At the start of the year, the outlook for risk assets looked relatively straightforward. Growth was firm, inflation was falling, AI and rearmament were supporting a durable capex cycle and last October’s Busan meeting between President Trump and President Xi Jinping heralded a period of G2 stability. This created a relatively constructive environment for risk assets.
Four months later, that backdrop looks less straightforward. AI optimism has accelerated but so has evidence of economic strain. S&P 500 earnings expectations for 2026 have risen sharply since March, yet the upgrades remain heavily concentrated in a small group of AI beneficiaries. The equal-weight S&P 500 remains below its late February peak, pointing to an equity rally that is concentrated in a narrow group of companies. Meanwhile, more cyclical and consumer-facing parts of the market continue to struggle, a sign that headline index strength is not being matched by broad confidence in end-demand.
At the same time, the energy backdrop has deteriorated markedly. Oil inventories have continued to fall, moving closer to levels at which operational constraints become more concerning. Further depletion risks feeding through into the real economy in ways markets may not yet fully appreciate.
For now, the US consumer has been partially insulated by roughly $47 billion in tax refunds and an additional $63 billion in tax relief, offsetting an estimated $25 billion hit from higher gasoline prices. But those supports may fade by September. As the fiscal cushion diminishes and the effect of higher energy costs accumulates, pressure on real disposable income may become more visible. Consumer confidence already points to that fragility, with the Conference Board’s index still subdued and expectations weak by historical standards.
This is the tension investors must now navigate. Our indicators still point to a reflationary macro environment, which would normally support a positive view on equities. But the margin for error has narrowed. Markets can still perform, particularly if AI spending remains durable. However, investors are increasingly relying on sustained AI-led growth, resilient consumers and limited economic fallout from higher oil prices all holding at once.
The challenge is that the two forces now doing the most work in that outlook — the pace of AI adoption and the path of geopolitical risk — are unusually hard to read. Both are moving quickly and both depend on decisions that have not yet been made. In some cases, the decision-makers themselves may not yet know where their thresholds lie.
Waiting until everything is clear is not a strategy. But neither is assuming that every uncertainty will resolve favourably.
Let’s start with AI. The theme has accelerated because the technology has taken another meaningful step forward and markets are beginning to understand what it could mean commercially. The first leap came in 2025 with the emergence of more capable reasoning models and the growing commercial use of inference-time computing. The second leap arrived in early 2026 with the rise of agentic AI. Markets realised that AI agents materially increase inference demand, creating bottlenecks across the AI supply chain, from GPUs to data centres. Token usage accelerated as pricing power improved across parts of the ecosystem, reflected indirectly in a roughly 70% increase in the spot price of Nvidia H100 GPUs in the on-demand market since April.
The bullish case for AI is increasingly straightforward: demand for compute continues to grow and unit economics are improving. That inflection since March helps explain why markets have become more optimistic about the sustainability of AI spending.
Will this continue? It certainly could. AI is a transformative technology and transformative technologies tend to require enormous amounts of capital. The recent supply chain bottlenecks are not necessarily a sign of excess. Scarcity and rising prices are a normal feature of any large product buildout; indeed, they are part of the way the price mechanism resolves scarcity.
At the same time, it is probably too early to extrapolate recent price action into a lasting improvement in AI economics. Some investors, including Goldman Sachs, argue that rising inference demand could drive durable improvements in pricing and monetisation across the ecosystem. Perhaps. But previous technological revolutions generally matured through scale, competition and falling unit prices, and AI is unlikely to be entirely different. It may be that markets are correct to remain cautious about the valuation implications for AI-exposed companies, even as they become more bullish about the technology’s long-term utility.
The Strait of Hormuz is the other major question mark. As the economic cost of continued disruption rises, so does the pressure for some form of resolution. The difficulty is that the path to that resolution remains deeply uncertain. Much depends on variables that are hard to handicap: the stability of the Iranian regime, domestic tolerance for prolonged economic stress, how long Western public opinion will tolerate the status quo, whether Israel escalates and whether regional powers such as the UAE, Saudi Arabia, Turkey or China alter their positions.
Even in the event of a reopening, normalisation will not be instantaneous. Tanker positioning, refinery restarts and broader supply-chain adjustments could take well into late 2026 to stabilise. Reopening the Strait may remove the worst-case scenario for markets without immediately repairing the economic damage already working its way through the system.
Markets still appear to be treating the Hormuz disruption as temporary rather than systemic. Yet shortages, depleted inventories and demand destruction could continue to weigh on real economic activity long after the immediate crisis has eased.
Our baseline remains that equities can perform reasonably well in a reflationary environment, particularly if AI spending remains sustainable. But that view is now more fragile than it was at the start of the year. Market leadership is narrow, the consumer economy looks softer than the headline indices imply and the effects of the energy shock have yet to fully pass through the system.
Markets may be right to look through the disruption, but only if several things hold at once: AI spending continues to accelerate, consumers absorb the squeeze from higher energy prices and the economic fallout from the oil shock remains contained. That is a narrower path than markets seemed to be pricing only a few months ago.
