AI After the Euphoria: Enter the Infrastructure Phase

Artificial intelligence has powered one of the strongest equity cycles in recent years. Momentum has now moderated. Valuations among the leading names have become elevated. A number of high growth names have struggled. Headlines increasingly question sustainability as energy constraints and monetisation challenges come to the foreground.

The key question for investors is how to position for what comes next.

Intrasia Wealth View

We see the AI theme transitioning from narrative to execution. The next phase is likely to reward cash generation, contract visibility and balance sheet discipline. The companies that supply the essential infrastructure for compute, networking, power and cooling appear better placed than those that rely on distant or uncertain end market payoffs. This framing is consistent with the reality that power availability has become a defining constraint on data centre capacity and therefore on the pace of AI deployment. 

The durable demand set

• Semiconductors and networking. Compute intensity continues to rise. Bottlenecks are increasingly visible in high throughput networking, memory and advanced packaging. Scale providers with clear roadmaps and pricing power remain advantaged.

• Data centre power and cooling. AI workloads are energy and heat intensive. Power distribution, uninterruptible power systems, switchgear and thermal management now sit at the core of reliable capacity planning. These categories benefit from slower product cycles, installed base replacement and service attachment. The importance of electricity availability and price has become a central determinant of where and how fast infrastructure can be built. 

• Specialty materials. As manufacturing complexity rises, fabs require high purity chemicals, contamination control and advanced filtration. Suppliers embedded in customer processes face high switching costs and can hold pricing power.

• Utilities and midstream, selectively. Regions that can offer reliable and affordable electricity are positioned to capture multi year investment. Grid upgrades and contracted volumes can support steadier returns where regulatory frameworks are constructive. This follows from the broader observation that power has become the binding constraint for the AI buildout. 

The new limiting factor

Electricity is becoming the scarce input that ultimately sets the cadence of AI capacity. Several major hubs already show limited near-term headroom. Lead times for transmission, substations and interconnections are long. In cost terms, power availability and pricing now function as a new form of capital cost for operators. Locations that can secure long dated power arrangements or accelerate grid access are likely to take share. 

Concentration risk remains elevated

A narrow group of very large companies continues to drive a disproportionate share of index level returns. This raises portfolio fragility because a single theme can dominate outcomes despite the appearance of diversification. Broadening return sources across regions, sectors and styles can reduce dependence on a small leadership cohort and restore balance. This perspective aligns with previous Intrasia Wealth analysis on the risks of mistaking concentration for strength. 

Portfolio implications

• Favour paid layers of the stack. Suppliers that are paid early and predictably in the value chain, including chips, networking, power and cooling, tend to monetise before downstream adopters.

• Use volatility to improve entry points. High quality names can experience sharp price moves around headlines and earnings. Predefined accumulation levels and trim bands help capture dislocations without chasing momentum.

• Follow the bottleneck. Leadership often migrates with the constraint. As pressure shifts from compute to networking to power, opportunity shifts with it.

• Keep concentration in check. Balance mega cap exposure with infrastructure, utilities and selected materials to lower single theme risk. This is consistent with our earlier guidance on avoiding over reliance on narrow market leadership. 

• Insist on cash discipline. Emphasise companies that convert revenue into free cash flow and that allocate capital prudently, particularly in multi year capex cycles.

How Intrasia Wealth frames positioning

• Core allocation. Diversified exposure to semiconductors and high bandwidth networking, sized with volatility in mind and rebalanced methodically.

• Targeted satellites. Select positions in power systems, thermal management and specialty materials that link directly to data centre buildouts.

• Opportunistic sleeve. Add on idiosyncratic weakness where fundamentals remain intact. Harvest into strength to manage risk.

• Risk controls. Position sizing, staged entries and portfolio level drawdown limits to maintain resilience through policy and sentiment shifts.

Conclusion

The AI theme remains intact. Leadership is rotating toward the infrastructure that enables it. Intrasia Wealth’s view is that the next phase will favour mission critical suppliers with order visibility, pricing power and cash generation. Electricity has become the key physical constraint, which elevates power, networking and thermal infrastructure within portfolios. A selective, valuation aware and diversification minded approach is best placed to compound through this stage of the cycle.

Source: Morningstar