Part I showed the pattern: Steam, Electricity, Internet, AI: four General Purpose Technology waves, each bigger and faster than the last. Each time, the infrastructure layer captured the value. This is where it gets numerical.
I couldn't find this comparison anywhere in financial media. So I made it.
Not opinions. Not predictions. Nineteen years of what actually happened. Four asset classes most people choose between, plus one honest benchmark: real estate, MSCI World, Nasdaq, Semis, and Berkshire Hathaway. And then, based on that, what might happen next.
The Setup
Four asset classes. Nineteen years. Same starting capital. Real Estate: the classic store-of-value choice. MSCI World / S&P 500: the "responsible" global choice most financial advisors recommend. Nasdaq 100: tech-focused, concentrated in fewer names. And Semis (SOX): the semiconductor sector index. The one nobody talks about.
Plus one benchmark: Berkshire Hathaway (BRK). Not a market index. A reference point for what disciplined, defensive active investing looks like over the same period.
Period: 2006–2024. Two major crashes (2008, 2022). One pandemic. One rate cycle. Real conditions. Not cherry-picked years.
Year-by-Year: 2006–2024
Each cell is the annual return. Green = gain. Red = loss. Darker = bigger move.
| Real Estate | MSCI / S&P 500 | Berkshire (BRK) | Nasdaq 100 | Semis (SOX) | |
|---|---|---|---|---|---|
| 2006 | +0% | +7% | +18% | +8% | 0% |
| 2007 | +0% | +4% | +11% | +18% | −2% |
| 2008 | −3% | −37% | −32% | −42% | −52% |
| 2009 | +0% | +27% | +3% | +55% | +75% |
| 2010 | +2% | +16% | +13% | +20% | +14% |
| 2011 | +4% | +2% | −5% | +3% | −11% |
| 2012 | +5% | +17% | +17% | +18% | +7% |
| 2013 | +7% | +30% | +33% | +37% | +42% |
| 2014 | +7% | +20% | +27% | +18% | +30% |
| 2015 | +7% | +4% | −12% | +10% | −2% |
| 2016 | +8% | +12% | +23% | +7% | +38% |
| 2017 | +9% | +22% | +22% | +33% | +40% |
| 2018 | +9% | −4% | +3% | 0% | −6% |
| 2019 | +6% | +32% | +11% | +38% | +62% |
| 2020 | +4% | +7% | +2% | +48% | +53% |
| 2021 | +14% | +38% | +30% | +28% | +44% |
| 2022 | −4% | −13% | +4% | −29% | −35% |
| 2023 | −7% | +22% | +16% | +49% | +67% |
| 2024 | +2% | +33% | +25% | +34% | +13% |
Methodology note: MSCI World / S&P 500 and Nasdaq 100 are EUR-denominated ETF annual returns (JustETF). SOX is the USD Philadelphia Semiconductor Index (totalrealreturns.com) — the USD/EUR rate moved from ~1.27 (2006) to ~1.08 (2024), a modest tailwind for European investors that slightly reduces SOX returns in EUR terms. Real Estate = Destatis/GlobalPropertyGuide nominal price index, prices only. All data approximated. Past performance does not predict future results.
The Compounded Result
Same €10,000, invested January 2006. Left alone for 19 years. Not touched.
| Asset | Est. CAGR | €10k becomes | Multiple |
|---|---|---|---|
| Real Estate (incl. est. rent yield) | ~6.5% | ~€32,000 | 3.2× |
| MSCI World / S&P 500 | ~10–11% | ~€67–74,000 | 6.7–7.4× |
| Berkshire Hathaway (BRK) | ~11.4% | ~€80,000 | ~8× |
| Nasdaq 100 | ~16% | ~€185,000 | 18.5× |
| Semis (SOX) | ~18% | ~€233,000 | 23.3× |
Real estate (incl. rental yield): ~€32,000. Semiconductor index: ~€233,000. Same capital. Same nineteen years.
The total return estimate factors in ~2.5% rental yield on top of price appreciation. Even then, real estate is still 7× behind. The "safe" choice and the "high-return" choice weren't even in the same league.
This is in the most documented, regulated, scrutinised market on earth. Nobody manipulated it. Nobody predicted it either. It just happened.
And almost no one is talking about it.
One more benchmark worth naming. Warren Buffett’s Berkshire Hathaway compounded at roughly 11.4% annually over the same period. Better than the S&P 500, and better than nearly any active fund run over the same stretch. €10,000 became approximately €80,000.
Berkshire is not a bet on growth. It is a deliberately defensive portfolio: large cash reserves, insurance companies, railways. The 2022 column shows it clearly: BRK held flat while Nasdaq fell 29% and Semis fell 35%. The most successful long-term investor in history, running a careful defensive portfolio, still underperformed the infrastructure layer over nineteen years. Which makes the gap harder to explain away, not easier.
Still less than half of what a passive semiconductor index delivered. No stock selection. No research calls. No genius. Just owning the right layer.
Why Did This Happen?
Because the internet wave (the last General Purpose Technology shift) needed chips. Every server, every smartphone, every data center, every router. The infrastructure layer of the last regime change was made of silicon.
The Nasdaq holds the tech companies that built on top of this infrastructure. The semiconductor index holds the companies that made the infrastructure possible. One layer deeper. One step closer to the bottleneck.
And in exponential technology curves, the bottleneck captures the value. Not the products built on top of it. The layer that everything else depends on.
Why not just hold the Nasdaq? Because the Nasdaq 100 is not a clean signal. It contains 100 companies, but in the last decade roughly seven of them (the Magnificent Seven) generated the majority of its returns. The rest dragged the average. You were paying for diversification into legacy. The Power Law was doing all the work, and you owned the dilution around it.
Semis are different. They're not a bet on which AI company wins. They're a bet on the layer that all of them depend on. When OpenAI beats Google, both need chips. When the next model surpasses both, it needs chips too. You don't pick the winner. You own the material every winner is built from.
That's the bottleneck. One layer beneath the Nasdaq. And that's why the numbers look different.
This is not an original idea. It's a pattern. Verified in every major technological wave since the steam engine.
Now: The Next Wave
We are at the beginning of the next General Purpose Technology shift. AI is not a product. It is infrastructure: the new electricity layer of the global economy.
And it requires more chips than anything in history.
The question isn't whether to pay attention. The question is: what does the next twenty years look like if this pattern repeats?
Three scenarios. Each anchored to an observable historical reference point.
2026–2045: Three Scenarios
This is not a forecast. The question is only: which era does the AI wave most resemble?
| Scenario | CAGR | €10k → (20 yrs) | Historical anchor | What it means |
|---|---|---|---|---|
| Floor — The Bubble Pops | 5–8% | ~€27–47k | Post-2000 Nasdaq crash: −60%, five-year recovery | AI hype peaks, semis drop 50–60%. Same as Nasdaq after 2000: brutal half-decade, then recovery. |
| Base — History Repeats | 18% | ~€270k | SOX 2006–2024: same bottleneck logic, same duration | AI repeats what semis did in the internet wave. Same bottleneck logic, same duration. If history rhymes, this is the base case. |
| Ceiling — Bigger Than the Internet | 30%+ | ~€1.9M+ | SOX 2016–2021: avg. ~38% pa over six years | SOX averaged ~38% pa over six years (2016–2021) — peak years, not a sustainable 20-year average. A sustained horizon of 20–25% annually, if AI demand exceeds the internet wave, is not fantasy. 30%+ in exceptional years is possible. 30% compounded for two decades would be historically unprecedented. |
The Floor is real. SOX crashed 52% in 2008. Can you hold through five years of pain for the recovery?
The Base is simply: history repeats. Not a moonshot. The exact rate SOX already delivered, applied to the next infrastructure wave. And yet most people treat even this as extreme, because they're comparing it to a savings account, not to the last nineteen years.
What I Take From This
I didn't start with the semiconductor index. I started with a question: has anyone ever verified that owning the infrastructure bottleneck actually outperforms everything built on top of it?
I wanted historical evidence. So I pulled the Nasdaq 100 first, focused on fewer names, the clearest proxy for the last tech wave. I couldn't believe the numbers. 18.5× in nineteen years. That number barely shows up in financial media. And when it does, it's never next to a real estate chart.
Then I went one layer deeper. If the Nasdaq captured the tech wave, what enabled the Nasdaq? The bottleneck layer: chips. The actual silicon. The companies that made everything else physically possible. SOX: 23×. Better than the Nasdaq. Better than anything else in the comparison.
Warren Buffett-level returns. Without Buffett's skills. Without stock-picking. Without active management. Just the conviction to own the right layer and hold through the crashes.
That was the moment I stopped researching and started deciding.
I now invest primarily in Semis. No MSCI World. No real estate. No legacy indices. No diversification into the past.
Not because I am certain. Because the data shows that "safe" and "high return" have been structurally opposed for nineteen years.
The MSCI World is, at its core, a weighted average of yesterday's winners. Many of them are the companies being disrupted by AI, sitting alongside the ones driving it. In a regime change, owning everything means you own the decline alongside the ascent.
Real estate scales linearly. You buy it, you manage it, you pay for it in time and leverage. It anchors mobility in a world where capital needs to be fast. The data says: real estate with rental yield delivers ~€32,000 after nineteen years. The semiconductor index delivers €233,000. Same starting point.
You can disagree with my conclusions. You can't disagree with the table.
The question is whether you believe the next wave is bigger or smaller than the last.
I know where I stand.
Want my exact setup?
Write me directly.
Which ETFs, what allocation, how I think about rebalancing. Happy to share — just not something I publish openly.
mail@thoreschwemann.com →Newsletter
Want to find the 5x AI processes in your company? Get the framework. Unsubscribe anytime.
Not financial advice. No specific investment products are recommended. Scenarios are illustrative projections anchored to historical patterns, not forecasts. Past performance does not guarantee future results.