Is Big Really Beautiful? Ten stocks, 40% of the index, and the risk hidden in your pension

SCM Direct · Alan Miller’s Blog

Super companies, super funds, and the crowd psychology that normally ends in tears.

There is an old rule in investment that nobody writes down, but everybody eventually learns: the moment something becomes too big to question is normally the moment you should start questioning it.

Right now, big has never been bigger.

Ten stocks now make up 40% of the S&P 500

How record concentration in Nvidia, the S&P 500 and SpaceX has raised the risk hidden inside ordinary index funds and pensions.

Nvidia became the first company in history to be worth $5 trillion last October¹. It has since given roughly $1 trillion of that back.

Nothing went wrong with the business. The crowd simply stampeded sideways into memory chip makers instead: Micron, Intel and AMD added about $2 trillion of combined value in three months². The crowd has not sobered up. It has moved to a different bar.

How concentrated is the S&P 500 in 2026?

The ten largest US stocks were a record 40.7% of the index at 31 December 2025, on RBC Wealth Management and FactSet data³. Published 2026 measures range between roughly 36% and 43%, depending on the date and the methodology used⁴.

Between 1990 and 2015 that figure sat between 18% and 23%. Concentration is therefore around double its long-run norm.

The average price/earnings ratio of those top ten names is around 50⁵.

In plain terms: if you own a US tracker, more than 40p of every pound is riding on ten companies making broadly the same bet on the same technology.

A single exhibit for the prosecution

Deutsche Bank’s annual “Charts to make you go WOW” pack, published this month by Jim Reid’s team, supplies it⁶.

Kioxia, a Japanese memory chip maker, was worth about $8 billion a year ago. At its recent peak it had risen roughly 46-fold, making it the largest company in Japan. It joined the Nikkei three months ago.

Forty-six times, in a year, in a business whose product is a commodity that has bankrupted producers in every previous cycle.

The same trade has seen Samsung and SK Hynix help the KOSPI triple, pushing South Korea’s stock market past Europe’s largest exchanges by value. When Reid’s team says today’s market “carries echoes of 1999”, they are being polite.

The largest flotation in history, and the fastest reversal

In June, SpaceX raised $85.7 billion at $135 a share⁷. The stock jumped 19% on day one and gained more than 40% in its first two sessions.

By 16 June the company was worth around $2.7 trillion — nearly eight times the $350 billion its own board had set in a private tender just eighteen months earlier⁸.

At the peak, that was roughly 145 times last year’s revenue of $18.7 billion, at a company that lost $4.9 billion in 2025⁹.

Elon Musk has suggested the company “might be able to reach approximately” $1 trillion of revenue by 2030. That would require revenues to grow more than fifty-fold in four years. Even Robert Greifeld, the former boss of Nasdaq itself, admitted the stock was “trading not on fundamentals” but on aspiration¹⁰.

Then the crowd changed its mind. In the six weeks since that peak, more than $1.2 trillion has been erased — what Bloomberg describes as one of the largest market capitalisation wipeouts in history.

The shares closed on 28 July at $116.49, some 14% below the IPO price, valuing the company at around $1.5 trillion. That is still roughly 80 times revenue, and all before it has published a single set of results as a public company; the first is due on 4 August⁸.

Every investor who bought at the float, or in the euphoric first fortnight, is already underwater. The tears in this blog’s title are not a prediction. For this crowd, they have already begun.

I have nothing against aspiration. The rockets are genuinely magnificent. I have quite a lot against paying 145 times revenue for it.

How the AI boom is being funded: $570bn of debt and disputed depreciation

The AI-related borrowing, data-centre securitisation and depreciation assumptions driving the boom — the parts of the story being ignored.

Bull markets are sustained by stories. This one is increasingly sustained by debt and by accounting assumptions, and the details deserve far more attention than they get.

The debt

Morgan Stanley estimates roughly $570 billion of AI-related debt issuance in 2026, with $236 billion already priced by the end of May — four times last year’s pace¹¹.

Barclays counts data centre asset-backed and mortgage-backed securities — loans bundled up and sold on to investors — rising from $4 billion outstanding in 2020 to $61 billion by mid-2026. It now describes Oracle’s credit default swap as a liquid hedge on AI capital spending generally¹².

Think about that for a moment. The market has invented an instrument for betting against the boom, and it is one of America’s largest technology companies.

The depreciation

Depreciation is the dullest word in finance and currently one of the most important. The value of hundreds of billions of dollars of AI kit depends on how long it stays useful — and the companies themselves cannot agree.

During 2025, Meta stretched the assumed life of most of its servers to five and a half years, reducing reported depreciation by around $2.3 billion over nine months, according to its own quarterly filings¹³.

Amazon went the other way. It cut a subset of server lives from six years back to five from January 2025 and booked a $920 million accelerated depreciation charge, explicitly citing, in its FY2024 Form 10-K, the “increased pace of technology development, particularly in the area of artificial intelligence”¹⁴.

Identical machines. Opposite conclusions.

Meanwhile Meta’s $27 billion Hyperion data centre financing carries bonds due in 2049, secured against a campus whose lease initially runs for four years — with the guarantee sitting in the footnotes rather than on the balance sheet¹⁵.

The underlying point is blunt. The hyperscalers — the handful of giants that own the world’s cloud computing capacity — are now spending more on capital expenditure than they generate in operating cash flow¹⁶. The most cash-generative businesses in history have collectively spent their way past their own cash flow, and the difference is being borrowed.

None of this means the technology fails. It means enormous, confident, levered numbers are resting on assumptions that the people writing them cannot agree on. Crowds do not read footnotes. That is rather the point of crowds.

Why global and emerging-market trackers hold the same bet

Global and emerging-market index funds carry the same AI-chip concentration risk — and what genuine diversification looks like instead.

The standard advice at this point is to diversify away from expensive US mega caps into international and emerging markets. Here is the uncomfortable bit.

Is an emerging markets tracker actually diversified?

On Deutsche Bank’s like-for-like methodology at 8 July 2026, the top ten holdings of the MSCI Emerging Markets index are 39.4% of it, against 36.7% for the S&P 500. On that consistent measure, the emerging markets index is the more concentrated of the two¹⁷.

MSCI’s own factsheet shows why. TSMC alone is 14.5% of the standard EM index; with Samsung and SK Hynix, three chip makers account for nearly 29% of what is sold to investors as broad emerging markets exposure¹⁸.

Buying a conventional EM tracker today is, somewhat ironically, a large bet on exactly the same AI hardware cycle.

How much of a “global” fund is America?

The MSCI World index, the benchmark behind most global tracker funds, is now 72.5% United States on MSCI’s own figures. One of the most popular global trackers in Britain was 73.5% America on Bloomberg’s fund data this week, with semiconductors its single largest industry group¹⁹.

Even on the widest possible measure — every listed company on earth, including all emerging markets — the United States is roughly half of world equity value, on about 25% of world GDP and 4.5% of world population²⁰.

The crowd has not spread its bets. It has crowded into every market in the world and bought the same handful of shares.

What we hold instead

This is exactly why, at SCM, our substantial emerging markets exposure is held through EM small cap rather than the conventional MSCI EM index. The difference is not cosmetic.

In the standard EM index, the largest single stock is 14.5% of your money. In the EM small cap index, the largest of roughly 1,800 holdings is about 1.2%, and the entire top ten is around 7%²¹.

One is a semiconductor bet with an EM label on it. The other is nearly two thousand businesses spread across two dozen economies.

Given how extreme concentration in the conventional index has become, we are researching ways to diversify our emerging markets exposure further still, away from the MSCI EM index and its embedded chip-cycle risk. Genuine diversification takes actual work, and that is what investors should be paying an investment manager for.

The madness of crowds, updated for 2026

What tulips, the Nifty Fifty, 1980s Japan and the dot-com bubble teach us about today’s market.

Charles Mackay published Extraordinary Popular Delusions and the Madness of Crowds in 1841, and nothing in today’s market would surprise him.

Crowds do not gradually become wrong. They become wrong all at once, together, while congratulating each other on their brilliance.

Tulips in 1637. The Nifty Fifty in 1972. Japan in 1989, when the grounds of the Imperial Palace were notionally worth more than California. The dot-com bubble in 2000, when Cisco briefly became the most valuable company on Earth.

Cisco was, and remains, a superb business. Its share price took over two decades to regain its March 2000 peak.

That is the point people always miss. Bubbles are rarely built on rubbish. They are built on genuinely wonderful companies at prices that assume the future arrives on schedule, in full, with no competition and no accidents.

Nvidia is a magnificent business. So is SpaceX. So was Cisco. The question is never “is this a great company?” It is “what am I paying, and what has to go right?”

What GARS, Woodford and Fundsmith teach us about giant funds

The same crowd psychology that inflates giant stocks inflates giant funds — and the ending is depressingly familiar.

I know, because I have watched this before, and said so at the time.

Standard Life GARS

Older readers will remember it. At its peak in May 2016 it held £26.8 billion and was the largest fund in Europe²². Advisers piled clients in. Consultants blessed it. The marketing promised positive returns in all market conditions — a claim that should always make your hand move instinctively towards your wallet.

In 2016, at the very peak of its popularity, I wrote a blog asking whether even Einstein could understand GARS, confessing that I could not comprehend its “Swedish flattener v Canadian steepener” strategies, and quoting Warren Buffett’s rule: never invest in a business you can’t understand. In 2018 I went further, and called it potentially one of the greatest mis-selling scandals in the UK²³.

I was told I didn’t understand its sophistication. Quite right. Neither, it turned out, did anyone else.

The fund missed its own cash-plus-5% target three years running, suffered record outflows, and was quietly merged out of existence in 2023 as the second-worst performer in its entire sector over three and five years²⁴. The crowd that rushed in rushed out, and the ordinary savers who arrived last, as always, paid the bill.

Fundsmith

Let me say clearly that Terry Smith’s long-term record is real: since 2010 the fund has still, just, beaten the MSCI World. But the recent history is sobering.

Fundsmith Equity has underperformed the index every year since 2021. In the first half of 2026 it lost 2.9% while the MSCI World gained 11.2%. It now lags the index over ten years. Assets have fallen from a peak of around £29 billion to roughly £12 billion²⁵.

Here is the part that fascinates me as a student of investor psychology. For fifteen years the Fundsmith mantra was “Buy good companies. Don’t overpay. Do nothing.” Investors were told, repeatedly and with great certainty, that trading was for fools.

Then, in the first six months of this year, the fund turned over 51.8% of its portfolio — a record — selling a list of holdings that had underperformed the index by an average of 46% over the previous year, and openly conceding that momentum now matters²⁶. The “do nothing” fund did more in six months than it used to do in a decade.

The letter blames passive funds and momentum for the underperformance, which is a curious argument given that a simple quality-factor ETF, charging a quarter of Fundsmith’s fee, has comfortably beaten the fund over one, three and five years²⁷. If cheap, rules-based money can beat you at your own stated style, the problem is not the referee.

To his credit, Smith has also warned that this passive-led momentum market will end badly, noting that momentum’s relative performance is at a 30-year high, beyond even the dot-com extreme²⁸. On that specific point, he and I agree completely. Which makes the decision to start chasing momentum now all the more remarkable.

The pattern never changes

I take no pleasure in any of this. Genuinely. But there is a lesson every investor should know: when a fund manager believes they walk on water, they normally drown.

Not because they are stupid — they rarely are — but because ego is the most expensive item on any fund’s fee schedule, it never appears in the KIID (the short factsheet a fund must give you), and it compounds faster than returns.

We saw it with GARS. We saw it with Neil Woodford, whose devoted crowd found the exit doors locked. Brilliance, adulation, size, certainty, underperformance, excuses, capitulation. In roughly that order.

Corporate ego follows the same script. Companies promising $1 trillion of revenue by 2030, or building a base on the moon with shareholders’ money, may deliver. History suggests the confidence peaks before the share price does.

How SCM builds a genuinely diversified portfolio

Not panic, and not pile in. Both are crowd behaviours.

At SCM the discipline is the same one we have applied since Gina and I founded the firm in 2009, and it is deliberately boring.

None of what we do is glamorous. Neither was warning about GARS in 2016, or about property funds days before they suspended, or writing in January 2025 that the market’s love affair with a handful of technology stocks was stretching the elastic band close to snapping — shortly before trillions were wiped off the Nasdaq²⁹.

We are steady Eddies, and we tend to be early, which is uncomfortable. But it is the only place value has ever been found.

Big can be beautiful. Nvidia and SpaceX may change the world, and some giant funds will recover. But beauty at any price is not investing. It is infatuation — and infatuation, whether with a stock, a rocket or a fund manager, normally ends in tears.

Alan Miller is Chief Investment Officer and Founding Partner of SCM Direct.

Alan co-founded SCM Direct with Gina Miller in 2009 and has publicly flagged risks ranging from Standard Life GARS in 2016 to the concentration of technology stocks in July 2024. In 1997 he launched the UK’s first equity long/short hedge fund, achieving a 17.2% annualised return over 9.5 years. He managed the Jupiter Investment Trust, later New Star Investment Trust, from 2000 to 2006, delivering a 47.5% NAV increase against a 0.2% decline in the FTSE All-Share.