Market realities, bubbles: Is Peak America leading to lower returns ahead?

There is considerable debate among investors as to whether the US equity markets are in a bubble?  Are we experiencing another 1999/2000 internet bubble all over again, this time with artificial intelligence (AI) being the focus? This issue is relevant for all markets because if the US is in la-la land, a steep market fall is inevitable, with obvious consequences for asset allocation and all financial markets. For context, after the bubble burst in March 2000, the Nasdaq fell by 78 per cent and the S&P 500 by 49 per cent over a 30-month period. It took 15 years for the Nasdaq to cross its year 2000 bubble peak of 5,000, such was the extent of damage.

 

The case for the bubble is straightforward. On the classic Shiller cyclically adjusted price-to-earnings (CAPE) ratio, which takes rolling 10-year real earnings as its denominator, current US valuations are excessive at about 40 times, having only been higher in the 2000 bubble at 45 times. We are today well above the 1929 peak of 32 times. This is clear bubble territory.

 

Market concentration has never been higher, whether we look at the top five or 10 stocks, or even the top decile of companies. Today, the top 10 stocks in the S&P 500 account for almost 40 per cent of the index, much higher than 25 per cent in 2000. The top decile of companies makes up 75 per cent of the market, an all-time high.

 

The US now has nine companies (eight of them technology firms) with market capitalisations exceeding $1 trillion, and three valued at over $3 trillion. There are also three technology companies earning about $100 billion in annual profits each. Yet, despite this scale of absolute profits, analysts expect earnings for these giants to grow at 15 per cent per annum over the coming five years.

 

The market has also been highly bifurcated, with the average stock rising only 13 per cent in 2024, despite the S&P rising by 25 per cent. This 12 per cent performance gap between the market-cap-weighted and equal-weighted S&P was also seen in 2023, but it is unusual. Over longer periods, there is typically little difference between the two index constructs. The last time we witnessed such a large divergence was in 1998–99.

 

We also have the surge in capex on AI. Just four companies (Alphabet, Amazon, Microsoft and Meta) have plans to spend $320 billion on technology capital in 2025, the majority of which will go for chips and data centre infrastructure. Including smaller companies like Oracle and others, this figure will exceed $400 billion. For context, this is 10 per cent of India’s gross domestic product (GDP) being spent by a few large US technology companies in just one year, solely on AI. What is the return on this capital expenditure? No one knows, beyond the assertion that the risks of underspending are far greater than those of overspending.

 

These companies are now spending all of their annual free cash flow and about 20 per cent of sales. Thankfully, no debt. Does this not smell like the over-investment in fiber optics and telecom that sowed the seeds for the tech bust of 2000? No one knows and only time will tell, but we can see the bear argument.

 

Similarly, profit margins for corporate America are near all-time highs, and we have had more than a decade of earnings growth exceeding nominal GDP by almost 4 per cent per annum. If you go back to look at longer data series, over the last 70 years, corporate profits always lagged nominal GDP. The median nonfinancial company in the S&P has a net profit margin of 12.5 per cent, this was 5 per cent in 1995. We seem to be in a period of corporate overearning.

 

The US economy is running a fiscal deficit of 7 per cent at full employment, with a debt/GDP ratio of 100 per cent, $28 trillion of federal government debt, with an absurdly short average maturity. Amazingly, back in 2000, the expectation was that the US government would go debt free by 2013, and have zero debt even in 2025! How wrong were the analysts!

 

This, in a nutshell, is the bear argument: All-time high valuations on above-trend earnings, a massive capex boom with unclear returns, and an economy running above trend — supercharged by massive fiscal indiscipline and AI investment.

 

American exceptionalism is taken for granted, and with a 70 per cent weight in global indices, we may be at peak US. This may or may not be a bubble, but it is certainly a recipe for poor prospective returns.

 

The bulls argue that inflated valuations are primarily driven by mega-cap stocks. While the top 10 companies in the US trade at 27 times forward earnings, the remaining 490 companies trade at only 20 times. They also contend that investment in AI will lead to a sustained uptick in labour productivity, signs of which are already visible. A productivity surge could drive sustainably higher GDP and earnings growth.

 

The US has energy security and the lowest costs, is the leader in technology and AI, has the best long-term equity returns in the world, and is about to enter an era of much lower taxes. It still has the best demographics among the large Western powers. Bulls argue that the Shiller CAPE ratio is backwards looking and does not take into account the surge in productivity and earnings that AI will bring. Earnings will surpass trend. With more than 50 per cent of US equity mutual fund and ETF assets now passive, we are not seeing the rampant speculation and day trading witnessed in 1999-2000. This market continues to climb a wall of worry and Nvidia of today is not Cisco of 2000. There remains no alternative to the US given the structural weaknesses of the EU, Japan and China. The US deserves to trade at a premium and will continue to do so.

 

The bulls also point to price action. The Nasdaq had tripled between 1995 and 1998, and then went up by 86 per cent in 1999 (remains its best annual return). In the final five months before the Nasdaq peaked in March 2000, it had risen by 88 per cent. While tech has done well, we are nowhere near this type of steep price acceleration today. They also point out that when the crack happened, we saw the Nasdaq declining by more than a third within the first month. It lost 10 per cent in the first week. Again, despite the shock of the Deepseek model, we have not seen this type of drawdown. The Nasdaq ultimately succumbed to rising rates and a recession hitting the US in 2001. We seem to have gone through this phase unscathed. Investors expect the Fed to cut rates from here onwards, not raise them, and there seems to be little risk of a recession on the horizon.

 

I remain unconvinced by the bull arguments. I believe we are at “Peak America.” Ideally, we would see a gradual derating of US assets, allowing risk appetite to remain healthy and capital to flow overseas. A sharp de-risking episode could trigger volatility and drag down all markets. I hope the US is not a bubble, as a bubble burst could lead to a 40-50 per cent decline. When a bubble pops, all bets are off — global markets and investors would reduce risk exposure, benefiting no one.

 

It is far better for the US to gradually underperform global markets, as we have seen from the beginning of 2025.

 

The author is with Amansa Capital