Yes, we believe it will. Excitement around artificial intelligence (AI) and its related technological advancements has been a key market driver since early 2023. Investors are rightly enthusiastic about the potential for AI to support labor productivity and growth. We share this enthusiasm, while acknowledging the uncertainty in quantifying these benefits. Despite AI’s long-run promise, we think recent exuberance has led to a disconnect in market pricing. We advise investors to take stock of valuations and portfolio concentration, and modestly tilt toward more reasonably priced segments of the market.
Transformative technological innovations fundamentally change how we live. Previous innovations led to impressive productivity gains in the 1960s and 2000s. In the 1960s, with significant advancements in semiconductor chips, we swapped clunky mainframes for faster, more convenient minicomputers. In the 2000s, penetration of the internet—followed by the advent of the smartphone—built on innovations from the prior decade. However, productivity gains in the 2010s were disappointing. While accurately identifying and measuring productivity drivers is fraught with uncertainty, the disappointing 2010s trend may be partially explained by the fact that innovations during this period were incremental rather than transformative. It may also be the case that gains were challenging to incorporate into data. For example, certain digital services that were previously paid for are now free on the internet, thus, understating economic output.
We think AI represents a transformative opportunity, which could lead to enduring productivity improvements. The proliferation of AI assistants is already underway. Higher profitability from increased efficiency will benefit businesses across the economy, particularly those relying heavily on manual in-office tasks. Currently, companies bringing AI solutions to market, such as chip manufacturers, are enjoying record revenue increases. Demand is also soaring for associated software, cloud computing, data centers, and power generation.
History shows that investors become extremely optimistic about future growth options from new technologies. This can lead to a market frenzy, which creates asset-price bubbles that inevitably culminate in bubbles bursting, much like the “dot-com” bust. That cycle provides an extreme example of exuberance; the US tech sector soared 53% on average per year in the five years ended March 2000 (the cycle’s peak). In retrospect, earnings growth forecasts were much too optimistic. At the cycle peak, analyst estimates for 2001 growth were 73 percentage points higher than the level of growth that was realized. Furthermore, peak-to-trough, equity investors lost 80% and it took 16 years to retrace to peak levels. A sobering reset indeed.
We do not think the current AI rally will follow the path of the dot-com bust, given businesses have healthy balance sheets and real earnings this time around. But we do think investors should diligently monitor the risk/reward balance going forward. Overall, valuation of the US tech sector—which makes up nearly 90% of the developed markets IT sector—has expanded from a 26x normalized price-to–cash earnings multiple in December 2022 to 41x today. This is staggering relative to the developed markets index, where the multiple only increased by 3.2x. Still, some of this optimism has been warranted. While AI darling Nvidia’s stock rally of over 600% in the same period and trailing price-to-earnings multiple expansion from 62x to 72x is incredible, results and forward guidance have repeatedly exceeded analyst expectations.
Nonetheless, investors should be mindful that uncertainty exists. One reason is that reality could disappoint versus expectations as there will inevitably be errors in forecasting the magnitude of productivity gains. Disappointment could come in many forms. For instance, we do not know how AI will displace existing technologies, rendering them obsolete; to what extent the labor market will suffer from lost jobs; or how regulations will curb use cases for AI. Another source of uncertainty is the time lag between the very large investments needed to support AI growth and realizing the intended benefits.
Taking this altogether, we advise modest tilts to market segments with more reasonable valuations such as small-cap and value equities. The normalized price-to–cash earnings multiple for developed markets small-cap equities, for instance, is the least expensive it has been relative to developed markets equities over the last 20 years. Similarly, the relative valuation of developed markets value equities versus developed markets equities sits at the 8th percentile of data over the same time period. Let’s not forget, these recommended strategies will also benefit from AI-driven productivity gains, even if their current prices don’t reflect it!
Sehr Dsani, Senior Investment Director, Capital Markets Research