Why one of the great wealth-creation machines of all time can still be a hard buy at the wrong price.
of the index sits in its ten largest companies. The other 490 share the remaining 62%.
The index has extraordinary long-term strengths. The question is whether today's price, concentration and interest-rate backdrop leave enough room for error.
A great asset is not the same thing as a great entry price.
The case for
World-leading companies, record profit margins, deep capital markets, a large slice of revenue earned abroad and a 126-year record of recovering from crises.
The case for caution
Ten stocks make up about 38% of the index. The 10-year Treasury yield touched its highest since 2002 on 1 Oct. Housing affordability is at a 21-year low and consumer confidence at a 12-year low.
$1 invested in U.S. equities in 1900 became
by the end of 2025, before adjusting for inflation (UBS, 126 years of data).
Prices rose too. A dollar in 1900 has the buying power of about $38 today. After inflation, that same dollar became
The gold line shows what prices did. Treasury bills only just stayed ahead of it, while equities finished about 3,300 times ahead.
This is the strongest argument against a simple "bubble means crash" story: earnings have been beating forecasts by a wide margin.
Q2 2026 earnings came in about 26.5% above analyst estimates, against a 5-year average beat of 7.0%. Alphabet and Amazon drove much of it: without them the beat was 10.9% (7 Aug data). Q3 revenue growth is estimated at 12.1%, almost double its 10-year average of 6.3%.
If earnings keep compounding this fast, today's index level can become less expensive over time without needing a crash. Price is only half of the valuation sum.
An S&P 500 fund is a U.S. index, but a big share of what its companies sell is sold abroad.
268 S&P 500 companies that reported international sales, Q1 2025.
That can help when growth is stronger abroad or the dollar weakens. It can hurt when tariffs, currencies or overseas demand move against U.S. firms.
Goldman counts every S&P 500 company, and about a quarter of them report no international revenue at all. S&P Global counts only the 268 that disclose it, which tend to be the more global ones.
In Q2 2026, companies with over half their sales outside the U.S. had blended earnings growth of 74.7%, against 35.5% for domestic-focused firms. Without Alphabet, Exxon Mobil and Chevron, the international group falls to 27.5%.
A fall is not unusual. How long the recovery takes depends on what caused it.
* A 9.1% fall, so under J.P. Morgan's 10% threshold. Selected from J.P. Morgan's ten events; the chart was drawn on 22 Apr 2026.
| Type of bear market | Avg fall | Avg length | Avg recovery |
|---|---|---|---|
| Structural (financial bubbles) | −60% | 32 months | 97 months |
| Cyclical (economic cycles) | −28% | 17 months | 11 months |
| Event-driven (sudden shocks) | −28% | 6 months | 11 months |
A P/E ratio is the price you pay for each $1 of a company's earnings. Whether the S&P looks expensive depends on which ruler you hold up: forward P/E uses profits analysts expect, trailing P/E uses profits already reported.
Forward P/E has actually fallen, from 20.4 on 30 June, because earnings estimates rose faster (+8.9%) than prices (+2.7%).
Earnings yield is 1 ÷ forward P/E (1 ÷ 19.2). My calculation.
Flip the P/E upside down and you get the share of price that a year of earnings represents. Today that is about 5.2%, roughly what a 10-year Treasury pays.
Stocks can still win, because earnings can grow and a bond coupon cannot. But today you are being paid almost nothing extra to take the risk.
The top-ten share stayed between about 18% and 23% from 1990 to 2015.
In 2025 the ten largest were about 41% of index weight but were expected to deliver only about 32% of its earnings. Investors are paying up for growth that has not fully arrived.
In Q3 2026 the S&P 500 rose about 2%, but the equal-weighted version, which counts every company the same, fell about 1.5%. Tech did the lifting.
The official "technology" weight understates how much of the index rides on mega-cap tech and AI.
The share S&P reports for the Information Technology sector.
For an investor thinking about AI and digital platforms, the real exposure is wider than the tech bucket.
Gold bars hold Alphabet, Meta, Amazon and Tesla. "Other" is the remainder: staples, energy, utilities, materials and real estate.
Adopting a technology and earning a return on it are two different questions. The price you pay still matters.
One way to read it
$2tn targeted IPO valuation ÷ $4.6bn of Anthropic's 2025 revenue, per its IPO prospectus as reported on 28 Sep 2026. The company also reported a 2025 net loss of $42bn, including a roughly $34bn accounting charge.
The counterargument
Revenue grew 12-fold in 2025. Anthropic told investors its annualised revenue run rate hit $65bn at the end of July 2026, up from about $9bn at the end of 2025 (CNBC and Bloomberg, 17 Aug). Against that figure the same valuation is roughly 31 times revenue.
SpaceX listed on Nasdaq on 12 June 2026. S&P's rules require 12 months of trading and GAAP profits, and S&P declined on 4 June to fast-track mega-cap IPOs. The earliest entry is mid-2027.
The IPO is reportedly expected after the November midterms. So an S&P 500 fund holds neither company today. Nasdaq and Russell indexes did change their rules to admit SpaceX earlier.
10-year Treasury yield at the 30 Sep close (Federal Reserve H.15 data). It briefly hit 5.34% on 1 Oct, the highest since 2002 (Bloomberg).
The Fed's rate is only the short end. The long end has been selling off on its own.
Illustration: $100 received in ten years, discounted back to today. The higher the rate, the less distant profits are worth, which hits growth stocks hardest.
None of those steps is automatic or immediate. Strong earnings can outrun high rates for a while, which is what has happened so far. The risk is that the longer yields stay high, the more pressure builds through refinancing, mortgages, credit and investment decisions.
Higher long-term rates reach households through mortgage affordability, construction, collateral values and spending.
Below 100, a median-income family cannot afford a median-priced home. July's 68 is the lowest in 21 years.
Fewer new homes mean weaker residential investment and less demand for materials, appliances and furniture.
When home values fall, households generally spend less. Fed research finds a positive link between housing wealth and consumption.
Lower collateral values can make borrowing harder and weaken household balance sheets.
Consumer spending is about two-thirds of U.S. GDP, so a broad slowdown eventually reaches company revenue and margins.
The lesson: housing matters most when falling prices combine with leverage and weak credit plumbing. A house-price dip on its own is a drag. A dip that forces banks to stop lending is a crisis.
Affordability is under strain, rates are high and consumers are pessimistic. Those are real risk channels.
Barr's speech is about affordability and supply, not mass defaults, and about half of mortgages carry rates of 4% or lower. Nothing in the data above shows a credit-system break.
The six-month outlook for income, business and jobs fell for the third month in a row. Readings under 80 have historically pointed to a recession within a year.
The strongest argument against the cautious thesis, taken seriously.
2026 earnings growth is estimated near 32%, with record margins. Estimates have been revised up, not down.
If profits grow fast enough, valuation multiples can fall without prices falling. The forward P/E already slipped from 20.4 to 19.2 in three months while the index rose.
UBS finds technology stocks ultimately delivered superior multidecade returns despite the dot-com collapse.
Global revenue gives exposure to growth outside the United States.
AI productivity, margins and revenue justify much of today's optimism.
Rates stay high, growth slows, and valuation multiples compress.
Six conditions would make the risk and reward more attractive. Here is where each one stands today, so you can decide for yourself.
None of these has a right answer. All of them have a wrong one: not having asked.
A 20-year investor can tolerate a very different drawdown profile from someone who needs the money in three years. Remember that an average structural bear market took 97 months to recover.
Regular contributions spread your entry prices over time. A lump sum is more sensitive to the starting valuation.
Your job, share options, existing shares and pension can all add exposure that is invisible if you only look at one fund. The S&P 500 is priced in dollars, so if you earn and spend in euro, exchange-rate moves add to or subtract from your return.
The average cyclical or event-driven bear market falls about 28%. If a fall of 20–30% would make you panic, the problem may be portfolio construction rather than the index.
Political and geopolitical shocks can move markets violently. Long-term allocation decisions should rest on your plan, not on one day's news.
Why I respect it
Long-term wealth creation, strong and rising earnings, global revenue, powerful innovation and repeated recovery from crises.
Why I'm cautious today
About 38% of the index in ten companies, AI expectations, a 10-year Treasury yield above 5%, the weakest housing affordability in 21 years and the lowest consumer confidence in 12.
Money Made Simple, plain-English money for Ireland.
Figures re-verified against the sources below between 1 and 4 Oct 2026. Links are on the next page. Where providers disagree, the range is shown.
FactSet Earnings Insight25 Sep 2026: Q3 earnings growth 29.1%, CY2026 32.0%, Q2 net margin 17.0%, forward P/E 19.2, trailing 25.8, price target 9,275, 59.9% Buy ratings. 28 Aug: Q2 beat 26.5%. 7 Aug: beat excluding Alphabet and Amazon 10.9%, 5-year average 7.0%. 4 Sep: Q3 estimates +1.2% in Jul–Aug, typical decline 1.7%. 4 Aug: growth by international exposure. 27 Feb: CY2026 estimate 14.6%.
UBS Global Investment Returns Yearbook 2026$1 in U.S. equities in 1900 grew to $124,854 nominal and $3,296 real by end-2025; 9.8% and 6.6% a year; the U.S. is 62% of world equity value; technology delivered superior multidecade returns after the dot-com collapse. Released 3 Mar 2026.
J.P. MorganBear markets since 1929 (Apr 2025): average −39%, 18 months, 39 months to recover. Recovery days from Bloomberg data, 24 Apr 2026; counts are trading days.
S&P Global Market Intelligence and Goldman SachsFor 268 companies disclosing international sales in Q1 2025, 35.9% of revenue came from outside the U.S. Goldman: foreign sales were 28% of S&P 500 revenue in 2024, and about a quarter of firms report none.
RBC Wealth ManagementTop-ten share about 19% in 1990 and 23% in 2000; 18–23% from 1990 to 2015; a record 40.7% in 2025; about 41% of weight against about 32% of earnings.
Index weights (approximate)Sector weights, end Aug 2026: Information Technology about 37%, Financials 12%, Communication Services 10%, Consumer Discretionary 9%, Health Care 9%, Industrials 9%. Top-ten share about 38% (37–39% across providers). Magnificent Seven 33.9% (Stock Analysis via Motley Fool, Sep 2026). These move daily and differ by provider.
Federal ReserveFOMC decision, 16 Sep 2026 (3.75–4.00%, unanimous, first hike since 2023). 10-year yield 5.29% on 30 Sep (H.15). Governor Barr, 23 Sep 2026 (affordability index 68, price and income figures, mortgages). Chairman Bernanke, 10 Feb 2011. Monetary Policy Report, 24 Feb 2009.
The Conference Board29 Sep 2026: index 81.9 (Aug revised to 88.6), Expectations 63.6, Present Situation 109.3 (Aug 117.2), 12-month inflation expectations 6.1%. Petrol $4.45 against $3.13 a year ago (AAA, via CNBC).
Bond marketsBloomberg and Reuters, 1 Oct 2026: 10-year yield briefly 5.34%, highest since 2002. Treasury par curve for 30 Sep is rounded to one decimal because aggregators differ by a basis point or two on some maturities.
Anthropic and SpaceXReuters and CNBC on the Anthropic prospectus, 28 Sep 2026. CNBC and Bloomberg, 17 Aug 2026, on the $65bn run rate. CNBC, 12 Jun 2026, and S&P DJI's 4 Jun 2026 decision (as reported by SpotGamma) on SpaceX.
Market levelsS&P 500 close of 7,651.54 on 30 Sep, up 2% in Q3 and almost 12% in 2026; equal-weight index down 1.5% in Q3 (CNN, CNBC). Brent back above $100; seventh month of the Iran war (Bloomberg via Yahoo Finance).
Buffett“Buffett on the Stock Market,” Fortune, 22 Nov 1999.
My calculationsEarnings yield (1 ÷ forward P/E), the discount-rate illustration, and the ~435× and ~31× revenue multiples are mine, not published figures.
Open any link to check a figure yourself. Page numbers show where each source is used.