What does Samsung’s 3.2x forward PE 2027 actually mean if the forecast behind it requires roughly $1.25 trillion in operating profit over three years?
Marcos Milla ranks Samsung at 3.2x forward PE 2027, Micron at 5.2x, and NVIDIA at 16.2x. Samsung and Micron appear remarkably cheap, but a multiple is only as reliable as the earnings beneath it.
Samsung’s cheap multiple embeds an extraordinary forecast
One author questions the expectation that Samsung will generate roughly $1.25 trillion in operating profit over the next three years. According to that author, meeting the forecast would require at least $40 trillion in global AI spending, or approximately $13 trillion annually. Stated current worldwide IT spending is $7 trillion.
That does not prove the forecast is wrong. It does suggest that Samsung’s 3.2x multiple may reflect an extraordinary earnings assumption rather than an ordinarily cheap business.
Today’s shortage cannot establish three years of profits
The counterargument is visible in the physical market. Memory demand currently exceeds supply. Producers are allocating supply and raising prices. The Samsung author expects DRAM to enter another generational run, with demand substantially exceeding supply.
Apple is also facing major MacBook Air shortages amid a memory and silicon crunch.
The scarcity is real, but its duration and financial impact remain unknown. A shortage today does not establish three years of exceptional profits.
The uncertainty sits inside the denominator
Cyclical businesses often look cheapest when expected earnings are highest. If supply allocation, price increases, and product delays persist, today’s low forward multiples could prove genuinely cheap.
If the cycle changes before those earnings arrive, however, the denominator falls and the apparent bargain can disappear. The low multiple does not remove uncertainty. It embeds that uncertainty inside the earnings forecast.
Pabrai prefers optionality to required optimism
Mohnish Pabrai puts memory and AI in the too hard pile. He acknowledges that memory manufacturing is protected by patents, scale, engineers, process expertise, and what a Micron CFO called “black magic.”
Still, Pabrai says he cannot know where the industry will be in three or five years, or whether one producer will pull ahead.
His contrast is Kaspi. The company produces $2 billion in annual cash flow, trades at roughly five to seven times cash flow, and has a dividend yield approaching 10%. Its Kazakhstan operation is still growing, while Turkey is an uncertain moonshot.
In Pabrai’s thesis, the core cash flows provide downside protection even if Turkey fails. The uncertain expansion is optionality, not a condition required to justify the valuation.
Samsung at 3.2x forward PE 2027 may offer more upside, but that estimate depends on a memory cycle whose scale and duration are contested. Kaspi’s valuation is tied to cash flow Pabrai considers relatively certain.
Higher bond yields raise the burden of proof
One speaker cites a 10 year Treasury yield of 4.6% and a 1 year yield of 4%. When government bonds offer material yields, equities dependent on distant or highly cyclical forecasts must clear a higher bar.
Cheapness should come from cash flows that survive scrutiny, with uncertain growth left as optionality, not from assuming today’s scarcity lasts for years.