Q3 Outlook for Investors: Keep the future, hedge the crowd

Key points:

• AI remains structural, but it is no longer one simple trade: Investors should stay exposed, but separate AI builders, users and efficiency winners because each faces different risks from capex, pricing, funding costs and proof of monetisation.
• Lower oil prices may bring relief, but the Fed is still constrained: A US–Iran peace framework could ease inflation pressure, but sticky services, wages, tariffs and fiscal spending mean investors should still prioritise pricing power, cash flow and balance-sheet strength.
• Concentration is becoming the hidden portfolio risk: Market-cap indices have worked well, but they increasingly embed exposure to AI, mega-cap growth and momentum. Equal weight, quality defensives, income and energy/power infrastructure can help broaden the portfolio without abandoning long-term growth.

The AI story is not over. But the easy phase of the AI trade may be.

Q3 may be less about asking whether AI is real, and more about asking whether portfolios have become too dependent on one version of the AI story: US mega-cap leadership, rising capex, expensive compute, abundant funding and market-cap indices that keep rewarding the same winners.

The message for investors is not to abandon AI. It is to keep exposure to the future, while hedging the crowd.

That means treating AI as more than one trade, building resilience against sticky inflation and higher-for-longer rates, and reducing accidental concentration in portfolios that look diversified on the surface but are increasingly driven by the same mega-cap growth factors.

AI reality check: Still structural, but tactical risks have risen

The first phase of the AI trade was built on scarcity, excitement and capital deployment. Investors rewarded companies with exposure to chips, cloud, data centres, memory, networking, software or power.

The next phase will likely be judged differently. The market will want evidence of monetisation, utilisation, margins and returns on invested capital. That is a higher bar.

Based on Bloomberg sector data as on 11 June 2026, S&P 500 Information Technology EPS growth is expected to remain very strong: around 58% in Q2 2026, 52% in Q3 and 43% in Q4. That is impressive, but it also means investors are already expecting a lot. When expectations are that strong, companies do not simply need to grow. They need to keep beating forecasts.

Valuations and positioning make that challenge even more important. Bloomberg valuation data as of 11 June 2026 shows the S&P 500 trading at around 21.5x forward earnings, while Information Technology is around 24.8x. That is not bubble territory by itself, but it leaves less room for error when earnings expectations are elevated and AI has become the default long trade across semiconductors, mega-cap platforms, cloud, memory, data centres, power equipment and parts of industrials.

There are three tactical risks to watch.

1. Token pricing may make AI demand more price-sensitive
The first wave of AI adoption was about experimentation. Enterprises wanted to test use cases, launch pilots and show they were not missing the next productivity wave.

The next wave is likely to be more disciplined. CFOs will increasingly ask whether AI tools generate enough revenue, productivity or cost savings to justify the cost of usage.

If AI remains expensive to use at scale, some enterprises may slow adoption. If token prices fall sharply, usage may improve, but revenue and margin expectations for some AI providers could be challenged.

The risk is not that AI demand disappears. The risk is that AI demand becomes more price-sensitive just as markets have been pricing it as almost unlimited.

2. Funding risks are more relevant
AI capex is capital-intensive. It needs chips, servers, memory, power, cooling, land, grid connections and financing.

That was easier when investors believed rate cuts were coming. The market has moved from expecting cuts to debating whether the Fed may need to stay higher for longer, or even signal hike risk if inflation re-accelerates.

Higher rates raise the cost of capital and reduce the present value of future earnings. That matters because many AI-linked stocks are long-duration growth assets.

The AI trade can remain structurally right, but still be tactically vulnerable if funding costs rise.

3. A tactical capex peak does not require AI to collapse
AI spending does not need to fall for markets to worry. It only needs to stop accelerating.

The market has rewarded aggressive AI capex because it signals confidence. But if component prices, power costs, financing costs and infrastructure bottlenecks keep rising, investors may eventually ask whether the return on that spending is good enough.

That is the key risk for Q3: not an AI bust, but a reset in expectations. If investors begin to question whether capex growth, token economics and funding costs can all move in the right direction at the same time, the most crowded parts of the AI trade could become more vulnerable to disappointment.

Inflation in the driving seat: The Fed safety net is at risk

On the Middle East front, markets have been oscillating between peace and war. One day, investors are pricing a US–Iran peace framework and a potential reopening of the Strait of Hormuz. The next, they are reminded that ceasefires can break, negotiations can stall and regional risks can return quickly. That makes the quarter’s inflation and Fed outlook more two-sided than one-way.

If the region moves toward peace, lower energy prices could ease pressure on transport, logistics, production costs and household bills. That would support the idea that inflation may cool again in Q3 and give risk assets some breathing room.

But investors should be careful not to confuse lower oil prices with the end of the inflation problem. The broader price picture remains sticky. Wage growth, services inflation, tariffs, supply-chain shifts and fiscal spending can all keep inflation above the Fed’s comfort zone, even if energy prices fall.

The opposite risk also remains alive. If the ceasefire breaks, nuclear talks stall, Hormuz reopening faces delays or regional tensions return, oil could rebuild its geopolitical premium. That would quickly revive inflation concerns and leave the Fed with even less room to turn dovish.

This is why the Fed backstop remains hard to justify. A softer oil impulse may support a Q3 relief rally, but it does not guarantee a clean return to 2% inflation or a quick rate-cut cycle.

For portfolios, that means quality still matters. Investors should look for companies with pricing power, strong balance sheets, visible earnings and cash flow discipline. Technology still has the strongest earnings profile, but it also carries the highest expectation risk. The broader opportunity may come from sectors that can defend margins without relying on perfect growth assumptions.

The Q3 playbook is therefore not about hiding from risk. It is about owning risks that can survive more than one macro outcome: a peace relief trade, a higher-for-longer rate environment, or another inflation scare.

Concentration risks: When passive becomes crowded

The risk is not that technology disappears as a market leader. The risk is that too much of the portfolio is now tied to the same set of assumptions: AI capex keeps rising, mega-cap earnings keep beating, funding stays available, valuations remain supported, and passive flows keep reinforcing the winners.

Mega-cap IPOs could add another layer of risk to index investing. As large private companies such as SpaceX, Anthropic and OpenAI eventually enter public markets and become index candidates, passive funds may become forced buyers after inclusion. That can increase concentration further and make indices even more dependent on a small number of very large growth companies.

While passive investing remains one of the best tools for long-term wealth building, the debate is whether investors fully understand what market-cap-weighted passive exposure now represents. When a small group of mega-cap companies dominate index returns, buying “the market” may actually mean buying concentrated exposure to AI, mega-cap growth, US exceptionalism and momentum.

That concentration can work beautifully on the way up. But it also means broad index investors may have more exposure to AI disappointment, post-IPO sell-offs, valuation compression or capex fatigue than they realise.

This is why boring diversification matters again. It is not about giving up on growth. It is about reducing dependence on a single risk factor and rebuilding exposure to parts of the market that have been overlooked simply because they are less exciting.

Investors may need to revalue three things that were easy to ignore during the AI-led rally:

• Earnings durability: companies that can keep growing even when rates stay high.
• Valuation discipline: sectors where expectations are not already stretched.
• Different risk drivers: exposure that does not depend entirely on AI capex, mega-cap momentum or falling bond yields.

This is where portfolio construction matters. Investors do not need to abandon market-cap-weighted exposure, but they may need to complement it with approaches that reduce dependence on the same handful of winners.

Equal weight is one way to do that. It is not a bet against technology. It is a way to reduce the risk that one crowded theme dominates the entire portfolio, while giving more room to companies and sectors with different earnings drivers, valuations and sensitivities.

Portfolio positioning: Keep AI, but build a stronger portfolio around it

This is not a market to abandon risk, but it is one where investors need to be more deliberate about which risks they own.

The goal is to keep exposure to long-term growth themes, but reduce the risk that one crowded trade does all the work.

1. Keep AI, but know which AI you own
Investors do not need to exit AI. But they do need to stop treating AI as one single trade.

The AI theme now has very different moving parts. Some companies are building the infrastructure, some are using AI to improve their own businesses, and some are making AI cheaper and easier to adopt. These buckets can perform differently depending on whether the market is rewarding capex, productivity or efficiency.

That distinction matters for Q3. If AI capex slows, the builders may be most exposed. If companies start showing real productivity gains, the users may get more attention. If token pricing and inference costs become a bigger concern, the efficiency winners could become more important.