Samsung is on track to book roughly $282 billion in operating profit this year. That's more than Nvidia will earn, and yet the market values Samsung's profits at a steep discount. Reuters' Breakingviews column laid out the gap plainly this week, and it deserves a slow read, because it explains something a lot of investors get wrong about how the market prices a business.

The short version: Samsung sells memory chips, and memory chips are a commodity. When you sell a commodity, your profits depend on supply and demand in a way that even excellent management can't fully control. Nvidia, meanwhile, sells chips that customers currently can't get anywhere else. Same industry, in the loose sense. Completely different pricing power.

I've watched this movie a few times now. Memory makers have these glorious stretches where pricing runs hot and the profits pile up, and then a capacity glut arrives and the same companies post quarters that make you wince. Micron has been riding the same wave, and I called it what it is: a cyclical memory business wearing an AI costume. Samsung's earnings look enormous right now because AI demand has tightened the memory market. The market isn't skeptical about this year's profits. It's skeptical about whether those profits survive the cycle.

Then there's the capital spending problem, and this one deserves real math because it's the heart of the whole story. Say a company earns $282 billion in operating profit. Sounds like a money machine. Now say that keeping the machine running means plowing tens of billions back into fabs every single year, forever, just to stay where you are. What's left for shareholders is the residual, not the headline. A software or design company with lighter capital needs converts far more of each profit dollar into something a shareholder actually keeps. The market pays for the dollars you keep, not the dollars you announce.

Nvidia sits at the other end of that scale. Its chip designs are manufactured by others, so its capital intensity is a fraction of Samsung's, and its software ecosystem keeps customers anchored in a way memory has never managed. That's why the market pays a premium multiple for Nvidia's earnings and a discount for Samsung's larger earnings. It can feel unfair if you own the cheaper one. It isn't unfair. It's arithmetic.

Does that make Samsung a bad investment? No, and that's not my point. Cheap cyclical stocks have funded plenty of retirements, including a few good stretches in my own portfolio over the decades. The discipline is sizing. A cyclical trading at a low multiple can stay cheap for years while the cycle turns against it, and the low multiple is often a warning label rather than a bargain tag. If you buy it, buy it as a cyclical, with position sizing that assumes the bad year is coming eventually. Because it is.

For most people reading this, the practical lesson is simpler than picking between the two. The whole dust-up comes down to this: profits and shareholder value are different things, separated by capital intensity, competitive moats, and where you sit in the cycle. A total market index fund owns both companies in sensible proportions and spares you the job of guessing which multiple expands next.

I've been doing this since the late '80s, and I still can't call cycle turns. But I can read a cash flow statement. It rarely lies.

As always, this is my perspective, not professional financial advice.