1 July 2023
Last week, Nvidia, the maker of computer chips that are particularly useful in training large machine learning models, surged over 27% on a strong sales report. Further gains this week brought Nvidia’s market capitalisation to over one trillion dollars and its price-to-sales ratio to over 38 times. This means that at its current 12-month trailing revenue rate, even if Nvidia generated a 100% margin, it would take shareholders 38 years to recoup their investment through earnings.
While advances in artificial intelligence and machine learning look set to have a significant impact on business, economies, and financial markets, it is not clear that investing in Nvidia or other potential AI beneficiaries at current prices will be a profitable strategy. Bulls argue that we are facing a multi-year technological revolution and that the best-placed firms will enjoy strong growth for years to come. AI threatens to disrupt many traditional businesses, so investors need to gain exposure to this theme, no matter the valuation. Bears argue that at a 38 times price-to-sales ratio, all the potential growth and more has already been priced in the share price, and as new capital floods into chip production, current incumbents will be challenged by new competition.
Predicting the future of these technologies and their impact on the world is difficult, but we can look at historical analogies. Over the past 30 years, many popular investment themes have promised high growth and rerated stocks above 10 times their annual revenue, from the dot-com boom of the late 90s through cannabis stocks, crypto, self-driving cars, and AI.
We back-tested the strategy of holding stocks with a price-to-sales ratio above 10 times as a proxy for expensive, high-growth, thematic winners. We compared the performance of that strategy to the performance of the Russell 3000 below. Each month, the strategy picks all the stocks in the Russell 3000 with a price-to-sales ratio above 10 times and equally weights them.
Source: Bloomberg, Mergence
The chart above shows the rolling 12-month relative performance of this strategy. The first thing that stands out is that there are periods where these types of stocks can generate very strong relative performance, up to 200% outperformance during the late 90s. But the chart below of the cumulative performance of this strategy makes it clear that over the long term, this strategy has been a poor way to allocate capital, underperforming the broad market materially over the period from 1996 to 2023.
Source: Bloomberg, Mergence
For this reason, we believe that investing in already expensive AI beneficiaries and tech stocks is not an attractive prospect for any long-term holding period. Advances in machine learning promise huge efficiency gains as well as disruption for global business. We will continue to research and study the potential impacts, but as always, turning this into alpha will require variant perspectives and mispriced investment opportunities, something that we don’t think is currently represented by large cap AI stocks.
Our Market Snippets email aims to provide concise insight into our investment research process. Each week, we highlight one chart that showcases our research, motivates our current positioning, or simply presents something interesting we’ve discovered in global financial markets.
For more of our current market views, please visit our website.
