Samsung, SK Hynix Q3 Profit Forecast Nears 190 Trillion Won. The Real Story Isn't HBM Alone

Analyst estimates put combined Samsung Electronics and SK hynix Q3 operating profit near 189.9 trillion won. The bigger shift is that AI demand is now tightening supply and lifting prices across server DRAM and NAND, not just HBM.

First, what that 189.9 trillion won figure actually means

As of September 21, financial data provider FnGuide's consensus put Samsung Electronics' third-quarter operating profit at 111.7 trillion won and SK hynix's at 78.1 trillion won. Combined, that is 189.9 trillion won.

This is a forecast, not a reported result. The estimate has actually come down from just days earlier, with a stronger won cited as the main reason. In other words, the memory boom underneath these numbers is genuinely strong, but what stock prices are pricing in isn't "current profit." It's how long this elevated level of profit can be sustained.

Samsung Electronics Q3 operating profit consensus: 111.7 trillion won (FnGuide, as of September 21) SK hynix Q3 operating profit consensus: 78.1 trillion won (FnGuide, as of September 21) Combined: 189.9 trillion won — still an unconfirmed market estimate

This supercycle started with HBM, then spread to ordinary memory

HBM stacks multiple DRAM dies vertically to deliver very high bandwidth to GPUs and AI accelerators. The catch is that HBM consumes more wafer capacity and equipment time per chip, and its back-end packaging is more complex. When manufacturers prioritize capacity for high-margin HBM and server products, supply of general-purpose DRAM for PCs and smartphones naturally tightens.

TrendForce projected third-quarter server DRAM contract prices would rise 13% to 18% from the prior quarter, with overall NAND flash contract prices up 10% to 15%. Global DRAM industry revenue jumped 59.5% quarter over quarter in the second quarter. AI demand is simultaneously expanding revenue from high-value products and squeezing supply of everything else.

A more recent fourth-quarter outlook from TrendForce, dated September 24, shows the trend hasn't reversed. TrendForce expects cloud service providers' AI demand to keep supporting both DRAM and NAND prices, with growing demand for KV cache in AI inference pushing enterprise SSD contract prices up 23% to 28% and overall NAND up 15% to 20%. Based on the data available now, it's too early to call the third quarter the peak of this memory profit cycle.

The key isn't rising HBM prices by themselves. It's the reallocation of production capacity. The more AI demand pulls in HBM and server DRAM, the more memory makers shift limited wafer capacity toward those high-value products. That tightens supply of general-purpose DRAM too. This is what separates the current cycle from past memory booms driven mainly by PC and smartphone replacement demand.

Same boom, different playbooks for Samsung and SK hynix

SK hynix remains the top player in HBM. According to Counterpoint Research, in the second quarter of 2026, HBM revenue share stood at SK hynix 50%, Samsung Electronics 33%, and Micron 18%. What matters is that Samsung's share jumped from 21% just one quarter earlier to 33%.

The picture looks different across all of DRAM. In TrendForce's tally, Samsung led second-quarter DRAM revenue with $60.98 billion, a 39.4% share, while SK hynix posted $38.59 billion and 24.9%. Samsung is benefiting from both expanding HBM4 shipments and rising general-purpose DRAM prices at the same time. SK hynix, with a higher share of revenue coming from HBM, is more sensitive to swings in the AI memory cycle.

Samsung ElectronicsSK hynix
Core leverageMassive general-purpose DRAM/NAND production base + recovering HBM4 shareLeading position in HBM + supply track record with AI accelerator customers
Recent signal33% HBM revenue share in Q2 2026, 39.4% overall DRAM revenue share50% HBM revenue share in Q2 2026, still the industry leader
Favorable scenarioHBM4 expansion and strong general-purpose memory pricing continue togetherHBM shipments and high-value product mix keep expanding
Key riskHBM yield/customer qualification issues, foundry losses, general-purpose memory prices peakingHBM share erosion, high earnings sensitivity if AI capex slows

CXMT is less a threat to HBM than a ceiling on ordinary memory

China's CXMT isn't shaking up the top ranks of HBM right now. Its more immediate impact is on general-purpose DRAM. CXMT announced on September 20 that it has moved into mass production of a fifth-generation DRAM platform, disclosing process geometry around 11.95 nanometers and an expanded lineup of LPDDR5X products.

This matters because the more Samsung and SK hynix shift capacity toward HBM, the more of a supply gap opens up in general-purpose DRAM, and Chinese suppliers can fill part of that gap. So a long memory supercycle doesn't mean every product category keeps the same pricing power. HBM has high barriers to entry, while Chinese supply could cap how far general-purpose DRAM prices can rise.

The first brake is already visible: currency

Memory makers generate most of their revenue in dollars. When the Korean won strengthens, the same dollar revenue translates into smaller reported profit in won terms. In a September 17 tally, Samsung Electronics' and SK hynix's third-quarter operating profit consensus had already fallen 7.5% and 6.5%, respectively, from a month earlier.

Nomura estimated that a 10% appreciation in the won could cut domestic memory makers' operating profit by roughly 12%. SK hynix's own half-year report disclosed a sensitivity showing that, all else equal, a 10% decline in the won-dollar exchange rate would reduce pre-tax income by about 4.75 trillion won. Even if memory prices keep climbing, a currency move in the opposite direction can flatten the slope of won-denominated profit growth.

Insight Times Editorial Desk

More by topic

Starting with this story's topics, then what readers are following now

Investor psychology, decision-making and books on investing

HBM

This story's topic77 storiesSee all →

Memory stacked next to the GPU.

Memory Is Still Short, but Less Short: Checking for the Peak This Week

Citi trimmed its 2027 DRAM and NAND shortage forecasts together. Server pricing remains strong, while consumer markets show weakness. This week's numbers show where the memory boom may peak.

Investing

The Real ASML Number Isn't the 3% Pop, It's the 30% Capacity Hike

ASML's rally isn't just an AI trade. Customers are already locking in 2027 tool slots, and ASML is raising Low-NA EUV capacity by roughly 30% to keep up.

Investing

A 3x P/E is not a discount. It is a verdict on how long memory profits last

Nomura's buy case on Samsung Electronics and SK Hynix does not stop at higher memory prices. It argues that if AI has turned memory from a cyclical component into strategic infrastructure, the formula the market uses to value both companies has to change too.

Investing

2027 Isn't the End: Citi's Case for a Memory Supercycle Running to 2031

Citi says the memory shortage timeline is stretching again. If AI shifts from "train once" models to continual learning, HBM, server DRAM and enterprise SSDs could all fall short at the same time.

Tech

AI ran out of memory before it ran out of GPUs

The HBM bottleneck is spreading into DRAM prices, server costs and the price of the next laptop or phone.

Tech

Has Memory Peaked? Prices Are Cooling, but the Shortage Looks Set to Run to 2028

Price growth has slowed and consumer demand is already breaking. Server and HBM supply remains short, and Micron says more than 75% of its 2027 volume is already committed to customers.

Tech

AI Chip Peak Check: Demand Still Rising, Risk Building in 2027-2029 Supply and Funding

DRAM and NAND contract prices are still climbing and HBM suppliers hold the edge. Power shortages, AI financing costs and simultaneous memory expansion are the risks for later years.

Investing

A building that houses servers and networking gear and supplies them with powe…

Nvidia and Palantir Put AI Inside the Supply Chain - The Real Shift Is That It's Learning Judgment

A single Vera Rubin rack holds 1.3 million parts. The real breakthrough in Nvidia and Palantir's new supply chain AI isn't automating that complexity, it's capturing the judgment calls that used to live only in planners' heads.

Tech

Is Space Cooler for AI? Why Musk Wants Data Centers in Orbit

The next limit on AI may be power and cooling, not GPUs. Elon Musk is looking to Earth orbit for an answer, but the test is cost per hour of compute.

Tech

Musk Sees the Next Phase of AI Infrastructure: Who Can Switch GPUs On, Fast

If neoclouds struggle under financial pressure, AI demand may not shrink. It may simply favor the few operators with the capital and execution to serve it, a shift that could matter for SpaceX and Tesla.

Tech

White House Swaps AI for "SI." The Real Investment Signal Is Infrastructure, Not the Name

An executive order renames artificial intelligence "Super Intelligence" and six frontier labs sign a voluntary safety accord. What investors should watch is the data center, power, nuclear and chip policy now gathered under one national-competitiveness frame.

Tech

The Cheaper Tokens Get, the Bigger Your AI Bill Gets

As AI shifts from training to inference, the token is becoming an industrial commodity. Falling prices are not cutting demand, because agents and automation consume tokens at a scale no human chat session ever did.

Investing

Marvell's $115 Billion Lumentum Deal: Is AI's Next Bottleneck Light, Not Chips?

Marvell's purchase of Lumentum points to a contest beyond GPU speed, in data transfer and power efficiency. Whether the price is right is not yet confirmed.

Tech

Workers in hard hats beside a communications equipment cabin, with a Starship rocket on its launch tower at dusk in the background

SpaceX Forecast at About $730 Billion in 2028 Revenue: The Bigger Opportunity Musk Sees

Sequoia Capital made a bold forecast, and Elon Musk replied, "I think we can do better." The future SpaceX is looking at is far bigger than the rocket launch business. The plan is to link satellite communications, AI data centers and orbital computing into one giant business.

TechEXPLAIN

NVDA

96 storiesSee all →

Who Hits $1 Trillion Next: Reading AMD and ASML by Different Rules

One company is chasing Nvidia's GPU empire. The other sells the machine that draws the chips for every advanced chipmaker, including Nvidia. Both get floated as 2028 trillion-dollar candidates, but the numbers investors need to watch are entirely different.

Investing

GPU

69 storiesSee all →

A chip built to draw graphics, whose ability to run thousands of identical cal…

The Next AI Chip Race Is About How Chips Connect

Applied Materials will lift its stake in Dutch equipment maker BESI from about 9% to 20%. The bet is less about control than about a new way to link chips.

Tech

Musk Borrowed $40 Billion for GPUs. He Also Warned AI Cloud Firms Could Go Bust

SpaceX's giant financing shows a new dividing line in AI: who borrows matters as much as how much is spent. Firms with other revenue may outlast those that rent out GPUs alone.

Tech

US AI Data Centers Are Headed for 90GW. Gigawatts Now Matter More Than Dollars

Goldman Sachs expects US data center capacity to grow from 64GW at the end of 2026 to 90GW at the end of 2027. The AI race is moving from buying GPUs to building power plants and transmission lines alongside them.

Tech

The AI Bubble Question Has Changed: By Late 2026, It's ROIC Over GPUs

There is no sign yet that AI investment is slowing. If anything, hyperscaler spending forecasts keep climbing. But as the scale of that capital has grown, the market's question has shifted from "is AI growing" to "when, and how much, does this capital come back as cash."

Tech

2027 Investing Needs Alignment, Not Prediction

A new investing book argues the hard part was never picking the next winner. It's staying aligned with where value and bottlenecks are moving while keeping old human weaknesses in check.

Investor Mind

Why Nebius Is Keeping Some 2027 AI Capacity Off the Shelf

Nebius mixes long-term contracts, short-term premiums and auctions. Holding back capacity pays only if the servers stay busy, and it is not yet clear the pricing will last.

Tech

Amazon Goes After Nvidia's Strongest Moat, and It Is Software, Not GPUs

Amazon's open-source TorchNeuron aims to let existing AI code run on its Trainium chips. The contest comes down to the total cost of AI compute and the return on Amazon's $220 billion in annual capital spending.

Tech