongs to a lexicon where 'monitoring closely' means worried and 'vigilant' means alarmed. In that lexicon, 'material risks that could work in the opposite direction' of cooling prices is about as loud as a Fed governor gets without using the word that everyone in the room was thinking. She did not use it. She didn't have to. The NABE audience — professional forecasters, the people whose models the Fed itself watches — heard it anyway.

The setting mattered. The National Association for Business Economics gathers the profession's forecasters — the people whose models the Fed watches. Cook did not come to soothe them. She said the risks from oil and AI have become 'more salient' in recent months, a phrase that, in central-banker syntax, is roughly equivalent to shouting.

The mechanics are ugly. War in the Middle East — the seven-month US-Israeli campaign against Iran — has repriced energy violently upward, with Brent back above $105. And the AI buildout, the largest private investment cycle in American history, is pouring demand into electricity, chips and skilled labor — cost pressures that land squarely in the price level. 'They present material risks that could work in the opposite direction' of the cooler price data the Fed has been celebrating, Cook said.

Start with the old enemy, because it is the better understood of the two. Oil at $105 a barrel does not just raise the price of gasoline; it raises the price of everything gasoline touches — freight, food, manufacturing, the long chain of costs that ends at the checkout. The seven-month US-Israeli campaign against Iran repriced energy violently upward, and energy is the one input no economy can substitute away from in the short run. The Fed can raise rates to cool demand. It cannot drill a well. That asymmetry is why oil shocks have humiliated every central bank that ever faced one.

Now the new enemy, which is stranger. The AI buildout — data centers, chip fabs, the power plants to feed them — is the largest private investment cycle in American history, and investment is supposed to be disinflationary in the long run: more capacity, more productivity, lower costs. But investment is inflationary in the short run, because building things requires buying things — electricity by the gigawatt, chips by the million, engineers by the thousand — and bidding for scarce inputs pushes their prices up. Cook's framing captures the timing mismatch: AI arrives first as cost, in the price level, and only later — maybe — as efficiency, in the productivity statistics. The bill comes before the benefit.

The September 17 cut is what makes the warning sting. Twenty-five basis points, the first cut since July 2023, delivered under a new chair — Kevin Warsh — as the opening move of normalization, the gentle return to neutral after three years of descent from 4.75%. It was meant to buy the economy room: cheaper borrowing, a little more air. Instead, the Fed may have cut into an oil shock, and the history of that specific error is written in the 1970s. In 1974 and again in 1979, easing met energy prices and produced stagflation — the word Cook would not say, the outcome every central banker is trained to fear.

Stagflation deserves its dread explained, because it is the trap with no exit. Inflation says raise rates; stagnation says cut them; stagflation says both at once, and the central bank must choose which pain to inflict. The 1970s taught the lesson at full cost: once the public stops believing prices will stay stable, expectations themselves become inflationary, and breaking them required the brutal rate regime of the early 1980s. Cook's 'material risks' are a long way from that abyss. But the direction of travel — easing into rising energy costs — is the direction the textbooks warn about, which is why two weeks after a cut, the market prices a hike.

In central-banker syntax, 'more salient' is roughly equivalent to shouting.

Beneath the mechanics sits the philosophical split the article names: what is AI, economically? For two years the Fed's working assumption treated the boom as productivity — the 1990s pattern, when the internet wave let the economy run hot without overheating, because technology was making each worker more productive faster than demand was rising. It was the golden scenario: growth without inflation, the free lunch of the digital age. Cook's darker read does not deny the lunch. It says the restaurant charges upfront: data centers devouring power, wages for scarce talent, chips at any price — all of it landing in today's price level, with the productivity payoff arriving, if it arrives, years later.

The honest answer is that both readings are true at different horizons, and monetary policy lives in the short horizon. A central bank setting rates for the next eighteen months cannot spend productivity gains that may arrive in five years. It must deal with the electricity bills and the wage pressure now. This is the cruelty of the AI moment for policymakers: the technology genuinely could be deflationary in the long run — smarter systems, cheaper services, abundant intelligence — while being unambiguously inflationary while it is built. Cook chose to warn about the build. The build is what the Fed has to survive.

The timing is exquisite, and not in a good way. The Fed only just cut — September 17, twenty-five basis points, the first since July 2023, under new Chair Kevin Warsh. It was meant to be the start of normalization. Markets have now flipped the script: futures price a 70–75% chance of a rate HIKE by October. In two weeks, the easing cycle may be dead.

Beneath the surface, the tension is philosophical. For two years the Fed treated the AI boom as productivity — the thing that could let the economy run hot without inflation, the way the 1990s internet wave did. Cook's framing is the darker read: that AI arrives first as cost — data centers devouring power, wages for scarce talent — and only later, maybe, as efficiency. Oil is the old enemy; AI is the new one.

Everything this week comes down to Wednesday and Friday. The PCE price index lands Wednesday; the payrolls report follows Friday. If inflation is still running near 3.7% and the labor market holds, Cook's 'material risks' become the committee's baseline — and the October meeting becomes a hike meeting.

The dollar is already telling the story. It touched a two-month peak this week while traders repriced the Fed's path upward. The rest of the world is paying attention: when the world's reserve-currency central bank pivots from cutting to hiking inside a fortnight, the shockwave reaches every borrower with a dollar loan — which is to say, everyone.

None of this was in the September script. The cut was supposed to buy the economy room. Instead the Fed may have cut into an oil shock — the classic error of 1974 and 1979, when easing met energy prices and produced stagflation. Cook did not use that word. She didn't have to.

This is why Wednesday and Friday are the whole week. The PCE price index — the Fed's preferred inflation gauge, the one its 2% target is defined against — lands Wednesday, and it will say whether inflation is still running near 3.7%, nearly double the target. The payrolls report follows Friday, and it will say whether the labor market is still tight enough to pass cost increases into wages. If both confirm the hot reading, Cook's 'material risks' stop being one governor's warning and become the committee's baseline — and the October meeting stops being a question and becomes a hike meeting with the question merely of size.

The dollar is already voting. At a two-month peak, the greenback is doing the Fed's tightening preemptively: a stronger dollar makes imports cheaper for Americans, which cools inflation, but it makes dollar debts heavier for everyone else. That is the global transmission of a Fed pivot — it does not wait for the vote. Every borrower with a dollar loan, which is to say most of the developing world, is already paying for a hike that has not happened yet. The market does not need the Fed to move. It needs only to believe the Fed will.

Western coverage — Reuters, the Wall Street Journal, CNBC — treats Cook's warning as the Fed's credibility test. The narrative is discipline: a central bank that just cut cannot afford to look surprised by the very risks it cited while cutting. The Journal's framing puts it bluntly — inflation at 3.7% with oil rising is not an easing story, and markets know it.

The Western lens also reads the Warsh angle. A new chair, a first cut, and two weeks later a governor warning of hike territory — the sequencing looks awkward, and Western commentary is not shy about saying so. The dollar's strength is presented as the market doing the Fed's tightening for it, prematurely.

The Western credibility frame has teeth because the sequencing is genuinely awkward. A new chair's first cut, followed within two weeks by a governor warning of hike territory, invites the question the Wall Street Journal is asking: did the Fed misread its own economy? The defense is that the data moved — the oil shock deepened, the AI costs clarified — and that good policy responds to data rather than defending past decisions. The prosecution is that inflation at 3.7% with oil rising was never an easing story, and the cut was a dovish reflex the data did not support. Both narratives will be litigated at the October meeting, in the only court that matters: the vote.

There is also the institutional angle the Western press keeps circling. Warsh is new, and new chairs are tested early — by markets, by colleagues, by events. A chair who cuts and then must hike within a month looks reactive; a chair who holds while inflation runs hot looks complacent. Cook's speech, in this reading, does Warsh a favor: it prepares the ground, moves the market's expectations, and lets the October decision arrive as confirmation rather than shock. Whether that was the intent or merely the effect, the ground has been prepared. The market now expects what the Fed just warned about.

Eastern coverage — Xinhua, TASS — reads Cook's warning as an admission of structural weakness. The argument: the American inflation problem is imported, self-inflicted by a Middle East war Washington chose to join, and now compounded by an AI boom that burns energy without producing goods. The Fed is fighting its own foreign policy.

The Eastern lens dwells on the oil mechanism with some satisfaction. A central bank that cannot control the price of oil cannot control inflation — and the war that lifted oil prices was, in this reading, an American war. Cook's 'material risks' are presented as blowback with a spreadsheet.

The Eastern 'blowback with a spreadsheet' frame rests on a simple causal chain: the war that lifted oil prices is a war Washington joined, the inflation now threatening the easing cycle is therefore imported and self-inflicted, and the Fed is fighting its own foreign policy with interest rates. In this telling, Cook's warning is an admission — that American inflation is no longer a domestic monetary phenomenon but the fiscal and geopolitical exhaust of American choices. A central bank cannot hike its way out of a war premium. It can only make borrowing expensive while the premium persists, which is a cure that treats the symptom by taxing the patient.

The AI half of the Eastern read is colder. An investment boom that burns energy without producing goods — data centers consuming power to train models whose economic payoff is prospective — looks, from this angle, like a particularly American form of cost-push: the world's richest economy bidding up the world's scarcest inputs, then asking its central bank to manage the consequences. The satisfaction in the Eastern coverage is not hidden. But beneath it sits a genuine analytical point: when the inflation is imported and structural, the interest-rate tool is blunted, and the Fed's dilemma is real regardless of who caused it.

The Global South lens — Al Jazeera, The Hindu, Business Day — reads the story as a rerun with worse costumes. A Fed pivot from cuts to hikes means capital flight, weaker currencies, and costlier dollar debt for the developing world — the 2022 playbook. Nigeria, India, Brazil: everyone has seen this movie.

What stings, in the South's telling, is the asymmetry. When America cuts, the world gets relief; when America hikes because its own war raised oil prices, the world pays the bill. The South's finance ministers will watch Wednesday's PCE the way farmers watch the sky.

The Southern rerun metaphor is precise about the mechanics. In 2022, the Fed's hiking cycle triggered capital flight from emerging markets, weaker currencies, and costlier dollar debt — the standard playbook of a strong-dollar world. Nigeria, India, Brazil: every finance ministry on earth has the contingency plans still warm. What stings this time, in the South's telling, is the authorship of the shock. When America hikes to cool its own overheating, the world pays but understands. When America hikes because its own war raised oil prices, the world pays for a war it did not choose — and the asymmetry curdles into resentment.

The South's finance ministers will watch Wednesday's PCE the way farmers watch the sky because the sky decides the harvest. A hot print means the October hike firms up, the dollar climbs further, and the capital-flight playbook opens to page one. A cool print means the easing cycle might survive, and the relief trade breathes again. Either way, the decision is made in Washington and paid for in Lagos, Mumbai and São Paulo. That has always been the structure of the dollar system. Cook's speech just reminded everyone who operates the machinery.

None of this was in the September script, and that is the story. The cut was supposed to buy room; instead it may have bought the classic error — easing into an energy shock, with a technology boom piling cost on top. The next ten days decide whether the easing cycle survives its second week: PCE Wednesday, payrolls Friday, and then the long wait for October. Cook did not come to Oakland to soothe the forecasters. She came to tell them the path they modeled is gone, and the new one points up. In central-banker syntax, that was shouting. The market heard it.