THE FACTS
- The global technology sector is booming, driven by excitement around artificial intelligence (AI).
- The Bank of England cautioned that AI-related equity market valuations appear stretched.
- McKinsey estimates AI could add up to US$4.4 trillion to global GDP annually.
With the global technology sector soaring again this year, driven by excitement around artificial intelligence, claims that a “bubble” exists have become more frequent.
Some critics argue that artificial intelligence (AI) darlings such as Nvidia have risen too far and too fast, that valuations are stretched and that investor exuberance, particularly around high-profile partnerships between AI names, suggests stock prices have become detached from the reality of future earnings.
“Bubble talk” makes headlines, and understandably sparks emotion – both from investors who have benefited from the surge in tech stocks and those who have missed out. So are these claims, which are often superficial in nature, overblown?
Every decade brings a new technology hailed as “the next big thing,” and sceptics are never far behind. Comparisons between the AI boom and the dot-com mania of the late 1990s are easy to make, but are simplistic and perhaps kneejerk – especially when one digs into what’s actually driving investor enthusiasm today.
Playing devil’s advocate, manias are often easy to identify in hindsight - and, in theory, should have been obvious at the time. During Tulip Mania in the Netherlands in the 1630s, some rare bulbs sold for more than a skilled worker’s annual salary or even the price of a house. In Japan’s late-1980s asset bubble, Tokyo land prices became so inflated that the Japanese Imperial Palace was estimated to be worth over US$1 trillion ($1.74t).
At the mainstream onset of the previous major technological breakthrough – the internet – traditional valuation measures were discarded altogether. During the dot-com boom, the traditional valuation metric Price/Earnings (P/E) came to mean “price-to-eyeballs”. Waves of investors had a avenous appetite for any company with “.com” in its name, regardless of revenue or profitability. Fear of missing out (Fomo) was rampant. While many warned of the excesses, they were drowned out by the prevailing euphoria.
Fast-forward to 2025 and the conversation feels much more nuanced.
The Bank of England recently cautioned that equity market valuations, particularly among AI-related names, appear stretched. It noted that the top five companies in the S&P 500 now account for roughly 30% of the index’s value – the highest level in 50 years – and that share valuations based on past earnings are at their most elevated in 25 years. However, the bank also acknowledged that valuations look less extreme when measured against expectations for future profits.
That distinction is crucial. The S&P 500 currently trades at about 25 times forward earnings, compared to a long-term average of around 15 times. Yet much of this premium reflects the larger weighting of high-growth technology firms, whose earnings are expanding at a faster pace. Nvidia, for example, trades on just over 30 times estimated forward earnings, but its revenues and profits are forecast to grow by roughly 50% through to fiscal-year 2026, and perhaps by a similar margin the year following. Unlike the late 1990s, today’s leaders are delivering strong fundamentals to justify investor optimism.
The Bank of England (BoE) did highlight legitimate risks: disappointing progress in AI capability or adoption, intensifying competition or bottlenecks in power, data and commodity supply chains, which could all trigger a revaluation. Conceptual breakthroughs that alter AI’s infrastructure requirements could also shift expectations. The bank’s caution perhaps partly reflects institutional memory – the BoE has acknowledged that it was too slow to recognise the credit bubble that preceded the 2007–08 Global Financial Crisis (GFC).
Still, the demand that investors are pricing into technology stocks today is real and measurable, particularly for AI chips and the associated infrastructure rollout. The largest AI research centres now generate close to US$20 billion in annual revenue. Unlike many firms during the dot-com era, these companies produce substantial profits and cashflows.
More importantly, AI is already delivering tangible economic value and productivity gains. Across sectors, companies are automating tasks, cutting costs and discovering new revenue opportunities through AI-enabled systems. From logistics to finance and healthcare, AI is optimising supply chains, detecting fraud and accelerating drug discovery. Even creative industries are being reshaped, with AI now composing music and designing advertisements. It is increasingly difficult to find a sector untouched by this technology.
Global consultancy McKinsey estimates that AI could add between US$2.6t-US$4.4t to global GDP (Gross Domestic Product) annually. These figures are grounded in real-world adoption. Expense-management firm Ramp reports that 44% of US businesses now pay for AI tools (up from just 5% in 2023), with average contracts valued at about US$530,000. Data from global payment platform Stripe show that the fastest-growing AI start-ups are reaching US$5 million in annualised revenue 1.5 times faster than the top software-as-a-service (SaaS) firms did in 2018. More broadly, US companies founded in or after 2022 are growing at 4.5 times the rate of those founded before 2020, with the advancement in generative AI one probable factor.
This explosive, monetisable demand supports a massive global investment cycle – building the infrastructure to power AI (semiconductors, data centres, cloud platforms, high-speed networks and the electricity to power it all). ChatGPT owner OpenAI alone has reportedly struck deals with chipmakers Nvidia, AMD and Broadcom worth as much as US$1t to secure the equivalent of 26 gigawatts of computing capacity, enough to power roughly 20 million US homes.
Comparisons between today’s data-centre buildout and the fibre-optic boom of the 1990s or the 19th-century railway mania are deceiving. This time, real, commercial demand already exists. AI applications have paying customers and data centres are running at near full capacity. The expansion is predominantly being financed by profitable, cash-rich companies – not speculative start-ups – with firms such as Alphabet (which owns Google) holding nearly US$100b in cash. And structurally, demand is likely to still be in its early stages: as every company seeks to integrate AI into operations, the addressable market continues to expand.
Scepticism in moderation is healthy, but there appears to be strong reasoning behind investor enthusiasm for AI. Those who dismiss AI as simply a “bubble” risk misunderstanding the scale and momentum of the foundational shift now under way.
As with any transformative technology, there will be both winners and losers, underscoring the importance of a highly selective approach to investing. As active managers, the investment team at Generate seek to identify companies with genuine, sustainable advantages while avoiding those driven by hype or unrealistic expectations. Through rigorous research, disciplined portfolio construction and continuous monitoring, the team aim to navigate the opportunities and threats presented by AI’s evolution – with the goal of continuing to capture long-term value while carefully managing downside risk.