Welcome to this week’s edition of Cape May Wealth Weekly. If you’re new here, subscribe to ensure you receive my next piece in your inbox. If you want to read more of my posts, check out my archive. This week’s article comes to you from David Höhl, author of Simple Alpha. I greatly enjoyed David’s well-organized writing about the state of artificial intelligence - and I’m very happy that he’s willing to share his thoughts in this edition of Cape May Wealth Weekly. Enjoy, and make sure to subscribe to his own newsletter.

While it has become fashionable to write about bubbles, the more people talk about one, the more it suggests that we are still far from peak bubble territory.

What I like about Templeton’s quote is that it shows that bubbles are not binary. They exist on a sliding scale. On the scale he laid out, we are probably somewhere between skepticism and optimism.

Skepticism is still visible in events like July’s AI sell off and in the fact that NVIDIA is trading at its lowest forward earnings multiple in 10 years. At the same time, optimism is clearly building, reflected in hyperscaler CapEx spending and elevated valuations, particularly in private markets.

However, the best way to see that we are probably not close to peak bubble territory is to compare today with the last major bubble: 1999. From the end of 1995 through the end of 1999, the Nasdaq roughly tripled, with 1999 alone delivering about +86%. From October 1999 to the March 2000 peak, the Nasdaq almost doubled in less than six months. Many companies were valued at 50 to 100 times revenue.

The picture today looks very different. Over the past five years, the Nasdaq Composite is up around 80%. The Nasdaq 100 currently trades at roughly 21.5 times forward earnings, close to its long term median. AI darlings such as Micron and SanDisk trade at single digit forward earnings multiples, not revenue multiples. Currently, a record $7.91 trillion is sitting in money market funds, waiting to be moved into equity markets.

But as we know, things can change quickly. I expect record IPO activity and lower interest rates next year to fuel a volatile rally that could bring us closer to peak bubble territory in H2 2027 and beyond.

Please note that the following thoughts are David’s personal investment views. They are meant for illustrative purposes only, and are not meant to be construed as investment advice (Anlagevermittlung or -beratung) nor tax advice (Steuerberatung).

Very likely. As described above, I think we are still in the early innings of a forming bubble.

What differentiates this bubble from many others is that AI is already creating real economic and scientific value. Hyperscalers are beginning to see the impact of their CapEx spending through accelerating cloud growth. Thrive Capital just reported to its LPs that AI agents are achieving 98% accuracy, completing tax related tasks 30% faster and resolving IT tickets 50% faster. (I wrote more about narrative violating AI facts here.)

Nevertheless, if you look at stock market history, fear and greed have always driven bubbles and depressions, and I do not think that will change.

Once people see others making millions in a matter of days, comparison kicks in. They want in too. Search volumes rise, the media picks it up and the whole thing becomes self reinforcing. The gap between real value and price keeps widening until the first major players start taking chips off the table and trust begins to crack.

With more retail investors in the market than ever before, the bubble will get bigger, but it will also make the fall more violent.

One thing remains certain: Moving out of the market entirely is extremely costly. If you missed the best 20 days of the S&P 500 in 20 years from July 2004 to July 2024, your return went from 10.5% to 3.6% annually.

I see 4 viable ways that could hedge your portfolio against an AI bubble burst.

First, Gold.

Gold has delivered a historical annual return of around 9.2% since the 1960s.

Across the last five market periods in which the S&P 500 fell by at least 20%, gold gained an average of 6%, according to a JP Morgan Report. This makes gold a useful stabiliser that can perform well during both strong and weak equity markets. In bull markets, however, it will naturally tend to lag equities.

Gold has also had an exceptionally strong run over the past decade, returning around 223%. Historically, periods of unusually strong performance have often been followed by more muted returns.

One way to gain exposure is through the Invesco Physical Gold ETC, which charges a fixed annual fee of 0.12%. Each certificate is backed by physical gold, with the bars held in J.P. Morgan’s high security vaults in London. Bonus: If you are tax resident in Germany, gains might be tax free after a holding period of more than one year.

(Note by Jan: For a deep-dive into investing in gold, check out our recent Primer.)

Second, High Quality Software Companies.

What became apparent during the AI sell off in July was how well high quality software companies held up. Profitable software businesses with double digit growth, such as Duolingo, Monday.com and ServiceNow, outperformed the Nasdaq by roughly 10% to 25%.

At the same time, these companies now appear less vulnerable when AI darlings rally. Many have already fallen 50% or more from their highs, meaning a significant amount of pessimism is already reflected in their valuations. This creates an interesting asymmetry: they can provide diversification during an AI infrastructure sell off while still offering meaningful upside if their fundamentals remain strong.

Third, Bitcoin.

This might surprise some people, but Bitcoin at around $65,000 still looks relatively cheap given its limited supply and growing global adoption by companies and governments. While AI stocks became the new investor darlings, Bitcoin moved the other way.

Bitcoin has historically moved with tech stocks, but that link is not fixed. During the July sell off, the Nasdaq fell around 5% while Bitcoin gained around 8%.

I expect Bitcoin to remain correlated with tech, but less strongly over time.

Fourth, Berkshire Hathaway.

Berkshire is home to cash flow heavy businesses, which historically tend to hold up better in crises than equity markets overall. In 2022, Berkshire gained 4% while the S&P 500 dropped 18%. During the trade war sell off in March 2025, Berkshire gained 3.6% while the S&P 500 lost 5.8%.

However, this hedge does not always work perfectly. During Covid in Q1 2020, Berkshire performed roughly in line with the S&P 500.

For now, I would not be too concerned about hedging against an AI bubble. I would rather make sure I am invested in the right companies benefiting from AI. As time moves on, I personally plan to invest more in high quality, cash flow generating software companies that benefit from AI adoption while the market still thinks they do not.

If multiples get closer to 1999 territory, I will simply build more cash, earn around 2.5% interest on it and stay flexible. As always, simplicity is key.

About the Author: David was a Partner at Interface Capital, where he invested early in Lovable ($13bn), The Exploration Company ($2bn) and ARX Robotics (>€500m). He has been investing in stocks since the age of 14 and now runs his own hedge fund. You can read his views regularly on his Substack.