Hey {{first_name|Investor}} -
I get some version of the same question almost every week. Which stock is about to double, which name in optics is the winner, where the money should go today.
If I actually knew which stock would jump the most over the next three months, I'd have quietly put everything I own into it and kept the secret to myself. I don't, and nobody selling you that promise does either.
I've watched the alternative play out too many times. Someone buys a name they found on Reddit, it slides twenty percent a week later, and because they never understood why they owned it, they panic and sell at the worst possible moment. The tip wasn't the problem. Owning something you couldn't explain was.
So I don't hand out hot picks. What I can give you is worth more and lasts longer: a way to look at any corner of AI infrastructure, figure out what really drives it, and reach your own call. Think of the old line about teaching someone to fish. That's the trade here, not this week's hot name, but a method you can run for the rest of your investing life. The one twist is that the lesson got much faster this year, because AI tools now do in an afternoon what used to take me weeks of reading.
This is part one of three. Today is the mindset and the overall process. Next week, exactly how I teach myself a field I know nothing about. The week after, how I judge a single company and decide when to buy or sell. Each part stands on its own, so start wherever you like.
Start with the sector, not the stock
Most people do this backward. They hear a ticker, get excited, and only then try to justify it. I go the other way. Before I look at a single stock, I ask which sector deserves my attention at all.
Right now the biggest theme of our time is AI infrastructure. Think of it as the picks and shovels of the AI gold rush. Whoever wins the software race, someone still has to build and power the machines underneath, and that layer is where I spend my time. Every new AI model runs on a physical stack: chips to do the math, memory to hold the data, fiber and lasers to move it around, and an enormous amount of electricity to run it all. That stack is real, it's expensive, and it gets bought whether or not any single app becomes a hit. Inside the big theme sit smaller sectors: compute, memory, optics, power, networking. Each is its own world with its own leaders and bottlenecks.
My first job is to choose the sectors I understand well enough to hold a real opinion on. Conviction gets thrown around a lot, so let me be plain about what it means to me. It isn't a strong feeling. It's the quiet confidence that comes from actually understanding why a sector matters, who the players are, and what could break the story. You can't borrow that from a newsletter, mine included. You build it.
Good news for anyone who feels behind. You don't need to master every sector. You need one or two you understand better than the average investor, and you let that edge do the work. I know memory and optics. I hold much softer opinions on the rest, and I size my attention accordingly. Depth in a couple of areas beats a shallow take on all of them.
How do you find your one or two? Start with what you already brush up against. Maybe your job touches data centers, maybe you followed the memory shortage in the news, or maybe you just find one of these worlds more interesting than the rest. Curiosity counts as an edge, because it's what keeps you reading long after the excitement fades. The sector you'll actually study beats the one that looks smartest on paper. Which corner of this build-out could you explain to a friend without notes? Start there. If the answer is none of them yet, that's fine, and it's exactly what part two is about.
Draw the map
Once I trust a sector, I draw a map of it. A map answers two simple questions. How does this sector break into themes, and which companies sit in each one?
Compute is the easy example. Start with the chips that do the heavy lifting. GPUs, the workhorses of AI training, point to NVIDIA ($NVDA) and AMD ($AMD). CPUs, the general-purpose brains running alongside them, bring in Intel ($INTC), AMD again, and a handful of others. That's the obvious layer, the one everybody knows.
Now stretch the map further, because that's where it gets useful. Someone has to manufacture these chips, so you add the factories. Someone has to package them, stitching several chips into one tight bundle so they act like a single brain. Someone has to test them before they ship. Each step is a link in the chain, and each link has its own set of companies. A good map keeps pulling that thread until you see the whole picture, not just the famous name at the front.
Memory works the same way. At the front sit the big makers of the chips that store data. Behind them are the companies supplying the special materials and machines used to build those chips, plus the ones that stack the memory into the dense modules modern AI demands. Same exercise, different sector. Draw the layers, then fill each with names. Do it two or three times and the motion starts to feel natural.
Compute and memory map cleanly. Other sectors fight you. Optics and power are the hard ones, full of unfamiliar terms and layers that don't sort themselves neatly. That difficulty is exactly why they reward the effort. When a sector is hard to map, most investors stop at the surface, and the ground underneath stays less crowded.
The reward for pushing that far: the most interesting name often isn't the famous chip at the front. It's a quieter company a layer or two back that sells into several big names at once and gets paid no matter which brand wins the headline fight. You find it only by drawing the map all the way down, which is why so few people bother.
You won't draw a perfect map on the first try, and you don't need to. Mine get redrawn constantly. The act of mapping forces you to ask where each company sits and what it does, and after enough reps you can build one for a new sector yourself.
You're not too late, and small is fine
A quick word for anyone in their late thirties or forties who feels like the train already left. It didn't. This build-out is measured in decades, not months, and the goal most of us share is to become work-optional over years of steady, sensible decisions, not to hit one lucky jackpot.
That long horizon is a gift, because it rewards understanding over speed. You don't need to know everything by Friday. Pick one sector, learn it properly, and add the next when you're ready. A single area you truly understand serves you far better than a dozen tips you can't defend. Slow and understood beats fast and borrowed, every time.
Conviction is something you build, not something you buy
Notice what these steps have in common. Neither involves a stock tip. Both are about understanding, and understanding is what holds up when the market gets scary.
When you own a position because a stranger online told you to, the first bad week shakes you out, because you have nothing of your own to stand on. When you own it because you understand the sector and mapped the company yourself, a rough week is just weather. You know why you're there. That steadiness is worth more over a lifetime of investing than any single call.
So that's the foundation. Pick sectors you can genuinely understand, map them until the whole chain is visible, and let real knowledge, not a hot tip, decide where your money goes. It's slower than chasing a ticker, and it's the only version I've found that lets you sleep.
Next Thursday I'll get practical about the hardest part: how do you learn a field from zero when the words alone make your eyes glaze over? That's where AI has quietly changed the game, and I'll walk you through the exact routine I use.
If you'd rather not wait week by week, my book Make Your Own Alpha lays out the whole method in one place. The title is the idea: an edge you build yourself is the only kind worth having.
Stay disciplined - Koh
Disclaimer: Nothing in this newsletter constitutes investment advice or a recommendation to buy or sell any security. Numbers and observations are as of publication. I may hold positions in companies discussed above. Always do your own research and consult a licensed financial advisor before making investment decisions.