Analysts at McKinsey take the view, as many do, that 2023 was the “breakout year” for generative artificial intelligence (gen AI). If nothing else, the performance of the tech-heavy Nasdaq over the past year suggests they aren't exaggerating. No company has attracted more column inches than AI chip designer Nvidia (US:NVDA), but the market stampede has also benefited established tech plays such as Alphabet (US:GOOG) and Microsoft (US:MSFT), whose growth prospects had previously been brought into question given their maturing marketplaces and increasingly tough year-on-year comparatives.
An accelerated rollout of gen AI has changed perceptions, yet any ramp-up in valuations on this scale invariably gives way to fears over the creation of a new asset bubble. It may be that the technology now commands the attention of business leaders as much as it does programmers, but it’s still early days. Organisations that have already embedded AI capabilities have understandably been the first to explore gen AI’s potential, but McKinsey says that “adoption remains concentrated within a small number of business functions”.
That seems at odds with much of the media coverage, which has tended to overstate the potential impact of the technology, at least over the short to medium term. Then again, back in the early 1990s, it would have been difficult to predict the extent to which the internet was to change commerce, news dissemination and so many other aspects of our lives. It might have felt prudent to wait until the dust settles from an investment perspective, but the past 18 months show people want to capture as much of the early growth of stocks like Nvidia as possible, thereby raising the risk of a steeper correction if the rollout of gen AI falters.
It may be, however, that even if tech ratings were to suddenly pull back sharply, the long-term prospects of a company such as Nvidia would hold up due to competitive advantages that are hard to replicate, so competitors would struggle to catch up to the company’s head start in its key competencies. It’s difficult to make a definitive call as to whether Nvidia possesses an unassailable moat, but when Arthur Sants outlined the investment case in late 2022, he made the point that although “new technology is often first adopted by enterprises, demand often kicks off when consumer applications appear”.
Some of those applications relate to the investment industry itself. But, as with other areas of the economy, the potential impact of gen AI and big data on quant and traditional investment strategies will elicit widely differing viewpoints. A review of machine learning experiments published within the International Journal of Data Science and Analytics contrasts the seemingly accurate market forecasts generated by many academic research models with real-world examples of machine learning-driven funds. Whether the technology will give rise to unique market insights is still open to question, but the review concedes that the numerical nature of financial markets provides an ideal environment for machine learning.
Ultimately, the potential cost benefits linked to the research function could tip the balance in favour of gen AI. Many large asset management firms already employ AI to synthesise large datasets sourced from books, academic journals, websites, and other sources within the public domain. Theoretically, the employment of gen AI should accelerate the process, although the fact that some funds obviously believe they can achieve alpha returns in this manner could stand either as an affirmation or repudiation of the efficient markets hypothesis, depending on your perspective.
On a tenuously related note, it's generally held that the performance of some funds can be hampered by their sheer size. Increased assets can lead to liquidity problems because their managers are sometimes forced to take outsized positions in companies to ensure they are meaningful in relation to the portfolio's overall assets under management. The view also exists that if a fund gets too big it can become effectively unwieldy, therefore less able to generate excess returns relative to the benchmark index.
There is a potentially strange parallel here with gen AI. It could be argued that alpha returns will be harder to come by if the technology becomes the default research tool in managed money. A corollary to this proposition is that capital allocations will become more uniform across the industry. The widespread adoption of the technology could therefore conceivably give rise to opportunities for contrarian strategies, although the danger also exists that it could silt up equity markets if investment perspectives become homogenised. Perhaps the emerging technology would be more fruitfully employed as part of a "smart beta" strategy – that is, adding something to passive strategies rather than replacing active funds.
