Before we dive in, here is the N-line summary.
Samsung Electronics recorded an operating profit of 89.4 trillion KRW in the last second quarter. They earned more than Nvidia. It's insane; they are the highest-earning private company on Earth for a three-month period.
However, on July 7th, when that announcement came out, the stock price dropped by 7%. It's a 'Lucky 777,' but why is this happening?
During the trading session, it even slid down to 9%.
A stock price drops 9% on the day it announces record-breaking performance. Does that make sense? Yes, it does. And understanding this one sentence is the entirety of today's article.
But you guys don't think Samsung Electronics is just about HBM, do you?
- Just how much money is 89.4 trillion KRW?
When numbers reach the trillions, it's hard to grasp. So let's break it down.
If you divide 89.4 trillion KRW by 90 days, it's about 1 trillion KRW per day. 1 trillion a day. If you break it down into seconds, it's about 11.5 million KRW per second.
In the 20 seconds it took you to read this one paragraph, Samsung earned 230 million KRW. They are earning it right now.
What's even crazier is this. Samsung Electronics' recent 3-year operating profit:
2023: 6.6 trillion KRW
2024: 32.7 trillion KRW
2025: 43.6 trillion KRW
The total for all three years is 82.8 trillion KRW. Yet, in the three months of Q2 2026, they earned 89.4 trillion KRW. They earned 6 trillion more than the previous three years combined.
Look at 2023 in particular. They earned 6.6 trillion KRW for the whole year. Now, they earn that in a week.
One more thing. That 89.4 trillion already has about 19 to 20 trillion KRW deducted for special employee performance bonuses. If that hadn't been deducted, it would be over 100 trillion. 100 trillion in three months.
- So why did the stock price drop on the day of the earnings announcement?
This is where it gets real. This is the point that amateurs never understand, and if you don't know this, you shouldn't buy Samsung Electronics.
To know the reason, you first need to understand the memory semiconductor market. Let's briefly revisit the 'Golden Cabbage Patch' story we learned before.
When cabbage prices skyrocket, cabbage farmers hit the jackpot. Seeing that, everyone starts planting cabbage. A few years later, there's a cabbage surplus, prices crash, and everyone goes broke together. Since the broke farmers stop planting cabbage, a few years later there's a shortage again, and prices skyrocket. An infinite loop.
Memory semiconductors are exactly like this. Let's use the term we learned last time.
It's called 'Cyclical.' You saw this in the SK Hynix article, right? No goldfish brains here who forgot, I hope?
Anyway, just think of it as a rollercoaster industry.
To give you an idea of how much DRAM prices have risen: in Q1 of this year, they rose 90-95% compared to the previous quarter, and in Q2, they rose another 58-63%. If you multiply the two, it's a 3x increase in just half a year. You've heard people saying computer RAM prices are insane. That's why. I heard the minimum for a gaming rig is now 5 million KRW. What a crazy world.
Now, here is the core.
The stock market doesn't look at how much you are earning now. It checks if you will earn even more in the future.
The forecast for DRAM price growth in Q3 is 13% to 18%. It went from 90% to 58%, and now down to 13%.
Don't get confused here. Prices aren't falling. They are still rising. It's just that the speed of the increase is slowing down.
In car terms: you were accelerating at 90km/h, then 58km/h, and now the acceleration is slowing to 13km/h. The car is still getting faster. But the person in the driver's seat knows: 'At this rate, I'll be hitting the brakes soon.'
Stocks hit those brakes 6 months to a year in advance. That's why the stock price drops even when earnings are at an all-time high.
In fact, the exact same thing happened in 2018. Samsung and Hynix kept breaking record earnings, but the stock price peaked 5 months early and dropped 40%. The news was setting off fireworks, but the stock charts were a funeral.
- Samsung is a golden cabbage patch, but they also run a kimchi factory
This is the critical point where Samsung differs from SK Hynix. Knowing this alone makes today's article worth it.
SK Hynix only sells cabbages. That's it. So when cabbage prices go up, it's just good. Period.
Samsung Electronics sells cabbages, but they also make and sell kimchi using those cabbages.
What I mean is that Samsung Electronics has two main parts:
The side that makes semiconductors (DS Division). This is the cabbage patch.
The side that makes Galaxy phones, TVs, and refrigerators (DX Division). This is the kimchi factory.
When golden cabbage prices skyrocket, the cabbage patch is rolling in money. But what about the kimchi factory in the same company? It gets hit with a cost bomb.
Do Galaxy phones not use memory? They do. TVs and refrigerators use it too. Samsung's own phone factory has to buy the semiconductors at the high prices Samsung set.
This isn't my guess; it's written exactly in Samsung Electronics' official disclosures. The phone and home appliance divisions mention 'increased cost burden' and 'limited profit improvement due to rising costs.'
In other words, Samsung Electronics is the only semiconductor company in the world shooting itself in the foot.
Therefore, we get this conclusion:
On the way up, Hynix rises more sharply. Because it's a pure cabbage patch.
On the way down, Samsung crashes less. Because when cabbage prices drop, the kimchi factory smiles.
Which one is better? The answer will be the exact opposite depending on where you think we are on the rollercoaster.
- You thought Samsung was #1 in semiconductors? You're only half right
This is where most amateurs are mistaken.
In the overall DRAM market, Samsung is indeed #1. As of Q1 this year, Samsung has 38%, SK Hynix 29%, and Micron 22%.
However, the most expensive and profitable item these days is something called HBM.
What is HBM? Many people explain it in complicated ways, but just think of it like this: if regular memory is a single-family house, HBM is a high-rise apartment built by stacking them. Because you put much more in the same space, it's much faster and much more expensive. This is needed for AI calculations. This is what goes into Nvidia graphics cards. Not hard anymore, right?
Now, let's look at the HBM market share.
SK Hynix: 58%
Samsung Electronics: 21%
Micron: 21%
Samsung is at about one-third. They are getting crushed by Hynix in the most lucrative market. This is exactly why Hynix's stock price has soared over the past two years while Samsung's has just crawled.
But the tables are turning.
On February 12th of this year, Samsung was the first in the world to mass-produce and ship the next-generation HBM, HBM4. They beat Hynix to it. In just four months, cumulative sales exceeded $1 billion, and at this rate, there's talk of reaching $10 billion by the end of the year. This is unprecedented for a new product's first year.
In terms of performance, Samsung is also ahead. Their speed is 13Gbps, while Hynix and Micron officially state 11.7Gbps. Since Nvidia requested 10 to 11, Samsung is producing goods that exceed the requirements. Honestly, these numbers aren't that important, so just read and move on.
What you need to see here isn't whether Samsung wins or loses in HBM, but what is currently reflected in the stock price.
Samsung's stock has long been discounted with the label of 'perpetual #2 in HBM.' When that label falls off, the stock price will jump before the earnings even improve. This is because the market's valuation of the company itself changes.
- Samsung has one more card that Hynix doesn't
Foundry.
What is a foundry? It's a factory that makes other people's chips for them. If Apple or Tesla says, 'We designed our chip, please make it,' they make it. In clothing terms, it's an OEM factory.
This market is dominated by Taiwan's TSMC with 71%. Samsung has 6.8%. It was simply a failing business. They've had trillions in losses for four consecutive years since 2022. Last year alone, the loss is estimated at 7 trillion KRW.
But the picture changed this year.
Tesla entrusted Samsung with the AI chips for autonomous driving. The factory built in Texas with a 54 trillion KRW investment has started running. Nvidia's Jensen Huang publicly said at an event in March, 'Samsung makes this.' The company itself forecasts that orders for the latest 2nm process will increase by 130% compared to last year.
So, the talk in the industry is that the foundry will turn a profit as early as Q3 this year. It's been four years.
Why is this important? Two reasons.
First, if the hole that was losing 7 trillion every year is plugged, profit increases by 7 trillion just from that. Even if they don't sell anything more, just by not losing money.
Second, and this is really important: memory is a rollercoaster, but foundries are less so. If Samsung establishes itself in the foundry business, the 'rollercoaster company' tag will fall off. Then the market's valuation goes up. Why do you think TSMC is valued so highly? Because they don't shake.
However, it's still a 'forecast.' It's not that they've made a profit, but that they likely will.
- This is where nine out of ten amateurs get fooled
Now it's time for you to open Naver Finance and look at Samsung Electronics' PER.
Let's clarify PER first. It's just a number representing 'how many years of earnings does it take to break even if I buy at the current price?' If the PER is 10x, you are buying it for 10 years' worth of profit, right? People say the lower it is, the cheaper it is.
As of mid-June, Samsung Electronics' PER was recorded at 24.8x, while Samsung's 5-year average is 13.6x.
'Oh? It's double the average? It's too expensive... I can't buy it.'
If you thought this, you've fallen right into the trap.
That 24.8x is calculated based on last year's profit. Since they earned 43 trillion last year, that's the basis. But the brokerage forecast for this year is 350 to 380 trillion KRW. They will earn more than 8 times as much.
If profit grows 8 times? The PER shrinks to one-eighth. 24.8x drops down to a single digit.
'Oh, then it's super cheap, buy everything!'
No? The real trap starts here. Remember just this one line today.
For rollercoaster stocks, the time when the PER is lowest is the most dangerous.
Why? Think about it. Why does the PER get low? It gets low because profit is at an all-time high. Profit being at an all-time high means it's near the peak. What comes next? A decrease in profit. Then the PER goes back up, and the stock price knows this and drops first.
Conversely, when is the PER 100x or not even calculable because of a deficit? When performance is at rock bottom. That is the floor.
So, you read rollercoaster stocks like this:
Buy when PER is high, sell when PER is low.
It's the exact opposite of regular stocks. If you don't know this and jump in saying, 'Samsung's PER is 5x, it's totally undervalued,' you will catch the exact peak. Scary, right?
- Where is the stock price now?
The closing price on July 23rd is 268,000 KRW.
The lowest price a year ago was in the low 60,000s. It has risen more than 4 times in a year. Then, in mid-June, it hit an all-time high of 374,500 KRW during the session, and in a month, it has dropped nearly 30% to its current position.
To summarize: it rose 4x in a year and dropped 30% in a month.
This is Samsung Electronics. If you thought it was 'safe because it's the national stock,' change your mind now. This isn't a savings account; it's one of the largest, most volatile casino large-cap stocks on Earth.
- So what are the risks?
I'll point out three.
Risk one: China is catching up.
You've heard a lot about a Chinese company called CXMT lately, right? Yes, ChangXin Memory. Their DRAM market share jumped from 3% in Q1 last year to 8% in Q1 this year. It nearly tripled in one year. They are currently trying to raise 6.5 trillion KRW by listing on the Shanghai stock exchange.
They can't make HBM yet. The consensus is that there's a gap of over 3 years because they can't buy lithography equipment. The problem is regular memory. They are pushing in with prices 30% cheaper.
Here, the golden cabbage patch and kimchi factory story comes up again. Samsung has a high proportion of regular memory, while Hynix has a high proportion of HBM. In other words, China's volume offensive hurts Samsung more.
Risk two: Shareholders don't get to eat all the money earned.
In May of this year, the Samsung labor union was on the verge of a general strike. Their demand was for 15% of the operating profit to be given as performance bonuses. If the operating profit is 300 trillion, they are asking for 45 trillion.
As a result, an agreement was reached, and 19 to 20 trillion KRW was taken out as bonuses in Q2. Employees in the semiconductor division receive up to 600 million KRW.
For reference, the performance bonus standard set by Taiwan's TSMC is at least 1% of operating profit. Samsung had a demand for 15%... Korea's labor unions are truly something else.
And it's not over. Shareholder groups have announced lawsuits, saying 'it is illegal to promise performance bonuses from profits before taxes are even paid.'
From a shareholder's perspective, the fact is this: the company is earning record amounts, but a significant portion of that money is going somewhere other than the shareholders.
Risk three: Dividends are stingy.
The dividend set by Samsung Electronics is 9.8 trillion KRW annually. This has been the same for several years. Based on the current stock price of 268,000 KRW, the dividend yield is about 0.6%. It's worse than a bank savings account.
Profit has increased 8 times, but dividends remain the same.
However, there is hope. This dividend policy expires at the end of this year. There's a high possibility a new policy will come out in Q4, and there's talk of special dividends. Furthermore, reports say SK Hynix is preparing a massive shareholder return in Q4, so if Samsung flinches here, the money will go to Hynix.
- So, should I buy it or not?
Choose from these three. there is no right answer.
First, it's still the first half of the game.
Brokerages believe next year will be better than this year. Even if new factories are built, products won't come out until 2028, but AI demand is continuously increasing. If the foundry turns a profit and HBM market share is recovered, Samsung will be promoted from a 'rollercoaster company' to an 'AI infrastructure company.' Out of 36 brokerages, 0 have a 'sell' opinion.
Second, earnings are good, but the stock price peaks first.
It's a replay of 2018. The moment it's confirmed that the speed of price increases is slowing down, even if earnings keep hitting records, the stock price turns first. The fact that the DRAM price growth rate dropped from 90% to 13% is already a reality. In this scenario, the current 260,000 KRW is not the floor, but halfway down.
Third, a cliff is coming in 2028.
China pours out volume, the factories being built now all start running at once in 2028, and AI investment slips even once. Then the cabbage patch will be overflowing with golden cabbages. Remember that it was only 3 years ago, in 2023, that Samsung's semiconductor division had a 15 trillion KRW deficit.
- What should I look at right now?
July 30th, 10:00 AM.
On this day, Samsung Electronics will announce its finalized Q2 results. The 89 trillion mentioned so far is just a rough number thrown out; on this day, the numbers broken down by division will come out for the first time.
Look for exactly three things:
How much the foundry deficit has decreased. Confirm if the turnaround to profit is real.
What tone the management uses when talking about HBM.
Whether they speak about the second-half price outlook with confidence or caution.
Look at the tone rather than the numbers. The 89 trillion is already known history. The market doesn't buy what it already knows.
Final Conclusion
You feel like a fool if you don't buy a company that earns three years' worth of profit in three months, but if you do buy, it's already risen 4x and dropped 30% from the peak.
This is what the area near the top of a rollercoaster always looks like. You only know if it was the peak or still the middle after it's passed. Anyone speaking with absolute certainty right now, whether bullish or bearish, is lying.
This is just my personal opinion based on disclosures and news, not a recommendation to buy. Protect your own account.
P.S. RAM enthusiasts should read this twice.
