Portfolio Manager, Rand Swiss

Viv is a senior analyst and investment specialist, focusing on international and local markets. He frequently appears in the media, contributing to channels such as: SABC, CNBC, SAFM, 702 and ETV. He is also a regular guest lecturer at a number of prominent business schools and advanced education programmes, and previously lectured at tertiary institutions such as UKZN and DUT.

You can follow Viv on YouTube at @RandSwiss for regular market updates and AI insights.

Beyond Nvidia: AI’s next bottleneck is energy

The defining feature of every major technology boom is not the breakthrough itself, but the constraint that follows it. Investors who identify that constraint early tend to capture the outsized returns.

In artificial intelligence, the first bottleneck was computing power. Graphics processing units became scarce, pricing power shifted decisively to suppliers, and Nvidia emerged as the dominant winner. More recently, memory constraints pushed companies such as Micron and SK Hynix sharply higher.

A new bottleneck is now taking shape. It is neither chips nor software. It is energy.

AI’s growing power problem

Training and running advanced AI models requires vast data-centre clusters that consume extraordinary amounts of electricity. Planned facilities over the next several years are expected to require gigawatts of power. One gigawatt is roughly enough to supply a mid-sized city.

The issue is not only rising demand, but timing. Hyperscalers can build a data centre in under two years. Connecting that facility to a high-capacity electricity grid can take five to ten years. The result is a growing interconnection backlog, rising power prices, and a grid ill-prepared for the scale of AI deployment now underway.

Energy, long treated as an operational detail, is becoming a strategic constraint.

Why batteries matter

Traditional solutions offer limited relief. Coal and nuclear projects are slow to commission and politically fraught. Renewable energy is scalable, but intermittent. That leaves energy storage as the most practical bridge between demand and supply.

Grid-scale batteries allow data centres to smooth peak demand, manage outages, and integrate renewables more effectively. They are becoming a critical part of AI infrastructure rather than an optional add-on.

A battery leader with a broader role

Contemporary Amperex Technology Co. Ltd, better known as CATL, is already the world’s largest battery producer. While its reputation has been built on electric vehicles, its growth opportunity increasingly lies elsewhere.

From data-centre backup systems to grid-level storage and autonomous infrastructure, CATL is positioning itself as a foundational supplier to AI’s physical layer. As energy constraints tighten, scale and reliability matter. Few companies are as well placed to meet that demand.

CATL is listed in Shenzhen and accessible via international platforms such as Swissquote.

AI is digital, but its build-out is physical

Much of the attention around AI remains focused on software and semiconductors. Yet the expansion of real-world AI infrastructure depends heavily on industrial inputs, particularly copper and steel.

Electric grids, charging networks, robotics, and automated logistics systems are all materially intensive. Supply growth in key metals has been constrained for years by underinvestment, permitting delays, and rising costs. As demand accelerates, pricing pressure is likely to follow.

A measured way to access the trend

For diversified exposure to copper at scale, BHP Group Limited remains a preferred option. The group combines large, long-life assets with a strong balance sheet and deep liquidity. It offers participation in the structural demand uplift linked to electrification and AI, without the volatility associated with early-stage technology stocks.

Portfolio positioning

Rather than relying on single themes or passive exposure, the Rand Swiss AI Portfolio is structured to identify and hold the key enablers of the AI ecosystem. Since its launch in June 2023, the strategy has delivered a cumulative return of +110.03%, compared with +56.36% for the S&P 500 Equal Weight Technology Index.

The approach is deliberate: allow winners to run, avoid forced rebalancing, and focus on long-term structural shifts rather than short-term narratives. That discipline has driven strong performance, albeit with periods of volatility that investors should expect in a concentrated strategy.

As AI evolves, the opportunity set is broadening beyond chips and software. Energy, storage, and materials are becoming increasingly central to the investment case.

Understanding where the next bottleneck lies may once again prove decisive.

Want to learn more? Sign up for our next AI: Quarterly Update webinar to stay ahead of the trends, bottlenecks, and stock picks driving the sector.