Daniel Casali, Chief Investment Strategist at Evelyn Partners, the UK wealth manager, comments: “In the classic TV series Blackadder Goes Forth, private Baldrick famously writes a poem about the war that consists solely of the words “boom, boom, boom.” It’s short, loud, and a little ridiculous—but oddly fitting for the frenzied excitement behind today’s Artificial Intelligence (AI) investment landscape.
"The boom in AI has dominated headlines and equity market performance, and some of the numbers are staggering. Since ChatGPT (a conversational tool powered by AI) was launched on 30 November 2022, the US tech heavy NASDAQ composite has doubled to add $18 trillion in market value, far outstripping the performance of the broader S&P 500 index. This index includes all the major players in the AI theme, including Alphabet (the parent of Google and YouTube), Meta Platforms (the owner of Facebook and Instagram), Amazon, Oracle, Microsoft, and chip designer Nvidia.
“Given the scale of this move, concerns are rising over whether AI-related stocks are another financial bubble in the making. In October, the Bank of England’s Financial Policy Committee warned that ‘equity market valuations appear stretched, particularly for technology companies focused on Artificial Intelligence’. Elsewhere, Bank of America's October 2025 Global Fund Manager Survey showed that 54% of institutional investors believed the AI boom is a bubble.
“It’s notoriously difficult to identify a bubble before it bursts. Rather than trying to time the peak, investors may want to focus on whether the AI investment trend can continue. Despite the prevailing risk factors which are common with all disruptive technologies that have far-reaching and profitable applications, we find there are solid reasons to support near term growth in the AI theme.”
Here, Daniel Casali explores what those reasons are:
Improved company profitability
One reason why the AI boom could be extended is its potential to capture productivity gains to enhance company profits. The theme is supported by investment in physical infrastructure, like data centres, to facilitate the surge in data creation from sources such as smart devices, e-commerce, and digital platforms, to name just a few examples. This data is the fuel behind AI models.
As firms harness this data through machine learning and analytics, they unlock predictive capabilities, automate decisions, and personalise customer experiences - boosting profitability. The results are becoming apparent. According to Stanford’s 2025 AI Index, 71% of companies using AI in marketing report revenue gains, while 49% using it in service operations report cost savings. These operational improvements are tied to margin improvement.
The forward profit margin for companies in the MSCI All Country World Index is near a record high of 12%, supercharged since Covid in 2020, when data creation surged due to remote work and online shopping.
Financing the AI investment cycle
The durability of the AI investment boom is a critical question for investors. A pullback in investment could risk broader economic weakness and bring down valuations beyond AI-linked stocks. However, the current AI investment cycle appears to be in its early stages. Unlike the dot-com era, where tech investment built-up over a decade, today’s investment acceleration really began only around two years ago after the launch of ChatGPT.
Looking forward, Nvidia’s CEO, Jensen Huang, predicts that global AI data centre spending would reach $3 trillion to $4tn per year by 2030, against an estimated $0.6tn in 2025. At first glance this roughly 40% per annum growth in capital investment looks fanciful. However, from a financing perspective, JP Morgan reckons it is manageable given that the IMF expects the global economy to expand by 20% in nominal terms over the next five years.
Critically, the finances of US non-financial corporates today are in rude health compared to 25 years ago. The difference between cash flows and capital expenditures (known as the corporate financing gap) shows a 0.5% of GDP surplus. In contrast, when the dot-com bubble burst in 2000, such US firms were running at a 3.4% of GDP deficit. Essentially, financing should not be a constraint on the AI investment cycle itself.
Equity valuations have room to rise further
AI stock valuations have risen, but they’re not yet close to historic extremes. Using the NASDAQ composite index, trades on 30 times forward earnings - well below the 72 times seen during the dot-com bubble in 2000, when speculative growth expectations drove valuations higher. Back then, many companies were early-stage ventures with unproven models, unlike today’s AI leaders such as Microsoft, which have diversified revenue streams and established profitability.
Investors will be paying particularly close attention to the valuation of Nvidia, a central player in the AI boom. It currently trades on 30 times earnings, which does not look especially high historically. To justify this premium, Nvidia will need to continue beating analysts’ earnings expectations. Nvidia is expected to report its latest earnings on 19 November.
Aside from expectations of company fundamentals, the macro environment appears supportive with global monetary and fiscal policy being eased. There is also the potential of a relaxation in regulatory rules governing US banks that could further enhance money flowing into the financial system to provide an additional tailwind for equity valuations.
The outlook from here - risks to monitor
Taken together, based on continued investment, company fundamentals and valuations, it would be reasonable to expect the AI investment boom to continue. Even so, there are some key risks to watch out for:
Energy supply. AI development and deployment are intensely power-hungry, requiring vast amounts of electricity for both model training and inference. The International Energy Agency (IEA) estimated that data centres, where the AI model training and development occurs, used around 1.5% of global electricity demand in 2024 and is projected to double in their base case scenario by 2030.1 However, there is room for upside to that ratio. Under a “lift-off” scenario of a more rapid adoption of AI and greater demand for digital services, the IEA projects that data centres could use 4.4% of electricity demand by 2030.
Without a significant ramp-up in grid capacity and generation, especially in regions hosting hyperscaler data centres, the pace of AI adoption could slow. Investors are increasing recognising that the AI boom is linked to the ability of the energy sector to keep up with its power-hungry needs. However, for now it seems that the AI adoption rate in the US economy continues to creep up.
Geopolitics. Risks from geopolitical tensions between China and the US has the potential to undermine the AI investment theme. In October, the US has imposed tariffs and export controls on advanced tech exports to China, while China has retaliated with restrictions on rare earths - critical for chipmaking. This tit-for-tat dynamic has weaponised technology supply chains, with high-end chips and rare earths now central to economic diplomacy.
Adding complexity to geopolitics, China’s DeepSeek, launched in January 2025, has demonstrated that competitive AI models can be built at a fraction of the cost and energy of Western hyperscalers. DeepSeek is an open-source AI company that develops advanced chatbots similar to ChatGPT, offering high performance and free access, which has made it one of the most downloaded apps globally. If widely adopted, DeepSeek could render billions in hyperscaler infrastructure obsolete, undermining the need for greater AI investment.
Circular vendor financing. The AI boom is being supported by vendor-financed deals, where major suppliers like Nvidia and AMD help fund their customers’ purchases—creating a circular flow of capital. For instance, Nvidia has invested $100 billion in OpenAI, which in turn committed to buying its high-end designed chips (Graphics Processing Units). Effectively, Nvidia is using its own capital to fuel demand for its products. AMD has structured similar deals, offering OpenAI equity-linked incentives to purchase its chips.
Oracle, another key player, has agreed to provide $300 billion in cloud infrastructure to OpenAI, much of which is powered by Nvidia and AMD chips. These arrangements blur the lines between supplier and customer, raising concerns about artificially inflated demand.
To conclude: Riding the AI wave
The AI boom is making waves in both headlines and markets. Since the launch of ChatGPT, tech stocks have soared, adding trillions of dollars to asset values. While some investors may worry about a bubble, the ability for AI firms to boost profitability and fund the AI investment cycle suggests valuations are justified and the rally has further to run. However, the massive energy demands, geopolitical tensions, and uncertainty around vendor financing are risks worth monitoring. Nevertheless, as Baldrick might say, the AI story still goes ‘boom’.
|
