The global push towards artificial intelligence is entering a more challenging phase, as investors begin to look beyond headline spending numbers and focus on how these massive investments will be financed and supported on the ground. With AI spending estimates ranging between $300 billion and $500 billion over the next few years, market participants are increasingly questioning whether such outlays can deliver meaningful returns.
On ET Now, concerns were raised about the funding shift behind this investment wave. What was earlier expected to be financed largely through internal cash flows is now increasingly being supported by debt, prompting unease among investors about balance-sheet stress and future returns.
Addressing these worries, Jim Walker, Aletheia Capital said the concern is justified, as AI companies will now be competing directly with governments in debt markets. He noted that it remains unclear whether governments or AI spenders will generate better returns on the capital they raise, adding that governments often earn little to no return on their investments and that a significant portion of AI spending could face a similar fate. While investors may still be willing to fund AI expansion through debt or equity, Walker stressed that financing itself is not the primary obstacle.
According to him, the real constraint lies in the physical economy, particularly energy availability. While companies can build as many data centres as they choose, there is insufficient electricity in the United States to power them. Walker pointed out that US installed power capacity currently stands at about 4,000 terawatt hours and is projected to rise by roughly 800 terawatt hours over the next five years, representing close to a 20% increase. This is a dramatic shift for a sector that has seen virtually no capacity growth over the past two decades.
He added that labour shortages further complicate the challenge. The US lacks a sufficient pool of skilled workers needed to build power generation facilities, transmission lines, and distribution networks. As a result, the biggest hurdle facing the AI industry is not the promise of a new digital economy or the excitement around how artificial intelligence could transform workplaces and lifestyles, but the inability of the traditional economy—energy and skilled labour—to scale quickly enough to support this future.
ET Now also highlighted the heavy inflows into US technology and AI stocks in 2025 and asked whether the recent cooling in the so-called Magnificent Seven could trigger a reversal of capital flows towards emerging and other Asian markets.
Walker said such a shift is likely, arguing that many investors have made a poor bet on US technology stocks. He said the real investment opportunities over the next five to ten years will be found in sectors tied to infrastructure and power, including steel fabrication, cable manufacturing, copper, and electric utilities. These areas, he said, are more likely to deliver sustainable returns as the global economy grapples with the practical requirements of the AI era.He added that this environment creates a strong opportunity for emerging markets to stand out, particularly as investors refocus on domestic policy-driven growth. Countries such as China and India, he said, are well placed to benefit from this shift due to ongoing domestic changes. Walker said his firm’s global strategy remains long on both China and India, reflecting confidence in their internal growth dynamics rather than reliance on overstretched global tech cycles.
As enthusiasm around artificial intelligence meets the realities of funding, power supply, and skilled labour, market attention may increasingly turn away from pure technology plays towards the foundational sectors and economies capable of supporting the next phase of global growth.