AI Stocks, Diversification & Conviction: What I've Learned

Prices are as of the October 6, 2026 close, from Yahoo Finance and the other sources linked below. Markets move fast, so verify current numbers yourself.

Two big warnings dominated the last couple of years: the dollar was about to lose its crown, and the AI bubble was about to burst. Neither played out the way the loudest voices said it would. I'm not writing this to gloat. I'm writing it because it taught me something about diversification, position sizing, and how to build conviction in a stock when everyone around me is nervous.

What Did Following the Loudest Predictions Actually Cost?

In 2024 the story was de-dollarization: BRICS would build alternatives and the US dollar would weaken for good. In 2026 the story flipped to "the AI bubble will burst." If you acted on both, you probably spent a lot of time on the sidelines.

Here is what sitting out cost. From the end of April 2024 to October 6, 2026, the S&P 500 rose from about 5,036 to about 7,819, roughly +55%. The Nasdaq Composite went from about 15,658 to about 27,600, roughly +76% (price only, before dividends). That is the 50–80% range of gains that a cautious investor can end up watching from the outside.

I'm not saying don't listen to experts. Some warnings are right, and the bubble debate is not over: well-known investors like Michael Burry are still publicly arguing it could pop sooner than expected (CNBC). What I try to do is pair what people say with what they do: where capital is actually being spent, and what people own with their own money. Words are cheap. Capex and positions are not.

Is the US Dollar Still Dominant in Global Trade?

Short answer: yes. It depends on how you measure, so here are the main yardsticks side by side:

So when I ask "what percent of global trade still happens in USD", the honest answer is: about 54% of exports are invoiced in dollars, and about 81% of trade finance runs through it. That is far above the US's share of world trade.

De-dollarization is not made up. The dollar's share of reserves has slid from roughly 70% around 2000 to about 57% now, and some BRICS trade is now settled in local currencies. But the Federal Reserve's review of the evidence found the dollar's overall international role has changed little over the past five years. The pattern I take from this: structural shifts happen over decades, but headlines price them in over weeks. The dollar index also wasn't a straight line: it climbed about 5% from January 2024 to January 2025, dropped to around 97 in early 2026, and sits near 102 today.

Are Companies Still Doubling Down on AI?

This is the data point I weigh most heavily. The four biggest US hyperscalers (Amazon, Microsoft, Alphabet, and Meta) have guided to about $725 billion of capital spending in 2026, around 77% more than the roughly $410 billion of 2025, with most of it going to AI chips, data centers, and infrastructure (AI Weekly).

That is a lot of people putting real money behind the AI growth story. But I don't read it as a guarantee. FactSet has noted that spending is starting to outrun cash flow and companies are tapping external financing (FactSet), and when Alphabet's report disappointed in July, it sparked a sell-off across the group (CNBC). Heavy spending is a reason to pay attention, not a reason to stop thinking about price.

META: The Dip Nobody Wanted to Buy

Meta is the clearest example I have of fear peaking near the low. The stock's lowest close of the last three months was $539.03 on July 30, a stretch when worry about Big Tech's AI spending was loud. At $738.88 it's now about 37% above that low, and up roughly 20% in just the last month.

I can't point to one reason for the rebound, and I'd be skeptical of anyone who says they can. The honest lesson is smaller: when a quality business is falling and the headlines are loud, it feels risky, and that feeling is often strongest right near the bottom. Of course, that's hindsight. At the time I couldn't have known July 30 was the low, which is exactly why I scale into positions instead of trying to catch the bottom (more on that below).

MELI: Why I Care About Diversification Beyond US Big Tech

Here's a point I think gets missed. Owning several large US growth stocks can quietly be one single bet. T. Rowe Price notes that AI infrastructure and hyperscalers make up nearly 60% of the Russell 1000 Growth Index (T. Rowe Price). If AI is going right for everyone, that's fine. If it isn't, many of my "different" holdings would fall together.

MercadoLibre (MELI) is a different engine: e-commerce and fintech across Latin America. In Q2 2026 its revenue passed $10 billion for the first time, up 50% year over year, with payment volume up 56% to $101 billion (Business Wire). It's not a clean story: net income was $466 million, down from $523 million a year earlier because of heavy investment, and the stock sits about 23% below its 52-week high. But it's a growth story that doesn't depend on the same drivers as the US AI trade, and that is what I mean by diversification: different geography, different business model, different risks.

Diversification goes beyond stocks, too. Some of my exposure is in assets with a very different risk profile, like Bitcoin (see how many public companies now hold it and the DCA calculator for how steady buying works).

NFLX: The One I'm Still Scared Of

If META is the dip I wish I'd leaned into, Netflix is the one I'm still wrestling with. It closed at $68.69, about 45% below its 52-week high of $124.86, and it's down about 10% over the past three months. It's lower even than when I ran my Netflix valuation walkthrough late last month.

I'm not telling anyone to buy it, and I haven't decided how I'll act on it. What I notice is the same feeling I had with META: the stock is falling, the news around it (a contested Warner Bros. Discovery bid, softer guidance) is noisy, and buying feels uncomfortable. That discomfort doesn't tell me the stock is a bargain, and it doesn't tell me it's a trap. It tells me I need a process instead of a mood.

That process is where position sizing comes in.

How I Build Conviction Instead of Chasing It

For me, conviction isn't a feeling of certainty. It's a short list of things I can write down:

1. A thesis in plain words, including what would prove me wrong.
2. A valuation check. I use the valuation scorecard so I'm comparing the same signals every time.
3. A small starting size, so I can stay calm and keep learning.
4. Adding on evidence, not on price. A stock being lower isn't a reason to add; improving fundamentals might be.
5. Watching what insiders and big spenders do, not only what they say on TV.

How Much of My Portfolio Do I Put in One Stock?

Here is my one hard rule: I never let any single stock or asset be more than 10% of my portfolio. That includes the ones I like the most, and it includes Bitcoin. Common rules of thumb for single-stock limits sit around 5–10%, with lower caps for riskier names, so I'm not inventing something unusual.

In practice the ceiling is the maximum, not the target. For higher-risk names, I stay well under it. In my AXON post I explained why I cap a volatile, richly valued stock at about 3–4% and start with only a fifth of that. The 10% ceiling is for the rare position I understand deeply and that has earned its place.

The downside math is why I like having the ceiling written down:

A stock like Netflix has already fallen about 45% from its high. Whether it keeps falling or recovers, a sensible size means either outcome is survivable and I'm still thinking clearly. That is what lets me hold a view with conviction instead of defending an oversized position out of fear.

What I Take From All This

• Loud predictions (the dollar collapsing, the AI bubble bursting) are worth listening to, but they're not a portfolio plan.
• Fear tends to be strongest near lows, but I can only see that in hindsight, so I scale in instead of guessing the bottom.
• Diversification means different drivers, not just different tickers.
• Position sizing is what turns "I might be wrong" from a disaster into a lesson.

FAQ

Is this financial advice?

No. It's education and my own perspective, and it says nothing about what's right for your portfolio.

Is the AI bubble about to burst?

Nobody knows. Valuations in parts of the AI trade are high and well-known investors are warning about a bubble, but the biggest AI companies are also very profitable and still raising spending. I don't try to time it. I diversify and size positions so a bubble bursting wouldn't wreck me.

Is the US dollar losing its role in global trade?

Slowly at the edges, not at the core. Its share of official reserves has drifted from about 70% around 2000 to about 57%, but it still covers roughly half of SWIFT payments by value and about 81% of trade finance.

How much of my portfolio should be in one stock?

There's no universal answer. Common rules of thumb cap one stock at roughly 5–10%, lower for riskier names. I never go past 10% in any single stock or asset, and I keep riskier ideas near 3–4%.

Why diversify if AI stocks keep winning?

Because many US growth stocks are effectively one big AI bet. Diversifying across geographies, business models, and asset types means one theme going wrong doesn't decide your whole outcome.