QQQ ETF: No Bubble Due To Productivity Surge But Downside Likely (NASDAQ:QQQ)

The Invesco QQQ Trust ETF (NASDAQ:QQQ) was trading around $437 at the time of writing, after gaining more than 40% during last year alone driven mostly by artificial intelligence. Market participants seem divided as to whether there is a tech bubble with some referring to events in 1999-2000 when volatility bit hard as shown in the chart below.

Noteworthily, the period before saw a productivity surge, and this thesis aims to show it is likely to be replicated this time around based on the potential of Generative AI, signifying we are not facing a tech bubble.

However, considering the uncertainty caused by inflation remaining sticky at above 3% together with the stock market’s fatal attraction with interest rates, amid inflated expectations relating to artificial intelligence, there are downside risks in the short term.

Downside Risks Due to Rate Cut Expectations and Some AI Hype

First, referring to my last thesis on QQQ in March last year, I had evoked the market’s irrational behavior mainly because tech stocks were advancing despite the high prevailing interest rates, while persistent inflation made it unlikely for the Fed to cut any time soon. Well, monetary policy has been kept unchanged, but, my Hold position proved wrong and the Invesco ETF gained 40% since then, because of investors’ love for the AI story.

In this respect, while the LLMs or large language models used by Generative AI are surely innovative, they inherit the bias problem, because it is people who train them after all. There are inaccuracies too. Improvements have been made with the latest version of ChatGPT, but, depending on the way it is used, it may also give rise to copyright issues as shown by the New York Times’ lawsuit against OpenAI for having used millions of its articles to train its algorithms.

Looking across the industry, there are traditional AI flavors like ML (machine language) and AGI (artificial general intelligence) which have proved useful in a wide range of use cases ranging from demand forecasts, automating customer service, optimization of inventory, and targeted marketing. These older technologies constituted around 91% of AI usage last year and have been around for years, and, are seeing renewed demand thanks to the ChatGPT eye-opener. Interestingly, not all require the latest fancy GPUs as demonstrated by China’s Baidu (BIDU) using older chips for its Ernie bot.

This is the reason why Nvidia (NASDAQ:NVDA) and Super Micro (NASDAQ:SMCI) which are selling the hardware or the very building blocks for AI may not necessarily see sustained demand and their future revenue growth expectations may be overblown. Thus, after their stocks have surged as per the table below, there is a higher likelihood of volatility in case of adverse market conditions.

On the other hand, some of QQQ’s other holdings like the AI infrastructure providers and enablers (as shown below) have not performed as well. Here, names like Amazon (NASDAQ:AMZN), and Microsoft (NASDAQ:MSFT) (which by the way is also a software beneficiary with Copilot) with their vast cloud platforms come to mind without forgetting Cisco (NASDAQ:CSCO) which is a key enabler for AI to be diffused across corporate data centers.

Next comes the software companies enhancing their products to use AI for generating more revenues like MongoDB (NASDAQ:MDB) and Adobe (NASDAQ:ADBE). However, once the product is AI-transformed we should not expect sales to automatically jump. Some investors learned this at their own expense when Adobe’s shares dipped by around 17% following lower sales and revenue guidance. At the same time, its Firefly Gen AI image creator tool faced some setbacks due to inaccuracies. Alphabet (GOOG) also encountered similar problems with Gemini AI.

Identifying a Bottom caused by a 17% Downside

Now, inflated market expectations as to the AI potential of some stocks combined with higher-than-expected inflation figures for February cast doubt on the ability of the Federal Reserve to cut rates in June. To this end, despite all the AI frenzy, when it comes to spending money, whether it is to build intelligent infrastructures, train algorithms, or develop software tools, CIOs are taking a more prudent approach before committing to new technology partners, according to Gartner. This may be caused by the cost of capital remaining relatively high.

Thus, monetary policy is a key ingredient that determines the performance of the market in general and tech stocks in particular because of their higher valuations. Hence QQQ trades at a P/E of 26.75x or 26.5% above the broader market or the SPDR S&P 500 ETF Trust (SPY) which itself is composed of nearly 30% of IT stocks. To identify a bottom, I look at QQQ’s historical performance during the Dot Com bubble which started bursting in March 2000 as charted below.

Here, one can notice that the 67% downside (from around $120 to $40) spanned about one year. In the immediate aftermath of the burst, there was a 33% downside to the $80 level, but this was temporary and the Invesco ETF settled around $100 for some time which corresponds to a 17% downside, which, by the way, is about the same drop suffered by Adobe.

Therefore, considering that the Nasdaq 100-tracking ETF is constituted of 57.64% of IT stocks, and currently trades at around $437, a 17% downside translates into a target of $362. This is relatively mild compared to the 67% bubble bursting and is justified by macroeconomic conditions not having deteriorated as expected in 2024, and GDP should still grow, albeit at a slower rate compared to 2023. Also, overall IT spending is expected to be better this year.

In addition, there is a productivity booster for the longer term.

Expect an AI-Led Productivity Surge

Thinking aloud, today’s situation is somewhat analogous to the 1991-1995 period when economic growth was sluggish and interest rates were at 5.5% (September 1991). However, things changed drastically in the second part of the decade when there was an acceleration in consumption, spending, and wage growth, all triggered by a productivity surge.

Looking deeper, this was enabled by new technologies such as computing, off-the-shelf software, and the Internet, all gaining rapid acceptance due to a tight labor market as corporations were forced to do more with fewer workers. Such a high level of innovation doubled productivity to an average of 2.5% in the 1995-2000 period, compared to only 1.5% in the first part of the decade or a 67% increase. It must be mentioned that some of this increase was also due to the third era of industrial globalization from 1989.

At that time IT investment as a percentage of GDP grew from 3% of GDP in the early 1990s to 4.9% at the end of 2000. This increased to about 8% in 2022 when comparing U.S. tech spending with GDP. In 2024, corporations should spend more on IT than last year as more capex is dedicated to cloud computing and Generative AI compared to general-purpose computing and software in the 1990s.

Noteworthily, in contrast to older flavors of AI like machine learning, Generative AI can understand natural languages (or those we commonly speak) and can potentially automate 60% to 70% of employees’ time compared to 50% previously according to analysts at McKinsey who further add that productivity gains ultimately depend on the rate of adoption and when Gen AI is aggregated with other technologies for example embedding it in existing software. Furthermore, in contrast to the 1990s when companies had to buy IT, today they can rent services by opting for cloud-based subscriptions which not only reduces the barriers to entry but can also accelerate time to market. Also, to use ChatGPT, you practically need no training but some practice to generate reports, in contrast to older versions of the technology where programming skills were required.

Again, according to McKinsey, for the U.S., which has one of the fastest rates of adoption of AI together with other developed countries, productivity growth could be 3.6%. This compares to the average of 1.6% in the 2019-2023 period according to the Bureau of Labor Statistics, thereby constituting a productivity increase of 125%.

Tellingly, a productivity figure of 3.6% is higher than the inflation rate of 3.2% itself which implies that AI-led gains in production output could offset the combined effects of deglobalization (on manufacturing) and supply chain overheads.

For this matter, one of the key factors that can determine the performance of the stock market today is interest rates and many have been pricing rate cuts. However, monetary policy will ultimately depend on the whole paradigm around supply-side inflation, and, to be realistic, we are living in a different world today, one characterized by deglobalization. The tendency towards more localization is exemplified by the CHIPS Act and Inflation Reduction Act which both encourage the consumption of “Made in America”, but, this will entail higher prices than “Made in China”. At the same time, the continuation of the war in Europe, a major conflict in the Middle East, and tensions in the Red Sea all add to transportation costs.

Therefore, as we venture into a period of low growth, sticky inflation, and sky-high governmental debt, a productivity surge could be the answer. Furthermore, in the same way as in the 1990s, this can lead to an increase in consumption and investment.

Downside Likely but No Bubble Bursting

In such a scenario, the Nasdaq could gain new heights, but for this to happen, output should rise enough to offset the effects of rising wages across all industries. Companies like AT&T (T) are already making it happen but it will take time to diffuse through the whole economy, most probably by 2025 judging by the 550K H100 GPUs Nvidia sold in 2023. Additionally, 1.5-2 million is anticipated to be sold this year, and it takes around 4-9 months for software companies to produce a stable version from the product conception stage.

In conclusion, this thesis has made the case for a downside, possibly around 17% caused by rates remaining higher for longer together with the hype built around the AI potential of some stocks. To this end, the price-to-earnings multiples of most of the stocks, except for Super Micro are either below or slightly above their five-year averages like for Taiwan Semiconductor Manufacturing (TSM) as shown in the table below. This shows that not all tech valuations have reached bubble status, but a double-digit downside can be caused by a knee-jerk reaction when a sell-off in one stock becomes contagious to others.

Still, for the longer term, Gen AI should deliver real productivity gains, especially important in a period of lower economic growth. As such, I do not foresee a bubble that can destroy 67% of QQQ’s value as was the case in 2000. Instead, it could suffer from a downside, with the 17% figure being more than half the roughly 40% slide suffered from its November 2021 peak to the December 2022 trough because of interest rates being hiked aggressively.