When Grantham Rings the Alarm Bell
Jeremy Grantham appeared on The Diary of a CEO on June 25, calling the AI boom the largest investment bubble in American history. He is forecasting a potential 70% decline and advising investors to hold zero US equities. I read it, sat with it for about a week, and kept coming back to the same thought: if he is right, my portfolio is exactly where the damage would happen.
His record on identifying speculative excess is among the best in the industry. When he says the US market is trading above dot-com peak valuations at 35 to 40 times earnings, I pay attention.
His prescription is to get out of US equities entirely, rotate into non-US indexes, hold bonds yielding four to five percent, add precious metals, and hold zero crypto. My portfolio looks almost nothing like that. I have roughly a third in Physical AI and robotaxi names. Another third is China. Crypto is under 10 percent. My only hard asset exposure is a residual position I have been trying to exit for months.
So the question I sat with was not whether to tweak a few positions. It was whether Grantham’s warning was serious enough to rethink the entire direction of the portfolio.
I have two problems with his escape route.
The first is structural. His prescription assumes that non-US markets will hold up when the US corrects. History suggests that when US markets experience a major correction, most global markets sell off initially as well. Rotating into international indexes is not protection from a crash — it is a bet on faster recovery during the aftermath. That is a different argument, and a weaker one.
The second is tactical. Even if you accept that bet, the easy money has already been made. Korea and Taiwan are semiconductors — correlated to the same AI capex story Grantham is warning you about. Singapore is defensive but not cheap. India has never been cheap. The broad EM mean reversion he is pointing to was last year’s trade. Investors rotating into international indexes today are buying what has already re-rated.
The China re-rating is a partial exception, and I am already there. But I could not convince myself that abandoning my current positioning for his prescribed rotation was a better risk-adjusted move.
Before deciding what to do, I mapped my portfolio against his warning. Only around 9 percent sits directly inside what Grantham would call the AI bubble — my active tech trading positions. Beyond that, roughly another third faces contagion risk in a genuine crash scenario. The positions most insulated are my Singapore-listed names and China healthcare holdings. So about half my portfolio feels some version of this risk. That is not a comfortable number, but it is also not the “flee everything” scenario Grantham is prescribing.
Here is the distinction worth making. Grantham’s actual targets are the hyperscalers and semiconductor names that have run 50 to 200 percent this year on the back of an AI capex arms race. My robotaxi names tell a completely different story. PONY is down 48 percent over the past year. WRD is down 28 percent. XPEV is down 28 percent. GRAB and UBER have de-rated roughly 19 to 20 percent over the same period. These names have been getting sold down consistently while US AI infrastructure stocks ran up. They are not riding the bubble. They have already been through a significant de-rating.
The question I kept asking myself is not whether they can fall further — of course they can. It is whether these businesses will be worth more five years from now than they are today. GRAB is profitable and growing revenue 24 percent year on year. WeRide reported 58 percent year-on-year revenue growth in Q1 2026. PONY reported 145 percent revenue growth in Q1. A market correction does not change any of that. That is a contrarian bet, and I am holding it.
Which brings me to the direct exposure — my active tech trading positions: AAOI, ONDS, and FROG.
I approach these names very differently from my Physical AI or healthcare holdings. For robotaxi names, I hold with conviction through volatility. For healthcare, I take a contrarian view and buy into weakness. For tech, I trade actively and manage risk through discipline. These are not long-term thesis positions. They are momentum trades, and momentum trades require exit rules.
I rank them by how far they have run. AAOI is up over 200 percent this year alone and nearly 300 percent over the past twelve months — an AI datacenter optical components play sitting squarely in the middle of the capex bubble Grantham is warning about. ONDS is down about 24 percent year to date but up over 300 percent in the past year, a speculative drone and autonomous systems name that is thinly profitable on a trailing basis. FROG was up about 50 percent year to date and over 100 percent in the past year, a DevOps SaaS business that is profitable on an adjusted basis with the strongest fundamentals of the three.
Interestingly, that didn’t change how I managed it. I sold FROG yesterday at a gain of just under 19 percent. It was trading above its previous 52-week high and the thesis was simple — take the profit off the table.
ONDS fell another 6% yesterday and I’m now down around 31%. The interesting part is that the business itself hasn’t deteriorated. Revenue continues to grow strongly, management raised guidance, and the order backlog remains healthy. The market is punishing AI sentiment rather than execution. That gives me enough conviction to hold, but not enough to average down before the chart stabilises.
AAOI fell nearly 13 percent yesterday — exactly the kind of move I was exposed to without a stop-loss. I am now sitting on a total loss of around 22 percent. The lesson I learnt from this mistake is straightforward: if you are trading semiconductor or AI infrastructure names that have run 200 to 300 percent in a year, a stop-loss is not optional. When sentiment shifts on the AI narrative — and Grantham’s warning is exactly the kind of catalyst that shifts sentiment — the unwind can be brutal and fast. Sometimes the biggest risk isn’t owning the wrong stock. It’s having the right stock but no exit plan.
The one structural gap Grantham’s interview reminded me of is one I have been meaning to address for months. I wrote about hard asset exposure earlier this year and concluded it was a missing piece in my portfolio. I still have none in any meaningful sense. Gold prices have been falling and I have not rushed to enter, but this is a useful reminder that I keep identifying the gap and not filling it. I will keep watching for a better entry into physical gold.
The honest answer to Grantham’s warning is not that I think he is wrong about valuations. He might be completely right. It is that a bubble warning does not automatically tell you where to hide, and rotating out of already-battered deployment-layer names into already-re-rated international indexes is not a trade I can make myself do on valuation anxiety alone.
In the end, I didn’t and couldn’t sell everything and run. I just got more honest about which positions deserve conviction and which ones need a leash.
The usual disclaimer applies. This is not financial advice. I am a retail investor sharing my personal portfolio thinking. Do your own due diligence before making any investment decisions.