AI Enthusiasm and the Illusion of Fundamental Value
There’s a growing concern that the current enthusiasm surrounding Artificial Intelligence (AI) is driving stock market valuations more by sentiment and psychological biases like excessive optimism, and overconfidence than by financial fundamentals. This overvaluation bias increases the risk of companies undertaking large, ultimately unprofitable, investment projects.
This discussion focuses on the interrelation of two key issues: capital investment and market valuation, particularly within the context of AI.
The Allure of AI Investment
Four of the largest U.S. companies actively engaged in AI development – Amazon, Alphabet, Microsoft, and Meta – have announced plans to significantly increase their capital investments, by a combined 45%. This substantial increase in investment shows tremendous optimism for the technology. Amazon’s portion of this planned capital expenditure (Capex) increase is the largest, representing almost a third of the total.
At the same time, the U.S. stock market has remained relatively steady, despite economic and political changes, with the S&P 500 only declining about 3% since the January 2020 election. However, this market stability could be masking underlying risks.
Mispricing Risk and the Dot-com Bubble
The conventional wisdom in finance dictates a corporate decision generates fundamental value if the net present value (NPV) of the associated cash flow stream is positive, or at least zero. This pertains to fundamental value, not necessarily market valuations. Therefore, it’s crucial to distinguish between these two concepts and to recognize when market valuations stray significantly from their corresponding fundamental values.
When significant gaps exist between market valuations and fundamental values, mispricing risk arises, additional to fundamental risk. During the dot-com bubble, this mispricing risk was extremely high. Overvaluation fueled many investments with negative NPVs, eventually leading to failures when the bubble burst. Investors should carefully consider this historical precedent when they evaluate AI-based capital investments in the current environment.
The NPV of any capital investment directly changes the fundamental value of the company. The firm’s overarching fundamental value is determined by the cash the company generates over time, which is available to both the company’s creditors and shareholders. This cash is termed the company’s free cash flow. The fundamental value of the company’s stock is derived from the portion of the free cash flows allocated solely to its shareholders.
Amazon as a Case Study
The case of Amazon provides a useful view of the degree to which its stock prices reflect psychological biases, rather than underlying fundamentals.
A minority group of sell-side analysts use free cash flow analysis to value Amazon’s stocks. At one time, the analyst team at Credit Suisse (now part of UBS), was one of them. Examining the history of their past and recent reports provides important insights into whether Amazon’s current stock market valuation is an accurate reflection of fundamentals.
On Feb 27, 2025, Amazon’s stock price closed just under $210. The UBS analyst team assigned a target price of $230. In October 2019, the team, led by Stephen Ju, established 2025 as the terminal year for their free cash flow analysis. In doing so, they divided their valuation number for Amazon’s stock into two components, one for the five-year period 2020 through 2024, and the second for 2025 and after.
It’s critical to understand that in 2019, for Amazon’s stock, the analyst team assigned approximately 80% of its value to the period beginning in 2025. As a result, the long-term issues, not the short-term, day-to-day market fluctuations that investors often obsess over, mostly drove the fundamental value of Amazon’s stock.
The significance of this long-term approach highlights why discussions about Capex are so critical, as these expenditures significantly impact long-term free cash flows.
Market Psychology and Analyst Forecasts
Market psychology ultimately determines the degree of rationality in the stock market. This psychology affects analyst estimates of free cash flows, and it affects investors’ responses to market events. In a rational market, if analysts’ forecasts of future free cash flows prove reasonably accurate, then their valuation estimates should be relatively unbiased.
Accuracy of Forecasts
Consider two critical questions:
- How accurate was UBS analyst team in forecasting Amazon’s free cash flows for the period 2020 through 2024 in October 2019?
- How accurate was their forecast of Amazon’s market capitalization at the beginning of 2025?
Based on analysis of the UBS analyst report and Amazon’s financial statements, the UBS team significantly overestimated Amazon’s free cash flows for the period 2020-2024. The Covid pandemic may have played a part in this. The ratio of the average annual free cash flow forecast to its actual value was four, implying a 300% optimistic bias.
If the analyst team had used the actual free cash flow values for their 2019 valuation, their target price would have been a mere 25% of that in their October 2019 report.
Further, the analyst team’s 2019 calculations provide an estimate for the value of Amazon’s stock at the start of 2025. The ratio of the actual value to the estimated value is 1.63, meaning the current market value of Amazon’s stock is 63% higher than what the 2019 analyst calculations suggested.
It’s critical to juxtapose the answers to these questions. The UBS team severely overestimated Amazon’s free cash flows and intrinsic value by 300% during the period 2020 through 2024, and yet in 2025 Amazon’s stock price is 63% above UBS’ valuation estimate from 2019. This outcome would be highly unexpected ina rational market. The evidence indicates that the market is far from rational, and that is the central point.
Capex and Bias in Forecasts
A significant reason why the analyst team overestimated Amazon’s free cash flows beginning in 2020 is that they underestimated Amazon’s Capex. In fact, they assumed that Amazon’s property, plant, and equipment would decline over time, eventually approaching zero.
The assumption of declining PP&E is unrealistic for a company like Amazon and has no precedent in Amazon’s history. It contradicts Amazon’s stated investment policy back then and, certainly Amazon’s massive planned investment described above. The speculation as to why the analyst team would make such an assumption remains, although one possibility is in considering that for a given level of cash flow from operations, lower Capex logically translates to higher free cash flows. Therefore, lower Capex, with all other factors remaining equal, leads to higher valuations.
It’s worth noting the balance sheet that the analyst team presents in their 2019 report omits values for the line item PP&E. Indeed, it’s the only line item in the balance sheet with blank entries. Investors interested in the analyst team’s forecast for PP&E would have to compute the numbers themselves.
In this regard, when developing target prices, analysts can utilize upwardly biased estimates of free cash flow to capture irrational exuberance, through a nontransparent technique. In 2019, the Credit Suisse/UBS team underestimated the magnitude of this bias.
Consider this fact: Suppose the analyst team had used the actual figures for free cash flows in 2019 to forecast the intrinsic value of Amazon stock at the outset of 2025. They would have arrived at a value that is one ninth of Amazon’s actual market capitalization. In other words, had the team used estimates close to those of realized Amazon free cash flows (rather than their actual estimates), it would have undershot Amazon’s 2025 market cap by almost an order of magnitude instead of by 63%.
The difference between Amazon’s market cap and the estimate of its fundamental value, based on actual free cash flows, is significant. This difference reflects the extent of bias in current market valuations for tech stocks, in particular for companies undertaking major AI-related capital expenditures.
The Sentiment Surge of 2017
In 2017, technology stock prices began to increase dramatically, reflecting a clear shift in investor excitement about this sector. That year Greg Ip, now the chief economics commentator for The Wall Street Journal, created a video titled “The Economy Needs More Amazons.” The website’s video description states: “The U.S. has observed a long period of low stock-market volatility and economic stagnation. The Wall Street Journal’s Greg Ip believes the economy and markets could benefit from more risk-taking companies – creative disrupters like Amazon.com.”
Amazon is a good company. However, irrational exuberance can, at times, lead to stocks of good companies becoming bad stocks. Amazon has been a popular stock over the last five years, returning an average of 17% per year. This is well above its required return on equity of around 10.5% (according to the analyst team’s October 2019 report).
The key point is that this return has not been generated by Amazon’s free cash flows (the basis of its fundamental value); instead, it was generated by investor sentiment, a biased reaction to underlying fundamentals. Remember that the analyst team severely underestimated the rise in price of Amazon’s stock, even while overestimating Amazon’s cash flow stream.
Conclusion
Investing in tech stocks like Amazon involves bets on future sentiment along with fundamentals. In an environment of very high expectations for AI and large capital expenditures, investors should understand the degree to which they are relying on sustained and increasing optimism to generate attractive returns.
Analysts, investors, and corporate executives should prioritize the differences between market capitalization and corresponding estimates of fundamental value. Sadly, there’s little evidence they are doing so.
