The persistent question echoing through financial markets today — ‘Are we in a bubble?’ particularly concerning artificial intelligence — finds a nuanced answer when scrutinizing historical data and market dynamics. Recent analysis, drawing from a Q3 Review & Update call for RWM clients, suggests that while concerns about market concentration and AI investment are valid, the current environment differs significantly from the dot-com era’s speculative fervor.
Market Concentration: A Historical Perspective
Much ‘kvetching,’ as one analyst humorously put it, has focused on the dominance of the S&P 500’s ‘Mag Seven’ or ‘big five.’ The top five companies currently account for approximately 27% of the index, a level comparable to the late 1960s and early 1970s. However, the trajectory of concentration since then reveals a critical distinction. The subsequent three decades saw a decrease in market concentration, primarily attributed to a more rigorous antitrust enforcement regime. During this period, ‘giant conglomerates were not in favor; M&A was cautiously watched,’ and vertically integrated industries were carefully monitored to prevent consumer disadvantage.
This trend reversed dramatically after a ‘very significant regime change in M&A and antitrust enforcement in the late ’80s and early ’90s.’ Over the past 15 years, the situation has ‘really gone postal,’ with the Magnificent Seven alone engaging in 846 mergers as of a year ago. The analysis posits that under traditional antitrust enforcement, these seven entities would likely be ‘several hundred competitive firms,’ many of which ‘would be S&P 500 companies in their own right.’ This perspective challenges the ‘concentration meme’ by highlighting that these dominant firms are, in essence, conglomerates formed from what could have been hundreds of standalone companies.
Global Context of Concentration
While US market concentration draws considerable attention, a global comparison reveals a different picture. Data indicates that many of the world’s largest and most advanced economies exhibit far higher levels of concentration. In countries like Canada, France, the UK, Germany, Italy, Hong Kong, Taiwan, and Korea, the top 10 companies often represent 60%, 70%, or even 80% of their respective equity markets. This places the US ‘on the relatively low end of the scale compared to the rest of the world.’ While acknowledging the US’s disproportionate share of global GDP (25%) and market capitalization (50%) relative to its population (<5%), the data suggests that if concentration is a problem in the US, it is a 'much bigger problem in the rest of the world.'
Valuation Metrics: A Sobering Contrast to 2000
Perhaps the most compelling argument against a broad market bubble comes from a direct comparison of valuation multiples between today and the dot-com peak. Heading into the dot-com implosion, companies like Intel and Microsoft sported P/E ratios of 47 and 60, respectively. Oracle reached 120, and Cisco peaked at 130. Today, the valuations are markedly different: Microsoft trades ‘under 20,’ Apple at 33, Google at 25, and Nvidia at 18. This means that at the dot-com peak, Microsoft’s P/E was approximately three times higher than it is today.
According to Ed Yardeni, the forward P/E for the technology sector today stands at 22, with the entire S&P 500 at 20.4. In stark contrast, the year 2000 saw the technology sector’s P/E at 55 and the S&P 500 at 25. The technology sector, therefore, was ‘2X as expensive as it is now’ at the turn of the millennium.
Nvidia: The AI Poster Child’s Rationalized Price
Nvidia, often cited as the ‘poster child for the claim of an artificial intelligence bubble,’ presents a particularly interesting case. Despite its rapid growth and prominence in the AI narrative, its P/E ratio today is the same as it was in 2019 – predating the pandemic, the CARES Act, the semiconductor bill, and ChatGPT’s widespread recognition. This ‘astonishing data point’ suggests that ‘its earnings have caught up to its price.’ Furthermore, the company ‘given up a trillion dollars in market cap this year,’ indicating that its price has become ‘a whole lot more rational relative to earnings.’ The analysis rhetorically asks, ‘when you see a chart like this, does it scream bubble to you?’
AI Investment: Misallocation as a Catalyst for Progress
A Deutsche Bank Research Institute chart highlights another critical aspect: the scale of private AI investment in the US. The US outspends China by ’20 times’ in this area, with China’s investment more than double that of the UK, Canada, or France. While this might appear as a ‘red flag for the people hyper-focused on a bubble,’ the perspective offered is that ‘every new technology comes with massive overinvestment and an over-allocation of capital toward that technology.’
Historically, this ‘misallocation of capital is ultimately a positive,’ even if individual investors suffer. Examples abound: the 19th-century railroad boom saw thousands of miles of track laid and most companies go bankrupt, only for survivors to consolidate and create a national network ‘for pennies on the dollar.’ Similar patterns emerged with telegraph, telephone, radio, oil, automobiles, television, aviation, semiconductors, and computers. A favorite example cited is the ‘bandwidth and fiber’ boom, where companies like Global Crossing and Metromedia Fiber laid extensive ‘dark fiber’ networks at immense cost, subsequently going bankrupt. Telecommunication and cable companies then acquired these assets ‘for pennies per mile,’ enabling the affordable, bandwidth-intensive services like YouTube and Netflix that define today’s digital landscape. This phenomenon, explored in Dan Gross’s book ‘Pop: Why Bubbles Are Great for the Economy,’ underscores that ‘somebody had to spend billions to build them and then go belly up.’ The sheer volume of ‘endless amounts of capital’ in the US, often ‘misinvested,’ is precisely why the country leads in AI, fostering an environment where overinvestment ultimately fuels technological advancement.
While the current market exhibits characteristics that warrant attention, particularly regarding capital allocation in emerging technologies like AI, the data presented does not align with the extreme bubble conditions witnessed in 1999-2000. It acknowledges the possibility of future misallocation and the eventual end of any bull market, but for now, the prevailing sentiment, supported by these metrics, remains cautiously optimistic.
