The End of the Cycle, The Beginning of Contracts

SEMICONDUCTOR NARRATIVE · 2026.06.08

The end of the cycle,

The start of the contract

Samsung Electronics · SK Hynix

Memory is no longer just a cyclical commodity trade — when beliefs change, prices change too

June 8, 2026 · Narrative composed after comprehensive research and discussion by the Goguma Team
Old Beliefs · Transition to a Contract Industry · Bull/Super-Bull Take-Profit Targets · Entry and Realization

Onion (Macro) · Acorn (Industry) · Peanut (Risk) · Carrot (Chart) · Sweet Potato (Comprehensive)

Prologue

On the day fear shook prices, the beliefs of an era were put to the test

On June 8, 2026, the KOSPI collapsed by over 8%, halting trading. The market reacted with an old instinct — "Memory is just a cycle, after all." However, on the same day, the person building the world's most expensive computers said in Seoul that he would "buy five more years' worth" of memory from two Korean companies. That instinct and that promise cannot coexist.

This report does not cover a simple decline, but the moment a belief changes. For decades, memory semiconductors have been treated as a "cyclical industry that rises and falls," and as such, they have received low multiples regardless of how good their earnings were. The June 8 crash was also a product of that old instinct. But at the bottom of the industry, something else is happening — memory is shifting from a 'market-price trade' where prices are haggled every quarter, to a 'contract industry' where demand is guaranteed for five years.

If this transition is real, the yardstick used to evaluate these two companies changes. And when the yardstick changes, the price changes. This report verifies whether this shift in belief is true, and if so, maps out how far the share prices of these two companies can go — based on brokerage figures and the latest industrial facts — up to the final take-profit price.

Chapter 1

Collapsed prices, and the fundamentals that weren't there

The name of Korean memory was not the cause of the June 8 crash. The trigger was the US — disappointment in the guidance of custom AI chip companies and interest rate fears fueled by strong employment. That shock spread to foreign and institutional selling, and the forced liquidation of debt-financed stocks amplified it into a circuit breaker. It wasn't that memory earnings had broken, but that the entire market pulled out at once.

So, even in the midst of the crash, three signals pointed in the opposite direction. Samsung Electronics and SK Hynix bounced back from the regular session lows to 303,000 KRW and 1,978,000 KRW respectively in after-hours trading, the exchange rate, which was the core driver of foreign selling, fell from 1,559 to 1,529, and US index futures held up without further collapse. Above all, both stocks defended their 60-day moving averages (approx. 270k for Samsung, 1.8m for Hynix), the lifeline of a trend, at the closing price. 60-day moving average (Samsung approx. 270k, Hynix approx. 1.8m) at the closing price. Fear shook the price, but it could not break the trend.

Jensen Huang · June 8, Seoul

"SK Hynix has been Nvidia's largest memory partner, and it will continue to be so. We are already procuring billions of dollars worth every year, and it will grow even larger." — On the same day, cooperation on HBM4, HBM5, and foundry was discussed with Samsung Electronics, and the Nvidia-SK Hynix partnership for multi-year next-generation memory for AI factories was formalized.

Prices were determined by fear, and the demand underlying those prices was solidified by the words of the end customer. To resolve this contradiction, we must first address why the market is so suspicious of memory.

Chapter 2

The old belief — "Memory is a cycle"

Forward P/E is about 6x for Samsung and about 5.2x for SK Hynix. Even as earnings head toward record highs, the market assigns only single-digit multiples. This is not evidence of undervaluation, but a trace of an old belief.

For decades, memory has been defined by 'the repetition of booms and busts.' It was a so-called chicken game industry where everyone builds factories when prices rise, prices crash when supply overflows, and companies endure losses until the next boom. Even at the peak of the 2018 supercycle, and even after the promise of 'escaping the cycle' in 2023, prices eventually collapsed. So the market learned: "When memory earnings are at their best, it is the most dangerous time."

This belief was carved directly into valuations. The fact that P/E is lowest when earnings are at their peak is because the market discounts it in advance, thinking "these earnings will soon break." This is why the paradox of a 5-6x P/E exists even now as net profit is doubling or tripling. The panic selling on June 8 was also an expression of this instinct. Memory stocks are the first and hardest hit by signals of crisis.

If this belief is correct, the current bull market is just another cycle peak and a crash is coming. But if the premise of that belief has changed this time, all calculations change.

Chapter 3

Where the belief breaks — From commodity trade to contract industry

The essence of the change occurring at the bottom of the industry is this: Memory is moving from a 'commodity trade where prices are haggled every quarter' to a 'contract industry where demand is guaranteed for years.' The very structure that created the amplitude of the cycle is changing.

Demand has changed — Infrastructure, not replacement

In the past, memory demand was tied to the 'replacement cycle' of PCs and smartphones. The speed at which people bought new phones was the memory cycle. But now, the primary driver of demand is AI data centers. Data centers consume about 70% of the memory produced worldwide, and this is tied to the long-term infrastructure investment of Big Tech, not the whims of consumers. Google, Meta, and Microsoft are not cutting their 2026 capex; they are increasing it. When demand shifts from 'replacement' to 'infrastructure,' that demand becomes a trend, not a cycle.

Supply has changed — From haggling to contracts

A more decisive change is the transaction method. AI customers now see 'securing volume' as a greater risk than price. Therefore, they are signing **5-year Long-Term Agreements (LTA)**with Samsung, SK Hynix, and Micron to lock in future volume in advance. HBM's 2026 volume is already sold out, and margins are 3-5 times higher than commodity DRAM. SK Hynix stated, "We will reduce the volatility unique to the memory industry through long-term contracts to elevate the market's assessment to the next level." The industry where prices fluctuated every quarter is changing into an industry where performance is visible in advance through contracts.

So the yardstick changes — From P/B to P/E

The market has begun to read this transition first. SK Securities changed SK Hynix's valuation method from asset-based (P/B) to earnings-based (P/E), citing "long-term growth and stability of memory within the AI cycle." This is not just a target price upgrade, but **a declaration that "we will no longer view memory as a cyclical industry."**This is why the recent memory rally is interpreted not as 'the resolution of undervaluation,' but as 'the restoration of trust in the structural improvement of earnings generation.' And on June 8, the most powerful witness to that trust appeared — Nvidia said, "We will buy for five more years."

Key points of Chapter 3. Demand is shifting from replacement to infrastructure (trending), transactions from haggling to contracts (reduced volatility), and yardsticks from P/B to P/E (revaluation). If these three changes occur simultaneously, the premise of the old belief that "memory is a cycle" collapses. That is the logical basis for the resolution of the 5-6x P/E undervaluation.

Chapter 4

Is it still a cycle? — Final counterarguments and their limitations

One should not naively believe in this transition. The phrase "this time is different" has echoed at every cycle peak. So we face the strongest counterarguments head-on.

The counterarguments are heavy. Some on Wall Street are already warning of the "end of the supercycle." Micron's gross margin of 81% is a record high, surpassing the 2018 peak (59%), and historically, memory stocks have turned 1-2 quarters before peak margins. If capex competition among manufacturers ignites again, an oversupply could come in 2027-2028, and the entry of China's CXMT into commodity DRAM will lower the price ceiling. Even if the structure has changed, the cycle eventually returns when supply catches up with demand.

However, this counterargument has a limitation in timing. The most honest leading indicator of the end of a cycle is inventory, and current inventory is normal at 2-4 weeks(at past peaks, it piled up to 15-31 weeks). Full-scale mass production of new facilities is mid-2027, and mass production of high-performance memory by China's CXMT has been delayed to 2026. In other words, the 'rebalancing point' where supply catches up with demand is 2027-end to 2028 at the earliest. Even if the counterargument is correct, there is still time until it becomes reality. The cycle will return someday, but until then, the fruits of the structural transition will be reflected in earnings and share prices first.

Conclusion of Chapter 4. The counterargument that "memory is a cycle after all" may be the truth of late 2027-2028. But until then — while inventory is normal and contracts support performance — the shift in yardsticks (P/B to P/E) will be reflected in prices. The conclusion is neither "buy" nor "sell." Enjoy the fruits of the transition, define the signal for the cycle's return in advance, and realize gains in parts when reaching the target price.

Chapter 5

When beliefs change, prices change too — Bull/Super-Bull Take-Profit Targets

The core is simple arithmetic. If memory is a contract industry, not a cyclical one, the yardstick shifts to P/E, and a 5-6x P/E is revalued to a normal multiple. We map out how far that revaluation pushes prices — above the actual target prices of brokerages.

🎯 SK Hynix — Direct beneficiary of the transition

Current price 1,911,000 KRW. The basis is its ~70% share of Nvidia HBM, multi-year partnership, net profit path of 101 trillion KRW in 2026 and 142 trillion KRW in 2027, and the performance visibility created by long-term contracts.

Range	Take-Profit	vs Current Price	Basis
Bull	3,000,000 KRW	+57%	SK Securities target 3m. Applying P/E yardstick to 2026 net profit of 101t
Super-Bull	3,800,000 KRW	+99%	Shinhan/Mirae target 3.8m. 2027 net profit 142t + contract industry revaluation + Nvidia multi-year partnership

→ Once the yardstick changes to P/E and cyclical doubts fade, **double the current price (approx. 3.8m)**comes into view.

🎯 Samsung Electronics — Revaluation of supply chain entry

Current price 295,500 KRW. Entered Nvidia's supply chain with HBM4, HBM5, and foundry, and secured supply for AMD's next-gen accelerators. The basis is the room for resolution of the largest undervaluation (6x forward P/E).

Range	Take-Profit	vs Current Price	Basis
Bull	500,000 KRW	+69%	SK Securities target 500k. P/E normalization (6→13x) + HBM4/foundry profit contribution
Super-Bull	600,000 KRW	+103%	Above Shinhan target 550k. HBM5/foundry full-scale + memory/foundry joint revaluation

→ Once the chaser settles into the supply chain and undervaluation is resolved, **the path to double the current price (approx. 600k)**opens up.

+57% ~ +103%

When the yardstick changes — Bull~Super-Bull upside vs current price (SK Hynix/Samsung Electronics)

Why this take-profit is not fiction. All figures are based on actual target prices proposed by brokerages (Samsung 500k~550k, Hynix 3m~3.8m), earnings forecasts (Hynix 142t net profit), and facts such as the P/B→P/E yardstick shift and Jensen Huang's June 8 multi-year partnership. Even the Super-Bull does not deviate significantly from the top of the consensus. However, this path assumes that the transition of "memory being revalued as a contract industry" is maintained until 2027-2028.

Chapter 6

How to walk the path to doubling — Entry and realization

If the target is double, the problem is not "whether to buy," but "how to ride it and where to get off." There are a series of gateways that will create volatility in the next two weeks — June 10 US CPI, June 16-17 FOMC (dot plot), and June 24 Micron earnings. Instead of catching a falling knife all at once, buy in stages while checking these gateways.

Entry in stages at each gateway

Timing	SK Hynix	Samsung Electronics	Weight
After 6/10 CPI	1,910,000	296,000	20%
After 6/16-17 FOMC	1,820,000·1,800,000	280,000·270,000	+30%
After 6/24 Micron	Follow-up upon confirming good earnings	Follow-up upon confirming good earnings	+30~40%

The core holding is SK Hynix, which Jensen Huang personally guaranteed; the split strategy is for Samsung Electronics, which has just entered the supply chain.

Realization in stages at take-profit

SK Hynix: Realize 1/3 of holdings at Bull (3m), another 1/3 at Super-Bull (3.8m), and hold the rest as long as the transition is alive.
Samsung Electronics: 1/3 at Bull (500k), further realization at Super-Bull (600k).
Realize positive targets in splits, but do not sell everything — because the final rise of the transition can be the steepest.

Trend check line. Acceleration of buying when foreign net buying turns, exchange rate stabilizes at 1,520, and 60-day line support overlaps. The day the thinking changes is **When Samsung falls below its 60-day moving average at 268,000 and SK Hynix at 1,790,000, while the exchange rate re-breaks 1,560,**is when both occur simultaneously.

The Seventh Chapter

The signal that the story is ending — the day the cycle returns

The shift to a contract-based industry hasn't eliminated the cycle, but merely postponed it. Therefore, pre-defining the signals for the cycle’s return is the only way to walk the path to a double-bagger return to the very end. Memory has always peaked alongside claims that "this time is different," and stock prices have historically peaked before margins reached their best.

Signals that the transition is nearing its end (shift to defense if two or more occur):

Deceleration of quarterly gross margin — passing the peak of margins
DRAM spot prices decelerating compared to the previous quarter
The moment HBM/DRAM weeks of inventory exceed 8 weeks (currently 2~4 weeks)
Slowing of Long-Term Agreement (LTA) renewals — cracks in the contract industry logic
Full-scale mass production by China’s CXMT + new capacity volume hitting the market in 2027~2028

In the short term, if the June 10 CPI exceeds expectations and the FOMC remains hawkish, the exchange rate may re-break 1,560 KRW. In that case, delay new entries, but as long as the 60-day moving average holds, do not shake your current holdings.

Epilogue

The End of the Cycle, The Beginning of the Contract

June 8 was a day when fear dominated prices. However, that fear may have been the final expression of the old instinct that "memory is, after all, a cycle." On the same day, at the foundation of the industry, demand was shifting from replacement to infrastructure, transactions from bargaining to contracts, and metrics from assets to earnings. The greatest witness to this change said in Seoul, "I will buy for 5 more years."

If this transition is real — and it stands upon the facts of 70% data center demand, 5-year long-term contracts, and the brokerage industry's shift from PBR to PER — then the perspective on these two companies must fundamentally change. The path opens up to 3.8 million KRW for SK Hynix, double its current price, and 600,000 KRW for Samsung Electronics, also double its current price. Climb this path by dividing it at each gateway, realize gains at target prices, and only descend on the day the signal for the cycle's return sounds.

Of course, the cycle has not disappeared. It has only been postponed. However, during this postponed time — until 2027~2028, when inventories are normalized and contracts support earnings — the greatest revaluation will occur exactly where old beliefs collapse. Not selling in fear, betting on the transition, and watching for the signal of the cycle's return — that is how to approach these two stocks.

"The belief that memory is a cycle has suppressed prices.
Where that belief wavers, prices are being re-evaluated according to new metrics.
The cycle has not disappeared — it is just that the revaluation comes first."

Samsung Electronics · SK Hynix — The End of the Cycle, The Beginning of the Contract · June 8, 2026
Narrative constructed after synthesis of GOGUMA Team research and discussions · Closing 6/8 · Based on Jensen Huang meeting, brokerage target prices, and industry data
Bull/Ultra-Bull exit prices are based on brokerage target prices (Samsung 500k~550k, Hynix 3M~3.8M) · earnings outlooks · and the logic for the shift from PBR to PER
This report is for informational purposes only and is not a recommendation to buy or sell. The responsibility for investment lies with the investor.