AI remains the decade’s most disruptive technology, with cloud platforms and semiconductor suppliers poised to benefit most.

Equity markets are entering a decisive phase of the artificial intelligence (AI) investment cycle; this is one marked by unprecedented capital expenditure, surging energy needs, intensifying competition among tech giants, and early signs of stress in parts of the ecosystem. Amid these dynamics, the AI megatrend potential remains intact. Here, we look at five reasons why this is the case, and which of these sectors are coming to the fore.

1. Valuations are high but not in bubble territory

One of the primary concerns remains whether AI stocks have entered bubble territory. While tech valuations are elevated, they remain far from the extreme multiples of the early-2000s’ dot-com bubble. The Nasdaq 100 is trading at around 26x forward earnings, representing just an 18% premium over its 10-year average, and significantly below the more than 50x seen during the early 2000s peak. Earnings, not higher valuation multiples or speculation, remain the primary driver: the tech sector is expected to deliver +25% earnings-per-share growth in 2026, with semiconductors leading at +46%.

2. Strong consumer demand with slower take-up by companies

Consumer adoption of AI applications is accelerating at historical speed. ChatGPT now has over 800 million weekly users, while Google’s Gemini has in excess of 650 million monthly users. That being said, take-up by companies is lagging behind due to privacy and accuracy concerns. This creates uncertainty for enterprise software companies which have underperformed this year and are yet to show a revenue acceleration driven by AI. OpenAI is playing a critical role in the AI ecosystem and has signed over USD 650 billion worth of contracts with cloud computing providers. However, its long-term projections, including the objective to generate an estimated USD 200 billion in revenue by 2030, appear highly ambitious and may prove unrealistic. The firm could also face a cumulative USD 115 billion cash burn between 2025–2029, posing risks for partners heavily dependent on its growth, most notably Oracle, which has a USD 300 billion cloud-capacity contract tied to OpenAI.

3. AI capex super-cycle still in early stages

Meanwhile, tech companies are aggressively investing to avoid falling behind in the ongoing AI ‘arms race’. Total capital spending from cloud leaders, i.e. Microsoft, Amazon, Alphabet, and Meta, is projected to rise 34% in 2026, fuelling robust demand for AI chips, networking, liquid cooling and cloud infrastructure. Nvidia’s data centre revenue is expected to surge from roughly USD 115 billion in the fiscal year 2025 to close to USD 483 billion by 2030 on the back of escalating GPU demand. However, the speed of investment raises questions about monetisation, depreciation, and financing. To justify an incremental USD 368 billion investment in Nvidia chips with a 10% after-tax return, Nvidia’s customers will need to generate around USD 1.4 trillion in additional revenue or cost savings.

4. Power demand: AI’s hidden constraint

In the AI ecosystem, electricity supply has emerged as one of AI’s most critical bottlenecks. Large data centres already consume around 3% of US electricity, a figure expected to rise above 8% by 2035. Grid capacity in the US is tightening rapidly, with capacity-auction prices rising nearly tenfold in two years. Some estimates point to a 107–200 GW power shortfall by 2030, equivalent to more than 100 nuclear reactors. These forecasts will underpin utilities with nuclear and gas-generating assets, as both stand to benefit. Increasingly, companies such as Vistra, NRG Energy, Entergy, and NextEra are signing long-term supply contracts with hyperscale data centres, which are locking in decades of steady cash flows.

5. Natural gas: the quiet winner of the AI boom

Similarly, natural gas power stations today account for nearly 40% of US data centre needs and are expected to remain the primary fuel source through to at least 2030. Gas producers anticipate an additional 7–10 billion cubic feet per day of demand by the end of the decade, with 40% of that growth coming from AI data centres. The largest US natural-gas producer, EQT, is highlighted as a top beneficiary. Recent long-term supply deals may support up to 10 GW of new power capacity, generating strong free-cash-flow growth through 2029. 

The cloud, semiconductors, and infrastructure: key investment opportunities

While we favour the technology sector, cloud-computing giants like Microsoft, Amazon, Alphabet, and China’s Alibaba are, in our view, well-positioned to monetise AI, benefiting from their substantial moats, strong unit economics, and growing backlogs. Semiconductor companies providing AI chips, especially Broadcom, appear poised to capitalise on AI adoption, with Broadcom's custom AI chips being developed alongside top cloud providers like Google. Nvidia currently controls more than 80% of the AI chip market but may face market share erosion as the AI landscape evolves. Networking companies are set to thrive as the data volumes generated by AI data centres continue to increase. Celestica, for instance, is becoming a significant player in providing custom networking hardware for Google’s AI infrastructure. Additionally, AI-driven demand could boost long-term returns for power suppliers, including Vistra, NRG, NextEra, and Constellation Energy.

While risks exist – from overoptimistic projections to rising debt loads and energy constraints – we believe the structural drivers remain powerful.

For long-term investors, diversified exposure across cloud platforms, semiconductors, and power-infrastructure providers may offer the most resilient positioning in an era defined by AI-driven growth.

The opinions expressed herein are correct as at 11 December 2025 and are subject to change without notice. Any forecast, projection or target, where provided, is indicative only and is not guaranteed in any way.