Investing in AI: is the hype still worth the opportunity?

Artificial intelligence (AI) has been a key driver of market performance over the past few years. As portfolio managers, we’re focused not just on what’s worked, but on where the next phase of opportunity - and risk - may lie. In this piece, we explore whether AI is still investable and where the next generation of AI winners could come from.

What are the risks and rewards of investing in AI?

AI promises major societal and industrial productivity gains, but also raises concerns over criminal use, data security, plagiarism and threats to creativity, accuracy, democracy and defence. 

It’s the most exciting investment theme of the day and the amazing share performance of its leading companies now means that, although fast growing, on traditional valuation measures they are very expensive. How do we position our client portfolios from here? We can all see what’s done well over the last few years. What do we think will do well in the next five?

Why has AI suddenly exploded in popularity? 

Recent AI market mania traces back to the launch of ChatGPT by OpenAI. ChatGPT became the fastest downloaded app up to 2022 - suddenly everyone was talking about AI.

However, AI isn’t new. The term was first coined in 1956 by a small group of technologists. So why has it suddenly exploded in popularity? The simple answer is the exponential increase in computing power*.

Finally, having the power to make machine learning work how technologists envisaged 70 years ago has unleashed enormous amounts of investment in AI. 

In 2025, just five companies - Amazon, Microsoft, Meta, Alphabet and Apple - plan to spend a record US$375bn on AI, up 70% from last year. Saudi Arabia has also pledged hundreds of billions towards AI chips and infrastructure. Explosive growth has also driven huge share price performance. The Bloomberg Mag 7 index (which comprises Microsoft, Apple, Nvidia, Amazon, Alphabet (owner of Google), Meta and Tesla) is up by 377% over the last five years. Nvidia, the poster child of the AI revolution, is up 1,524% over the period. If you exclude this set of companies, the rest of the US market is up a far more pedestrian 81%, which compares with the UK’s FTSE 100 equivalent return in US dollars of 96%. So, the phenomenon of “American exceptionalism” is entirely about the Mag 7, not the wider market. 

Are AI stocks too expensive?

However, even with growth, valuations have become very expensive. 

A warning from history: in early 2000, Cisco was the leading company behind the leading investment theme of the time – the internet. It was also very expensive, trading at 13x forward revenues with a market cap of US$259bn.

Today, Nvidia occupies a similar position for AI – the dominant provider to the dominant theme – and it's even pricier, valued at 14.3x estimated forward sales and a market cap of US$3.2tn.

What happened to Cisco? After losing around 90% of its value by late 2002 as the internet bubble burst, it eventually recovered to a market cap of US$246bn today – despite sales rising from US$18.9bn in 2000 to nearly US$54bn in 2024. Impressive revenue growth, but disappointing share price performance from its peak.

The recent release of China’s DeepSeek – reportedly far cheaper than Nvidia’s tech – offers a reminder of how quickly market leaders can be challenged. Nvidia’s share price fell 35% over two months before recovering.

As the graph below shows, Cisco’s history in the 2000s shows why valuation is a crucially important consideration for investors in growth companies. 

Will the next big AI companies be outside the names we know?

As we look ahead, we shouldn’t focus solely on what has already done well – that was the key lesson from the technology mania of the late 1990s. 

At the end of 1999, Amazon’s market cap stood at US$28bn. Today it’s US$2.2tn – a staggering 7,857% increase – but along the way, it lost over 90% of its value between 2000 and late 2001. 

Alphabet (Google) wasn’t even publicly listed in 2000, arriving in 2004 with a US$23bn valuation. It’s now worth US$1.9tn – an 8,261% rise.

We believe that as AI matures, many of the long-term winners may be companies we don’t yet know – or that aren’t even listed. Some of the best-performing investments of the past 25 years simply weren’t available during the last tech boom. And history shows that early hype often hides where real, lasting value lies. AI may prove no different.

Are we underestimating China’s AI potential? 

The next winners also may not be US or European companies. 

In a world of strong US-China competition, we should note that the Chinese university system produces around 3.6m STEM (science, technology, engineering and maths) graduates every year. It holds over half the world’s patents in quantum computing and 70% of AI patent filings. 

China is a difficult market for western investors for several socio-economic and political reasons, but it is likely that Chinese companies will be major players, if we are able to access them in future.

Why might the biggest long-term AI winners not be obvious yet?

AI is a transformative technology, in a way few other leap-forwards have been, comparable to the steam engine, electricity or the internet. 

In many of these seminal advances, the early leaders in the field became popular short-term investments, but the real beneficiaries of the revolution came later, and the earlier leaders were left behind as monetisation of the new technology moved in directions not immediately apparent. 

AI is changing the world - but patient, selective investing has always been the most effective way to capture lasting change for clients. As AI keeps evolving and it matures as an investment theme, we’re focused on cutting through the hype to invest on client’s behalf - not just in what’s working now, but in what we believe could lead the way next.

*In 2000, Intel’s Pentium 4 CPU had 42 million transistors. Today, Apple’s M3 ultra CPU has 184 billion transistors, an increase of almost 438,000%.