Not every big sell-off is an opportunity to double down on stocks

INVESTORS have been rewarded each time they have bought the dip in the last five years, especially in the US market.

Over that period, the S&P 500 index has risen 83 per cent, as at Feb 6. There is no disputing the fact that the market has done exceedingly well in the last few years.

But as Warren Buffett says, investors ought to “be fearful when others are greedy, and to be greedy only when others are fearful”.

Are investors greedy or fearful today?

The fact is that the market capitalisation-weighted S&P 500’s strong performance has been driven by eight companies – Apple, Microsoft, Nvidia, Amazon, Meta, Alphabet, Tesla and Broadcom.

The top 10 constituents of the index accounted for 36.2 per cent of its weightage as at Jan 31, making it less “diversified” than one might imagine.

Looking at the aforementioned eight counters, one single theme ties them all together to some degree: artificial intelligence (AI).

The promise of AI is broad and wide-ranging, and large investments have gone into developing the technology.

Boston Consulting Group said in its AI Radar 2025 report that 68 per cent of executives surveyed believe that AI will make their workforce more productive as they upskill their existing employees to meet AI needs.

Meanwhile, Facebook’s parent company Meta is investing US$65 billion to expand its AI infrastructure. In 2025, the company plans to bring online about one gigawatt of computing capacity with more than 1.3 million graphics processing units (GPUs).

ChatGPT creator OpenAI, Softbank and Oracle have also announced that they plan to invest US$500 billion in growing AI infrastructure in the United States as part of the Stargate Project.

These numbers are truly staggering, so you would think that now would be a great time to buy into Nvidia, or Meta, or simply the S&P 500 in general to capture the upside.

However, investors should remember that such expectations may already be priced in.

Lofty expectations

When news of China-based DeepSeek’s V3 model became more widely reported on Jan 24, Nvidia shed a record-breaking US$600 billion, or 17 per cent of its market cap, in a single day on Jan 27.

DeepSeek claims that it trained a fairly competent large language model in less than two months with 2,048 Nvidia H800 GPUs, which had their performance cut to meet US export restrictions to China.

Investors worried that the innovations demonstrated by DeepSeek’s engineers to achieve this would harm demand for GPUs in the long run.

This is not an unfounded concern, since the history of computing is littered with numerous instances of technology becoming vastly cheaper and more efficient.

In 1981, an IBM PC with 16 kilobytes of random-access memory and a single floppy drive cost US$5,351, after adjusting for inflation. Today, a US$35 Raspberry Pi 5 no larger than the size of a credit card has 500 times the performance.

Besides Nvidia, other tech companies have been subject to very high expectations as well.

On Feb 5, Google’s parent company Alphabet posted fourth-quarter revenue growth of 12 per cent, as compared with a growth of 13 per cent in the corresponding quarter a year ago.

Ordinarily, such performance would have been celebrated at other companies. Instead, Alphabet’s shares fell 9 per cent in after-hours trading. London Stock Exchange Group data showed that investors had expected US$96.56 billion in revenue, whereas the company posted US$96.47 billion.

Interactive Brokers chief strategist Steve Sosnick said that investors expect S&P 500 companies to continue growing their earnings by 13 to 15 per cent, which is a great reason to own stocks. However, this growth has been priced in.

“The balance of risks shifts against you because you have to exceed that 13 to 15 per cent (growth) collectively to continue to grow… On the other hand, if you just meet it or, worse, fall behind, that’s kind of disastrous,” he noted.

Sosnick recounted a time when IBM was a key winner in the computer revolution of the 60s and 70s, but was later disrupted by Microsoft and Intel semiconductor chips.

Early innovators do not always stay at the top, and even the Internet revolution has produced new winners since, such as Apple and Alphabet.

Sosnick said that even though mega-cap tech stocks have grown so much because they are hugely profitable with big competitive moats, they also risk being disrupted.

He added that if there is any hiccup or development that causes investors to change their investment thesis or, worse, run for the exits, it would significantly hurt those with high exposure to tech stocks.

In a report on Jan 7 commemorating the 25-year anniversary of his memo on the dotcom bubble, Oaktree Capital Management co-chairman Howard Marks wrote that asset bubbles are more a state of mind than a quantitative calculation.

If he starts hearing investors say “there’s no price too high”, he considers it a sure sign that a bubble is brewing.

Jumping in to buy stocks when they fall, while holding outdated assumptions, is a dangerous and undisciplined endeavour. That would be greed speaking, and that could be when investors ought to carry a bit more fear with them.