The thesis
“Nvidia and the AI chip stocks are in a bubble that is about to pop.”
How buffet read it
$10,000 split across NVIDIA, the VanEck Semiconductor ETF, and the iShares Semiconductor ETF.
Ten years back to today.
Dividends reinvested, rebalanced annually.
Benchmark: S&P 500.
NVDA 40%, SMH 30%, SOXX 30% · 2,513 trading days · benchmark SPY
$10,000 became
$0
The same money in the S&P 500 would have been $41,459 over 2016–2026 — the thesis came out ahead by $480,696.
The worst stretch
−54.9%, November 2021 to October 2022 — you would have sat through eleven months of that, and it took until May 2023 to get back to level.
the S&P 500 fell 33.7% at its worst over the same period. The thesis moved 38.9% a year against 18.0% for the index.
Where the thesis was wrong
The thesis was that AI chip stocks were about to collapse; over 2016–2026, the portfolio returned +5121.5% against the S&P 500's +314.6%. That result depended heavily on NVDA's +14272.5%, which alone pulled the blended return far above what the semiconductor ETFs contributed. In 2022 the portfolio fell −41.6% against the index's −18.2%, and the worst drawdown hit −54.9% between November 2021 and October 2022, nearly twenty points deeper than the index's worst stretch. Annual volatility ran at 38.9% against the index's 18.0%, so the ride that produced the gain was roughly twice as rough throughout.
Counter-test
Same money, same schedule, parked in 1–3 month T-bills instead: $12,510.
Year by year
| Year | The thesis | SPY |
|---|---|---|
| 2016 | +36.1% | +3.7% |
| 2017 | +60.4% | +21.7% |
| 2018 | −15.6% | −4.6% |
| 2019 | +70.2% | +31.2% |
| 2020 | +86.1% | +18.4% |
| 2021 | +73.0% | +28.7% |
| 2022 | −41.6% | −18.2% |
| 2023 | +135.3% | +26.2% |
| 2024 | +78.5% | +24.9% |
| 2025 | +44.2% | +17.7% |
| 2026 | +46.1% | +13.4% |
What this verdict does not include
- US-listed stocks and ETFs only, priced on end-of-day closes.
- No tax, no trading fees, no slippage, no bid-ask spread.
- Dividends are reinvested, because closes are adjusted for them.
- Survivorship is not corrected for: only instruments that exist today are held.
- Only 10 years of history exist for this. Read it accordingly.
- The forward claim — that AI chip stocks are in a bubble about to pop — cannot be tested with historical data; this backtest shows only what has already happened.
- No single 'AI chip bubble' instrument exists; NVDA is included as the named stock, with SMH and SOXX as the broadest semiconductor proxies in the catalogue.
- The 'about to pop' timing rule cannot be expressed; this is a buy-and-hold backtest only.
Keep this thesis
buffet re-runs it once a month and mails you the new number. Nothing else is ever sent.
Another thesis
Every verdict names something it could not test. That gap is usually the next thesis.
Run another thesis — $5