In his tome Extraordinary Popular Delusions and the Madness of Crowds, Charles Mackay defined mania as the phenomenon in which “millions become simultaneously impressed with one delusion and run after it”.
It is as old as greed and its anatomy is unchanged. In the early 1970s, the Nifty Fifty on the New York Stock Exchange represented hot stocks like IBM, Xerox, Coca-Cola, GE, Kodak, JC Penney and ITT. The belief was that these shares were worth any price. Howard Marks of Oaktree recalls the prevailing mantra: “You can’t be fired for buying IBM.” It wasn’t long before inflation, the Dracula of valuations, arrived. The S&P500 lost over 45 percent. The Nifty Fifty crashed over 70 percent. Everything “great” was toppled from the pedestal of popularity.
Today, money is chasing silicon chips for the proverbial gold under the rainbow. Money lured by returns is piling into AI stories. Taiwan’s TAIEX is up 50 percent, while Korea’s Kospi rose 96 percent. The returns come with concentration risk. TSMC alone represents 42 percent of Taiwan’s total market capitalisation. Samsung Electronics and SK Hynix account for half of South Korea’s market. The three companies account for over 25 percent of the MSCI Emerging Markets index. Tech accounts for over 45 percent of Japan’s Nikkei 225. ASML, ASM International, BE Semiconductor are hosted on Europe’s Stoxx 600 as well as global exchange traded funds.
In the US and Europe, the dot-com bloodbath haunts conversations. The markets are also troubled by the corralling of a large slice of savings into a handful of stocks. The Magnificent Seven tech stocks account for 35 percent of S&P 500—the modern passive index fund has become the ultimate financial anaesthesia.
The world has been here before. Each episode follows the same arc—a new technology emerges, capital floods a set of companies, these dominate indices until valuations uncouple from reality. The technology survives and investors need trauma therapy. Each episode is book-ended with the slogan ‘This time it’s different’. In the late 1990s, tech and telecom reached 41 percent of the S&P500. Cisco, which peaked at 200 times earnings, fell 86 percent. Qualcomm fell 88 percent. Global Crossing went bankrupt.
This time, too, the headline is bullish—for one, Scandisk is up 600 percent—but the headcount tells a different story. On Friday, when the S&P500 hit a new high, 204 stocks posted negative year-to-date returns , a count that usually moves money. Last year, the Mag7 grew 23 percent. Barely half of the rest grew at 11 percent. Yes, the market is visibly K-shaped. When Korea’s Kospi hit a new high this week, only 75 stocks were up, while 826 were down! This has implications for the market makers and users.
Concentration risk is essentially too many eggs in one basket. The architecture of financial markets has morphed into a high-stakes wager. The systemic risk is of not a hypothetical nightmare. It vested in millions of portfolios. In the US, the Mag7 account for over 30 percent of 401K retirement accounts. The mechanism is automatic. By design, 401Ks make the biggest contributions to the largest-cap companies. On average, every American who puts money in 401K invests $2,358 in Mag-7 stocks. This boosts valuations and also amplifies concentration risk as any dip in these major stocks impacts the portfolio.
The concentration risk is not limited to the US and there is no historical precedent to the global simultaneity. The 1970s’ Nifty Fifty crash largely hit the US; the 1980s’ Japanese bubble was limited there. Asian savings were mostly insulated from the 2000 Nasdaq crash. But today, AI stocks are embedded in American, Korean, Taiwanese, Japanese and European indices. There is no major global index one can buy with without the semiconductor effect.
There are many what-ifs here. What if an AI breakthrough slashes compute demand, a large data centre company trims capital expenditure guidance, or a cost blow-up forces CFOs to mull over AI tokens vs humans? The systemic risk is globalised and exposed to geographic choke points at the same time. The trillion-dollar Taiwanese giant TSMC has 72 percent of the global foundry revenue; compare it with the next— Samsung at around 7 percent. There is no geographic hedge for TSMC yet—over 80 percent of its manufacturing capacity is in Taiwan. A single geopolitical shock would cascade through every index on Earth.
The structural risk deepens when you examine the profound mismatch between corporate infrastructure expenditure and real downstream income. The need for capital has moved beyond the circular you-fund-me-I fund-you model. The estimates of borrowings range from $600 billion last year to $700 billion this year, and the earnings of the giants are still in two-digit billions.
The upcoming trillion-dollar IPOs of SpaceX and Open AI will soak liquidity, push up the cost of capital and increase concentration risk. Will they correct prices or crack the confidence? Bank of America’s Michael Hartnett says, “Strong price action, retail mania, slumping volume… so bubbly.” India has reasons to worry as the global equity and currency markets are intimately wired together.
Meanwhile, the Gulf war has pushed up inflation across the global economy. The world is bracing for a higher-for-longer interest rate regimes. The math is simple. By one estimate, at 3 percent interest rates, the Mag-7 trades are at 32 times earnings. At 6 percent, the price-earnings ratio could well compress to 20x. Can Big Tech earn enough to pay up?
The question is not whether the AI boom is real. It is whether the world has priced it correctly and if it has a Plan B. During a stampede, it is instructive to remember Mackay’s words, “We go mad in herds, while we only recover our senses slowly, one by one.”
Read all columns by Shankkar Aiyar
Shankkar Aiyar
Author of The Gated Republic and Accidental India.