The Federal Reserve is entering its July meeting with a problem that is far more serious than a single uncomfortable inflation print. The issue is not that inflation has suddenly returned to the levels seen after the pandemic. It is that the structure of inflation is changing again, just as policymakers had begun to believe that the worst of the adjustment was behind them. Oil has moved back above USD 100 per barrel. Fuel markets are becoming dangerously tight. The Trump administration is rebuilding its tariff wall. Artificial-intelligence investment continues to absorb extraordinary quantities of capital, electricity, equipment and labour. Each of these pressures operates through a different channel. Together, they are beginning to produce the kind of inflationary environment in which the Federal Reserve can no longer rely on patience alone. A softer-than-expected consumer-price report for June initially appeared to give policymakers time. The market reduced the probability of an immediate increase in rates, assuming that the disinflation process remained intact and that the Fed could maintain its current stance while waiting for more evidence. That interpretation lasted only a few days.

Expectations of a quarter-point rate increase at the July meeting rose sharply, reaching almost 40% at one point before settling somewhat lower. Investors are no longer debating only when the Fed might cut rates. They are again considering whether it may need to raise them. This reversal is more important than the exact probability attached to the next decision. It shows that the market has lost confidence in a predictable monetary path. For most of the previous cycle, investors assumed that inflation would continue to fall, allowing the Fed gradually to ease policy. That assumption supported bonds, equities, credit and emerging markets.

The new environment is very different. The Fed is no longer choosing between tightening and easing within a stable macroeconomic framework. It is being forced to respond to a sequence of supply shocks whose duration, intensity and transmission remain highly uncertain. The Middle East is the most immediate source of danger, but even here the conventional focus on crude oil misses the deeper problem. The real stress is increasingly visible in refined products. Petrol, diesel and jet fuel are the fuels that keep the global economy operating. They move goods, power logistics networks, support agriculture and maintain aviation. A shortage of crude oil matters, but a shortage of refining capacity can be even more damaging because crude oil cannot be consumed directly. It must first be transformed. That transformation system is now operating with almost no margin for error. Refineries across the world are running at exceptionally high utilisation rates. In the United States, processing levels are close to their highest seasonal point since 2018 even though national refining capacity has fallen by almost 600,000 barrels per day over the same period. Since the beginning of 2025, roughly one million barrels per day of processing capacity have disappeared from the American and European markets through closures. At the same time, export restrictions imposed by Russia and China following sanctions related to Iran have removed an estimated 2.5 million to 3 million barrels per day of refined products from global trade. The result is not simply an expensive oil market. It is a saturated fuel system.

This distinction matters enormously for inflation. When crude oil rises, governments and companies may attempt to draw on inventories, redirect supplies or adjust purchasing schedules. When diesel becomes scarce, the entire economy feels the pressure almost immediately. Diesel is the fuel of trucks, agriculture, construction, mining, shipping and industrial machinery. Its price is embedded in almost every physical product long before that product reaches the consumer. Higher diesel prices therefore do not remain confined to petrol stations. They enter food prices. They enter transport costs. They enter manufacturing margins. They enter the final retail price of imported goods. This is why the current situation represents a more dangerous form of inflation than a simple increase in crude oil futures. The fuel market is no longer absorbing the shock. It is amplifying it. The United States has already seen petrol exceed USD 4 per gallon in many states, while diesel has moved above USD 5. Russia is experiencing long queues and domestic shortages after repeated Ukrainian attacks on refinery infrastructure. France has suspended weekend traffic restrictions for fuel tankers to maintain supplies to service stations. These are not isolated disruptions.

They are evidence that the global energy system is losing redundancy. A resilient system contains spare capacity. A fragile system operates at maximum utilisation, assuming that nothing goes wrong. The current refining system increasingly resembles the second. This is particularly dangerous because refineries running continuously at full capacity become more vulnerable to mechanical failures, extreme heat, storms and maintenance problems. A single unplanned shutdown can therefore have a disproportionate effect on prices. The market can cope with strong demand when spare capacity exists. It can cope with geopolitical disruption when inventories are high. It can cope with refinery outages when alternative facilities are available. It cannot easily cope with all three at once. The economics of refining illustrate the severity of the imbalance. US refining margins, measured through the theoretical value of converting three barrels of crude into two barrels of petrol and one of diesel, approached USD 70 in mid-July. That was the highest level on record and more than twice the level observed before the conflict with Iran began. Such margins are not simply a sign of corporate profitability. They are a distress signal. They indicate that the market is willing to pay an extraordinary premium for the ability to transform crude oil into usable fuel.

Normally, high margins would attract additional production. That adjustment is difficult today because nearly every available facility is already operating close to full capacity. There is no large hidden reserve of refining capacity waiting to return. There is no obvious source of additional imports capable of filling the gap. There is no rapid infrastructure solution. This is why the ultimate balancing mechanism may become demand destruction. Either the conflict de-escalates and supply conditions improve, or fuel prices rise sufficiently to force consumers and companies to reduce consumption. From an economic perspective, demand destruction is simply another name for lost growth. Consumers travel less. Companies reduce activity. Transport becomes more expensive. Marginal production is postponed. Households cut discretionary spending to pay higher energy bills. The inflationary shock therefore gradually becomes a growth shock.

This is the environment in which the Federal Reserve must now make its decision. The central bank cannot produce diesel. It cannot reopen damaged refineries. It cannot guarantee the security of the Strait of Hormuz or the Red Sea. It cannot prevent attacks on Russian energy infrastructure. It can only influence the second-round effects. That means preventing higher energy prices from spreading into broader inflation expectations, wages and corporate pricing behaviour. This is why the case for tighter policy has strengthened even though the latest consumer-price report was relatively benign. Inflation data are backward-looking. Oil prices, shipping costs, refining margins and tariffs are forward-looking inputs. The Fed is not simply asking what inflation was in June. It is asking what inflation may become in September, December and the first half of next year. Several members of the Federal Open Market Committee have already begun to argue that the current stance may not be restrictive enough. Dallas Fed President Lorie Logan has suggested that rates may need to move slightly higher because inflation is not on a convincing path back to 2%. Cleveland Fed President Beth Hammack has emphasised that inflation currently represents a greater concern than employment. Both are voting members and could oppose a decision to hold rates unchanged. They are unlikely to be alone in their concern.

The minutes of the previous meeting showed that several officials had already considered scenarios in which inflation remained elevated due to AI demand, geopolitical conflict, or tariffs. Almost all believed that such a combination would probably justify higher rates. Those scenarios are no longer theoretical. They are beginning to materialise simultaneously. The artificial-intelligence boom is especially important because it complicates the conventional distinction between supply-driven and demand-driven inflation. AI may eventually increase productivity and reduce costs. Before it does so, it requires enormous physical investment. Data centres must be constructed. Electricity generation must expand. Transmission networks must be upgraded. Advanced semiconductors, memory chips, cooling systems and specialist equipment must be produced. Skilled workers must be hired. Land must be secured. All of this creates immediate demand for resources that are already constrained. The result is inflationary before it becomes productive. The sheer scale of investment illustrates the point. Alphabet has indicated capital expenditure of up to USD 205 billion this year. Other technology companies are pursuing similarly ambitious programmes. This is not another software cycle. It is an infrastructure cycle. Infrastructure cycles are capital-intensive, energy-intensive and labour-intensive. They compete with governments, utilities, manufacturers and households for financing and resources.

This creates pressure not only on consumer prices but also on long-term interest rates. The federal government must finance large fiscal deficits. Technology companies must finance the AI build-out. Energy companies must finance additional generation and grid capacity. Defence expenditure is increasing. Supply chains are being duplicated for strategic reasons. All these borrowers compete for the same pool of capital. The consequence is that long-term Treasury yields may remain elevated even if the Fed keeps the policy rate unchanged. This is one reason Kevin Warsh faces a credibility test so early in his chairmanship. The Fed can hold rates steady and argue that it needs more evidence. But if markets believe the central bank is underestimating inflation, long-term yields may rise anyway. A pause does not necessarily produce easier financial conditions. It can produce the opposite if investors demand greater compensation for inflation risk. This creates a counter-intuitive possibility. A modest increase in the policy rate today could reduce long-term yields tomorrow if it convinces the market that the Fed remains committed to price stability.

The logic is politically attractive. President Trump has repeatedly expressed concern about long-term borrowing costs. Warsh could therefore present a rate increase not as an act of economic restraint, but as a measure designed to prevent inflation expectations from driving Treasury yields even higher. That argument is not without merit. Bond markets often respond more positively to credible tightening than to ambiguous patience. The problem is that credibility cannot be reduced to one decision. An immediate increase would demonstrate determination, but it would also risk overtightening into a future slowdown. The fuel crisis may ultimately destroy demand. Higher tariffs may weaken consumption. AI investment may remain concentrated in a narrow group of companies rather than supporting the broader economy. Labour-market conditions may deteriorate. If the Fed raises rates just as these forces begin to slow growth, it could transform an inflation shock into a recession. This is why the decision remains so difficult.

The committee is not divided between those who understand inflation and those who do not. It is divided over timing. One group believes the Fed should act before inflation broadens. Another believes it should wait for evidence that the shock is passing into underlying prices. Both positions are defensible. The danger lies in waiting too long if the shock persists. The experience of the post-pandemic cycle remains deeply embedded in institutional memory. Central banks were slow to recognise that temporary price pressures had become persistent. They maintained overly accommodative policy and were then forced into rapid tightening. No policymaker wants to repeat that mistake. Yet the current episode is different. The earlier inflation shock combined extraordinary fiscal stimulus, excess savings, supply-chain disruption and labour shortages. The present shock is emerging from geopolitical fragmentation, energy insecurity, tariff protection and an infrastructure boom. The transmission may be slower. It may also be more durable.

This is why the next Fed meeting matters less for the quarter-point decision than for the framework Warsh establishes. Markets need to know what evidence would justify a rise. They need to know whether the Fed is primarily focused on current inflation or future inflation. They need to know how much weight the committee gives to energy, tariffs and AI-related demand. They need to know whether policy remains data-dependent or has become risk-dependent. The distinction is important. A data-dependent central bank waits for inflation to appear in official statistics. A risk-dependent central bank acts when the probability of future inflation becomes unacceptable. Warsh has deliberately reduced forward guidance, arguing that markets should not expect a predetermined path. That approach offers flexibility. It also increases uncertainty. Without clear guidance, investors are forced to construct their own policy scenarios. Some banks are preparing for further increases, while others continue to hedge against cuts. The market is no longer confidently forecasting the Fed. It is preparing for both extremes. This uncertainty will remain even if rates are left unchanged. A hold would not restore the previous easing narrative. It would simply postpone the decision. If fuel prices remain elevated, tariffs begin to affect import prices and AI investment continues to accelerate, the pressure for higher rates will intensify into September. If the conflict de-escalates, refining margins fall, and labour-market weakness becomes more visible, the case for patience will strengthen. The difficulty is that the Federal Reserve does not control which scenario emerges. It controls only its response.

For investors, the broader lesson is that inflation protection can no longer be treated as a temporary tactical allocation. The world is entering a regime in which energy security, industrial policy, technology investment and geopolitics are permanently connected. Oil is no longer merely a commodity. It is a strategic asset. Refining is no longer merely an industrial process. It is a geopolitical bottleneck. Tariffs are no longer merely a trade instrument. They are part of national-security policy. Artificial intelligence is no longer merely a technological innovation. It is a global competition for capital and energy. Each of these developments points towards higher structural costs. This does not imply that inflation must remain permanently elevated every month.

The Federal Reserve faces a choice. If it assumes that the old disinflationary order will return, patience remains reasonable. If it accepts that the world has entered a structurally more inflationary period, the burden of proof shifts. Rates must remain higher. Policy must become more pre-emptive. And credibility must be defended before inflation reappears fully in the data. The Fed is not simply deciding whether to raise rates in July. It is deciding whether yesterday’s monetary framework is still adequate for tomorrow’s inflation.