Learnings from the CFA Investment Roundtable

Tuesday 23 June 2026

By and large, artificial intelligence has been the dominant force in global markets in 2026. US indices are trading near all-time highs despite the conflict in the Middle East and a subsequent energy supply shock. The S&P 500 is at its most concentrated since the dot-com bubble in the late 1990s. The 10 largest stocks in the index - which include the AI hyperscalers Alphabet, Amazon, Microsoft and Meta - make up around 40% of the index’s total market capitalisation, up from roughly a quarter three decades ago.   

 The inaugural CFA UK Investment Strategy Roundtable Flagship Series was hosted on 3 June with one theme in mind: is the AI-led rally sustainable? Three Co-Chairs and four guest speakers discussed under the Chatham House rule whether they thought AI was a bubble, what investors can do to position their portfolios, how shifts in the market’s structure have affected asset prices and their predictions going forward.  

 
Is there an AI bubble? 

The roundtable was divided on whether the AI boom was a bubble or a secular bull cycle. “The economics of AI doesn't make sense at the moment,” one speaker said. They raised three key challenges: first, how long can hyperscalers sustain massive outlays without making a return; second, whether other layers of the AI stack beyond semiconductors would become profitable and third, there was no evidence yet that the technology was generating enterprise value for corporate America. “I would say the economics doesn't make sense yet, but we don't think it's a bubble. We think that a correction might be actually needed at some point,” they said. 

One speaker took a more bullish angle on the technology’s prospects. “I think the economics make a lot of sense,” they said. “Consumers are irrelevant in this situation, because consumers will end up consuming products that have AI in it”. Robust enterprise demand for AI would buttress the ability of hyperscalers to recoup their investments: “It is very likely that you can get that back, because all of a sudden, we've got enterprise demands for compute that are way beyond supply,” they said. That demand was in turn fuelling the data centre buildout. “There is no visible end in sight to thedemand from hyperscalers,” said another panellist.  

Stock market valuations are currently justified by the fundamentals, one speaker said. “Even if you’re looking at the 10-fold returns on Nvidia or Samsung or [SK] Hynix, their profits are up roughly the same amount,” they said. However, they added that there were signs of “manic behaviour” with large retail participation using leverage. In their view, one scenario that could stop the mania would be supply constraints impeding data centre construction. “If you can't finish the data centre, and there are already delays, then you can't spend money, and if you can't spend as much money as they want to spend, then the entire supply chain will suffer,” they said.  

Another panellist noted echoes of the dot-com era. “There’s a real deep fear of missing out (Fomo)… and it’s full on amongst the AI participants and among investors,” they said. Meta’s CEO Mark Zuckerberg remarked last year that he would rather misspend “a couple of hundred billion dollars” than risk falling behind his competitors. With forecasts suggesting that hyperscalers could spend roughly $3 trillion on AI by 2030, the panellist said that investors had to consider what a realistic return on investment would be. “When we think valuations on the growth side, the growth stocks that we're seeing as driving AI are very high, are stretched and you've got a lot of risk there,” they said.  

Several speakers highlighted that AI investments are supported by hyperscalers’ free cash flows. “What is interesting is what happens next, and it's that gap that's going to come between monetising those investments, getting the return on the investments, and running out of free cash flow,” one panellist said. They added that some hyperscalers could start to “look like a utility” rather than a growth stock because the infrastructure for AI has to keep being built. Another speaker pointed to what happened when the dot-com bubble burst: “These companies [that had] the biggest valuations, they [turned] into this type of utility that’s never cash flow positive and it can have dreadful returns.”  

How should investors position portfolios to protect themselves?  

One risk that investors should consider is over-exposure to AI. “There is a lot of circularity in AI, and a lot of concentration in just one thing in a lot of portfolios,” one speaker said. “When I talk to clients, they are often invested in US equities… and then on the credit side, you will have quite significant exposure on the AI theme.Then, if you're invested in private assets, you will have infrastructure, private credit, all of that is equally exposed to AI”. Investors that want to hedge against an AI-pervasive market environment should look at having a global portfolio, reduce concentration in any one single investment theme and ensure resilience in different macro scenarios, they said. “European equities, for once, are a good diversifier. They virtually have no AI”.   

Attractive investment opportunities also exist beyond the AI trade, a panellist said. “In fact, outside of the US… there's loads of opportunity to pick very good companies at very cheap valuations,” they said. In addition to European equities, the speaker recommended consumer staples and pharmaceuticals. “When you think that way, then you see all of these opportunities, and you don't have to think, ‘is it going to benefit from AI?” they said. “But we do think, ‘is this business disrupted because of AI? Is it fundamentally impaired?’ But most companies we talk to are going, ‘AI is a fantastic opportunity to improve process, deliver better service’”.  

Another speaker said that opportunities exist within AI for investors to benefit from disruption. “I think you can still stay within the AI space but really look for where the AI use cases will improve profitability,” they said. “I think that will be in places like banks, because they have a lot of data, a lot of customers, a lot of processes”. The speaker added that they saw similar opportunities among logistic, transportation and medical insurance companies – where insurance premium revenues are rising while claims remain steady. “These sorts of things are where the opportunities might be for companies that have lower margins, they can expand their margin and that operational leverage produces the strong earnings returns,” they said.  

 
Does market structure matter?  

In recent decades, structural changes in markets have channelled more capital in index-tracking vehicles. That has had an amplification effect on current asset prices, some speakers said. One panellist said that there were fewer contrarian investors today that could act as a counterweight to the market. “You've got this inelastic market behaviour where you can go a lot further from the general perception of fundamental fair value for a lot longer and that’s partly driven by passive [investing] and quant firms… trading price movements,” they said. Another major change was the shift from defined benefit to defined contribution schemes, making capital allocation more automatic and index driven. “I would say that for 20 years you've had lots of people saving into US equities, so it just drives constant flow into that market, regardless of what happens to some extent,” they said.   

Passive investing has grown substantially over the past three decades, accounting for over 50% of total equity investing in mutual funds and ETFs globally. That rise has meant active managers face a difficult choice between being underweight AI-related stocks or continuing to follow the benchmark higher. “You've got passive coming in and just reinforcing that passive has to buy the biggest stock,” another panellist said. “Where's the risk? I think it was Edward Chancellor who correctly said ‘passive investors are prisoners’. They’re prisoners of the shape of the index. Active managers are in that dilemma. We can either take career risk because we might massively never perform for a long period of time, or we can stick to our guns and maybe go down in a blaze of glory”.  

The long view 

 One panellist expects a “relatively benign” base case for the next 12 months depending on the reopening of the Strait of Hormuz, the vital energy chokepoint that has remained effectively shut since the war in the Middle East began. “I would say invest in AI but be mindful as well of the risks around it and the overall concentration in that theme,” they said. The speaker added that “the world is becoming less globalised, more fragmented, more focused on resources.” Investors should think about how supply chains change as a result and try to invest with “high conviction”, they said.  

Another speaker agreed that investors should “definitely” be invested in AI. “I do think that looking for better value and where the future profitability surge is going to come, it will come from AI use cases. I think about which companies have a lot of data, a lot of processes, a lot of customers and are spending to find uses for AI,” they said. However, the speaker added that inflation risks were on the horizon: “Every country is going to create their own ecosystem, as opposed to the world being globally efficient. That’s well underway on defense, on manufacturing, on AI”. For UK investors in particular, geographic diversification could help mitigate the impact of any potential currency crisis from rising bond yields.  

“We don't macro forecast, but we are concerned about the cycle,” another speaker said. “There's companies out there priced at 100 times sales, and there are companies out there priced at two times earnings. The dislocation is massive”. They added that investors did not necessarily have to invest while markets are euphoric with soaring valuations.  

 “If you waited until some of the heat came out of the [dot-com] market, you were able to pick up the winners and do very well afterwards,” they said.