The cleanest question in the mailbag this morning was not whether oil is up or down. It was whether oil still gets to explain growth. The reader's hunch was good: if the next leg of the economy is built around AI, chips, data centers, cloud contracts, and electrified industry, the old reflex of reading the world through barrels starts to miss the room.

But the conclusion needs discipline. The new economy is not a cloud. It is a load. It asks for transformers, interconnect queues, gas turbines, nuclear restarts, copper, cooling water, permits, and rate cases. The demand curve has moved from the pump to the substation. That does not make it cleaner, cheaper, or less political by default. It makes the bottleneck more local and harder to photograph.

The International Energy Agency's latest electricity work is blunt about the direction of travel: global electricity consumption is rising fast, and data centers plus AI are a visible new source of demand. In the United States, the IEA has said data centers could account for a striking share of electricity demand growth by the end of the decade. On the oil side, the same agency has been describing a slower, more contested demand path, with electric vehicles and changing industrial patterns cutting into the old certainty that every expansion is also a crude story.

Markets are catching up in their own nervous way. Last week, Wall Street pulled back from record highs after an AI-heavy rally, with oil and inflation worries doing the rude work of reminding traders that physical inputs still matter. The Federal Reserve's May financial stability report also put two things in the same frame that business rhetoric likes to keep separate: stretched valuations in AI-related assets and the risk that geopolitical shocks or energy prices unsettle the inflation path.

This is the point. The AI economy does not abolish oil politics. It adds another energy politics on top of it. The data center boom can soften oil's symbolic power while intensifying the fight over electricity. Utilities that were built for boring growth are being asked to accommodate an industrial surge dressed as software. Governors now compete for server farms the way they once competed for auto plants. Ratepayers are told the future is arriving, then asked to underwrite the poles and wires.

The old barrel economy had a brutal clarity. Tankers moved, prices spiked, consumers noticed. The wire economy is more administrative. It hides in interconnection delays, grid upgrades, capacity auctions, power-purchase agreements, and the quiet transfer of risk from tech giants to public systems. That makes it less visible, not less consequential.

So yes, the reader is right that oil is no longer the single master gauge of growth. But that is not liberation. It is a change in the pressure point. The newspaper's rule for this cycle should be simple: follow the load. When a company says AI will transform productivity, ask where the megawatts come from. When a governor announces a data center, ask who pays for the transmission. When oil falls, do not assume the machine got light. It may simply have plugged itself into a different wall.

The World Health Organization has declared the Ebola outbreak centered in the Democratic Republic of Congo and Uganda a public health emergency of international concern. That phrase is bureaucratic, but it matters. It is the global health system's way of saying that the danger has crossed from local tragedy into shared risk.

The reports available Sunday point to a serious and still-moving picture. AP reported more than 300 suspected cases and dozens of deaths. Reuters cited WHO figures that included suspected deaths and a smaller set of lab-confirmed cases. The mismatch is not unusual in an outbreak's early stage. It is part of the problem: the confirmed count lags the disease, and the suspected count carries uncertainty. The operational answer is not theater. It is testing, tracing, protective equipment, local treatment capacity, and clear communication before rumors outrun the virus.

Ebola is politically dangerous because it is visually legible in the worst way. It produces fear faster than funding. It allows faraway governments to look decisive by talking about travelers, while the real work is slower and closer to the ground. The WHO alert should move money, logisticians, field laboratories, medical teams, and community trust. If it only moves speeches, the world has learned nothing from the last decade of emergency rehearsal.

The strain reporting will matter, as will geography. Congo has lived through repeated Ebola outbreaks; Uganda has border and health-system experience of its own. That experience is valuable, but it is not infinite. Health workers burn out. Supply chains break. Communities that feel managed from above stop cooperating. A disease response fails first socially, then medically.

The front-page test for rich countries is whether they treat the alert as an invitation to assist or as a chance to pose. The best early sign would be boring: funding that arrives, labs that run, protective gear that does not disappear, and case data that improves. The worst sign would be familiar: travel panic accompanied by a faint hum of indifference to the places doing the actual containment.

The 2026 World Cup is no longer just a sports event approaching the United States, Canada, and Mexico. It is becoming a live test of which kinds of mobility the American state is willing to smooth.

The State Department has suspended a $15,000 visa bond requirement for World Cup ticket holders from Algeria, Cape Verde, Ivory Coast, Senegal, and Tunisia, according to AP. Those countries had been among the places affected by a broader visa-bond policy. The exception is narrow, practical, and revealing. It says the government can move quickly when the traveler arrives attached to a sanctioned global spectacle.

This is not only about soccer demand. FIFA has been selling a global event into a political climate defined by border fights, cost anxiety, and suspicion of international movement. The host country wants the tournament's money, television, tourism, and prestige. The border apparatus wants enforcement. The compromise is a ticket-shaped credential: not open movement, but event movement.

That matters because mega-events do not just reflect policy. They expose it. They create classes of visitors: the sponsor guest, the athlete, the journalist, the fan with a verified ticket, the ordinary applicant outside the magic circle. The country can tell itself it is merely making a tournament work. It is also deciding which foreign presence is valuable enough to accelerate.

The move may be sensible on its own terms. A World Cup that cannot admit fans becomes absurd. But the exception is also a useful x-ray. It shows that friction at the border is not a fixed natural law. It is administered, priced, waived, and justified. When the guest is attached to FIFA, the machine can bend. The question is what that says about everyone else standing in line.

The AI trade has spent the spring trying to levitate above ordinary macroeconomics. It has a clean story: software demand, chip scarcity, capital discipline somewhere in the future, and enough productivity vapor to perfume every valuation model. Then oil, inflation expectations, and the Fed returned to the room.

Reuters reported that Wall Street pulled back from all-time highs Friday as inflation worries rose and investors reassessed the path of interest rates. Semiconductor names, which had been central to the rally, were hit hard. Oil's rise was part of the immediate explanation. It does not need to destroy the AI story to matter. It only needs to raise the discount rate, squeeze consumers, and remind investors that a data center is a building connected to an energy system, not a theorem.

The Federal Reserve's latest financial stability report gives the discomfort a more official shape. It points to stretched valuations in equities, including AI-adjacent names, and notes the vulnerability of risk assets to shocks. That is central-bank language for a simple condition: when a theme gets crowded, the exit door becomes part of the asset.

There is a difference between the AI economy and the AI trade. The economy is capex, power, chips, labor substitution, software integration, and eventually productivity if the deployments work. The trade is multiple expansion, momentum, and the market's willingness to accept future margins before they exist. The first may be real and still take years. The second can become expensive by lunchtime.

Investors should be especially wary of companies that talk about AI savings while hiding AI costs. Compute is not free. Power is not free. Engineers are not free. Depreciating accelerated hardware is not free. The most comic phrase in corporate America right now is capital-light AI strategy.

The market's job is not to believe or disbelieve AI. It is to price duration, risk, and cash flow. Friday's retreat was a small event, not a verdict. But it exposed the weak joint in the current narrative: a supposedly weightless technology boom is already fighting for heavy resources.

The AI copyright fight is losing some of its philosophical haze and becoming what it was always destined to become: a fight over price.

A federal judge has been pressing lawyers in the Anthropic copyright settlement, where authors are seeking approval of a large payout tied to allegedly pirated books used in AI development. Separate litigation continues against other major AI companies, including Meta. The basic pattern is now familiar. Creators allege that copyrighted work was absorbed into training systems without permission. AI firms answer with fair-use arguments, technical distinctions, and warnings about crippling innovation.

The public debate often treats this as a moral binary. It is more likely to end as a market structure. If training rights can be licensed, priced, pooled, audited, and insured, then the industry gets a cost line. If they cannot, the industry gets rolling legal uncertainty and selective settlements. Either way, the fantasy period is ending.

The interesting part is not whether every model will reveal every training source. They will resist that as long as possible. The interesting part is whether courts and negotiators create a practical tariff on the cultural back catalog. Books, photographs, news archives, film scripts, and code repositories are not the same asset. A model can ingest them all, but law and bargaining will sort them differently.

That sorting will shape the next generation of AI companies. Big incumbents can absorb licensing costs and settlements. Smaller labs may be pushed toward cleaner datasets, synthetic data, narrow domains, or open corpora. Publishers may discover that their archives are valuable, but not as valuable as their press releases hope. Authors may find that collective action matters more than outrage.

The technology industry likes to describe this litigation as a drag on progress. That is too convenient. Price discovery is not anti-innovation. It is how a society tells a new industry what it has actually been using.

Cannes is useful when it behaves less like an awards corridor and more like a temperature reading for taste. This year, the temperature is anxious, defensive, and occasionally alive.

AP's Cannes coverage has the festival opening under the shadow of AI, streaming economics, and Hollywood's continuing uncertainty about what it is selling. That is not a new cultural crisis, but it has acquired a sharper machine edge. The question is no longer only whether theaters can compete with platforms. It is whether human taste can survive systems optimized to predict it.

The encouraging answer, as ever, is not a panel. It is a film that refuses to be a frictionless content object. Jane Schoenbrun's latest work arrived with the kind of genre pressure and queer unease that Cannes still knows how to amplify. The festival's value is not that it crowns the global consensus. It is that it sometimes lets difficult work look central for a few days.

The anti-AI rhetoric around cinema can become sentimental fast. Not every human-made movie is worth defending, and not every automated tool is an aesthetic crime. The actual defense of art is narrower and harder. It is the defense of surprise, locality, risk, bad timing, odd desire, and the unscalable choices that make a work feel authored rather than merely assembled.

Cannes is not innocent in this. It loves hierarchy, sponsors, ritual, and the red-carpet economy. It can turn rebellion into a gown before dessert. But the festival still has one function the platforms envy: it can make taste feel like an event rather than a recommendation. That may be old-fashioned. It may also be the last institutional advantage culture has left.

The fashionable line is that sports has slipped behind politics, AI, markets, and whatever else has captured the collective nervous system. I do not buy it. A thing is not irrelevant when states bend around it.

The World Cup's cultural power is no longer measured only by whether a casual viewer watches every group-stage match. That was never the right test. The better test is whether the event forces institutions to reveal themselves. On that measure, the tournament is doing fine. It has already become a border story, a tourism story, a security story, a pricing story, a labor story, and a legitimacy story.

A government that suspends a visa-bond requirement for certain ticket holders is telling us something precise. It is saying that sport can authorize movement that ordinary human purpose may not. That is not a reason to sneer at soccer. It is a reason to watch the tournament more closely.

The pageantry is not decoration. Pageantry is how power makes itself friendly enough to televise. Flags, opening ceremonies, corporate slogans, national anthems, crowd shots, and volunteer uniforms are not the opposite of geopolitics. They are geopolitics with choreography.

So no, the World Cup is not dead. It is mutating. The question is less whether people care about football in some pure way and more whether they will accept the event as a temporary operating system for public life. For a month, cities will price themselves differently, police differently, market themselves differently, and admit foreigners differently. That is culture. That is politics. That is a front-page story with cleats on.

The AI industry wants to be judged by capability. Investors want to judge it by revenue. Users judge it by whether the thing helps. All fair enough. But the most honest review of AI will arrive in a dull envelope from the utility.

A technology that needs vast data centers has entered civic life before it has earned civic consent. The product demo shows a model answering a prompt. The infrastructure demo is a transformer upgrade, a gas plant kept online, a nuclear restart debated, a transmission line fought, a county commission meeting extended into the night.

This does not make AI bad. It makes it industrial. The industry should be forced to talk like an industry. Where is the power coming from? Who pays for grid upgrades? Which customers get priority? What happens to rates? Is the company buying clean power that exists, or merely rearranging certificates while the grid burns whatever can be dispatched?

The public has been trained to ask whether AI will take jobs. It should also ask whether AI will take capacity. That question is harder to moralize, which is why it is better. Capacity is where slogans die. There are only megawatts, substations, queues, pipes, permits, and bills.

If AI produces extraordinary productivity, society may decide the buildout is worth it. But the burden of proof belongs to the industry asking everyone else to host the machinery. The future should be allowed to arrive. It should not be allowed to hide its meter.