The first half of 2026 has made the outlook difficult to read. AI-linked earnings have been exceptional: Nvidia's latest quarterly revenue reached $81.6bn, up 85 per cent year on year, while Micron followed record fiscal third-quarter results with striking guidance for the next quarter. Yet inflation remains stubborn, central banks are moving in different directions, and disruption in the Strait of Hormuz has shown how quickly an energy shock can re-emerge. The task for investors is to identify what markets are already pricing in, and what could prove those assumptions wrong.

Can AI earnings keep doing the heavy lifting?

The case for AI is no longer about distant potential. The revenues are real. Nvidia generated $75.2bn from data centres in its latest quarter, while Micron's cloud-memory and core data-centre businesses have become central to its results. Spending on chips, memory, networking, data centres and power is visible across corporate accounts.

That distinguishes this boom from the late 1990s. Technology accounts for more than a quarter of the S&P 500's trailing net income, while commanding 39.4 per cent of its market value. But strong profits do not, by themselves, settle the valuation debate. They simply move the question further along the chain.

Nvidia's sales confirm that its customers are spending heavily on AI infrastructure; they do not prove that those customers will earn an adequate return on that spending. The real test is whether AI can lift revenues, reduce costs and create durable competitive advantages beyond the current product cycle.

The bull case is that demand is broadening beyond the first phase of infrastructure. The bear case is that capital expenditure outruns monetisation, margins weaken and exceptional hardware revenues prove insufficient to validate every AI valuation. Watch spending plans, pricing, margins and, above all, evidence of returns. AI is real; whether eventual cash flows justify today's prices remains open.

Will central banks stay patient, or be forced to change course?

Monetary policy may be the market's most reliable source of surprise. In June, the Federal Reserve held its target range at 3.50-3.75 per cent, the Bank of England held Bank Rate at 3.75 per cent, though two members voted for an increase, and the European Central Bank raised rates by 25 basis points.

The encouraging view is that higher energy and transport costs prove temporary, allowing underlying inflation to keep easing. The less comfortable possibility is that the shock becomes embedded where transport bills feed into goods prices, wage pressure remains sticky, services inflation proves stubborn and expectations rise. A policy pause would then become harder to sustain than markets expect.

One inflation release or press conference will not decide it. Watch services prices, wage growth, credit conditions and expectations. Is inflation fading beneath the surface, or spreading through the economy?

Could Europe finally beat the United States?

Europe is cheaper. At the start of June, the STOXX 600 traded on 18 times forward earnings, against 28 times for the S&P 500. The gap reflects Europe's weaker growth record, but also demanding assumptions embedded in US equities.

Europe need not become the world's growth engine. It only needs to be less disappointing than investors expect. Better earnings revisions, firmer industry, fiscal support or steadier trade could make European equities attractive relative to an expensive US market.

The discount exists for reasons. The European Commission expects EU growth of 1.1 per cent in 2026 and inflation of 3.1 per cent after the energy shock. Europe remains vulnerable to energy costs, industrial weakness and soft exports. Higher rates or weak demand could squeeze margins and delay investment.

Watch earnings revisions, lending surveys, industrial orders, export demand and corporate margins. If they improve, Europe can outperform without matching America's technology leadership. If not, apparent cheapness may remain merely apparent.

Are investors becoming complacent about geopolitical risk?

The Strait of Hormuz is a reminder that lower oil prices do not necessarily mean lower risk. Currently, oil loadings continued after attacks on vessels and renewed US-Iran strikes. Yet shipping slowed, and Brent had fallen 10.6 per cent in the previous week before rising again as violence resumed.

The constructive reading is that physical flows continue, producers are adapting and an outright supply shock may be avoided. But partial flows are not normality. The costs of disruption can emerge later through insurance, freight, rerouting, delivery times and inventory draws, before reaching consumer prices or corporate margins. That is delayed inflation pressure that can complicate life for central banks.

Watch shipping volumes, vessel routes, loadings, inventories, freight rates and the oil futures curve. Physical evidence will say whether the risk is genuinely receding.

Will market leadership finally broaden beyond AI?

The final question is whether the rally can become healthier. As the second half begins, the technology sector accounts for 39.4 per cent of the S&P 500's market capitalisation. Include Alphabet, Amazon and Meta, classified outside the technology sector but among the biggest investors in AI infrastructure, and AI-linked companies make up more than half the index.

That concentration has rewarded investors while the leaders deliver, but leaves the index exposed if a handful of companies miss demanding expectations. Broader participation would make the market more resilient.

AI infrastructure needs more than semiconductors. It needs power generation, electrical equipment, grid upgrades, construction, cooling, networking and specialist industrial capacity. Utilities and industrial firms could benefit from the same cycle that has lifted chipmakers.

Still, a few good weeks for banks, smaller companies or industrials would not mark a new regime. Real broadening would show up in equal-weighted indices, more stocks participating in rallies and stronger earnings revisions beyond the largest technology names.

The second half will bring inflation releases, central-bank meetings, earnings reports, energy scares and geopolitical headlines. The task is not to predict every one, but to judge each against assumptions already embedded in prices.

Successful investing is rarely about getting every forecast right. It is about asking better questions, recognising uncomfortable evidence early and changing one's mind before the crowd is forced to do so.