Samsung SK hynix Earnings Face a Strong-Won Paradox Despite the Chip Boom

Samsung SK hynix earnings have encountered an unexpected constraint, despite record chip profits and continued demand from artificial intelligence infrastructure. A strengthening Korean won reduces the domestic value of revenue earned in dollars. That translation effect is now pulling analysts’ forecasts lower.

The change does not mean the memory boom has ended. High-bandwidth memory, server DRAM, and enterprise storage remain central to expanding AI data centers. However, currency movements can soften reported revenue and profit even when product prices rise.

That is the paradox confronting Korea’s two largest memory manufacturers. Better market conditions support their underlying businesses, while a stronger home currency makes those gains look smaller in financial statements.

The conflict matters beyond two quarterly earnings reports. Samsung Electronics and SK hynix carry enormous weight in South Korea’s stock market and export economy. Their results also serve as indicators for Nvidia suppliers, cloud infrastructure spending, and the wider memory cycle.

The most important contest is therefore not Samsung against SK hynix. It is operating momentum against currency translation. The next earnings season will reveal how much stronger memory pricing can offset the exchange-rate pressure.

A Stronger Won Has Changed the Earnings Calculation

The currency has not weakened chip demand, but it has changed how dollar sales translate into reported Korean earnings.

Memory chips are internationally traded products, and many contracts are priced in U.S. dollars. Samsung and SK hynix report their consolidated financial results in Korean won. When the won strengthens, each dollar of revenue converts into fewer won.

Consider a simplified sale worth $1 million. At an exchange rate of 1,500 won per dollar, that revenue converts into 1.5 billion won. At 1,400 won per dollar, the same sale becomes 1.4 billion won.

The customer bought the same quantity at the same dollar price. Yet the manufacturer reports 100 million won less revenue before considering hedging, costs, or other variables.

That mechanism has entered analysts’ semiconductor forecasts. Mirae Asset Securities reportedly reduced its dollar-won assumption from 1,553 won to 1,400 won. The change lowered its operating-profit estimates for both Korean chipmakers.

The firm cut its Samsung operating-profit forecast by 5.7 percent for 2026 and 4.2 percent for 2027. It reduced the corresponding SK hynix estimates by 4.8 percent and 5.6 percent, according to the published currency forecast.

Those revisions illustrate the scale of the accounting exposure. They do not represent a comparable fall in shipments or customer demand. They reflect fewer reported won from dollar-denominated business.

Currency exposure extends beyond sales translation. Semiconductor manufacturers import equipment, materials, software, and intellectual property from multiple countries. A stronger won can lower some dollar-based costs, partially offsetting the revenue pressure.

The timing of revenue and expenses also differs. Equipment purchases may enter cash flow immediately but affect earnings through depreciation over several years. Revenue translation can reach quarterly results much faster.

Hedging further complicates the picture. Companies can use financial contracts or matching currency expenses to reduce short-term volatility. However, hedging usually delays or limits the effect rather than removing long-term economic exposure.

The precise impact therefore depends on product mix, contract terms, costs, hedges, and the average exchange rate during each reporting period. A quarter-end spot rate alone cannot determine the result.

That distinction is essential for interpreting Samsung SK hynix earnings. A stronger won can reduce reported growth without signaling weaker competitiveness. Conversely, a favorable currency can flatter earnings without improving the product business.

The immediate change is still significant. Analysts now need stronger price increases, more shipments, or a better product mix to reach the same won-denominated profit forecast.

The AI Memory Boom Remains the Larger Force

Exchange rates are applying a discount to the boom, not erasing the demand that created it.

AI accelerators need enormous amounts of fast memory to train models and generate responses. High-bandwidth memory, or HBM, stacks multiple memory dies to feed processors more data while controlling energy use.

This requirement has moved memory closer to the center of AI system performance. A processor cannot use its theoretical computing capacity if data reaches it too slowly. Memory bandwidth can therefore become a system-level bottleneck.

SK hynix entered the current cycle with a strong position in HBM products. Samsung has worked to expand its advanced HBM business while benefiting from stronger prices across conventional DRAM and storage products.

The boom is broader than one premium memory category. AI servers also require standard server DRAM and enterprise solid-state drives. Supply allocated to those products can tighten availability elsewhere and support industry pricing.

Samsung’s second-quarter results showed how strongly that environment had already affected its business. The company reported 171.5 trillion won in consolidated revenue and 89.5 trillion won in operating profit.

Samsung’s semiconductor division generated 127.5 trillion won in revenue and 89.2 trillion won in operating profit during the quarter. Its memory business recorded quarterly highs for both measures.

The company said continued AI infrastructure investment supported demand for servers, HBM, enterprise SSDs, and server DRAM. It also expected supply constraints to persist despite some moderation in smartphones and personal computers.

SK hynix reported its own second-quarter results on July 29. The company’s official earnings release identified the quarter as another important performance milestone.

Independent coverage placed the two companies’ results inside the same AI-driven memory expansion. An earnings analysis described both companies as beneficiaries of stronger memory prices and advanced HBM shipments.

That underlying demand explains why analysts can lower profit forecasts while remaining optimistic about the companies. The new forecasts combine a less favorable currency with better expectations for memory prices.

Mirae Asset increased its assumptions for average selling prices, or ASPs, which measure the average revenue earned for each unit sold. The firm also raised its target prices for both stocks.

Its analysis projected Samsung’s DRAM ASP to rise 16.5 percent sequentially in the third quarter. It expected another 5.4 percent increase in the fourth quarter.

For SK hynix, the corresponding estimates were 15.8 percent and 7.2 percent. These are forecasts rather than completed results, but they show the offset analysts expect from pricing.

HBM contributes more than unit growth. It carries a higher value per bit and requires advanced packaging, stacking, testing, and close cooperation with accelerator customers. That can improve the product mix even when shipment growth moderates.

Next-generation HBM also increases capacity per accelerator. Mirae Asset said HBM use in custom accelerators from major technology companies should widen the customer base beyond traditional graphics processors.

That development would matter for both Korean manufacturers. Cloud companies are designing more proprietary AI chips to control cost, performance, and supply. Each design creates another opportunity for qualified memory suppliers.

However, qualification cycles are demanding. Memory must meet specific bandwidth, thermal, power, reliability, and packaging requirements. A broad market boom does not guarantee that every supplier wins an equal share of premium orders.

The chip cycle is therefore providing a substantial cushion, not immunity. Rising prices can overcome part of the currency impact. Strong HBM demand can improve margins. Neither force makes the exchange rate irrelevant.

Why Samsung SK hynix Earnings React Differently

The same currency movement reaches Samsung and SK hynix through different product portfolios, customer relationships, and cost structures.

SK hynix operates as a more concentrated memory company. That focus gives investors a relatively direct view of HBM, DRAM, and NAND conditions. It also leaves results closely tied to memory pricing and shipment volume.

Samsung Electronics contains a much wider collection of businesses. It sells memory, foundry capacity, smartphones, displays, appliances, network equipment, and other products. Each division carries a different currency and demand profile.

Samsung’s semiconductor unit currently provides most of the company’s earnings momentum. Yet weakness in mobile devices or consumer electronics can offset part of that strength.

The second-quarter figures demonstrated this contrast. Samsung’s mobile and networks businesses reported 33.2 trillion won in revenue and a 700 billion won operating loss. Its visual display and appliance operations also recorded a slight operating loss.

A stronger won can affect those businesses in different ways. Dollar revenue becomes less valuable when translated into won. Imported components can become cheaper, while regional pricing and local expenses create further offsets.

SK hynix presents a narrower calculation, but not a simple one. Its HBM mix, long-term customer agreements, equipment spending, and global manufacturing footprint all influence currency sensitivity.

HBM customer concentration deserves particular attention. A supplier with large dollar contracts can benefit tremendously from strong demand. It can also experience a visible translation impact when the won rises.

The two companies also occupy different positions in the HBM competition. SK hynix established an early lead in supplying advanced products for prominent AI accelerators. Samsung has been expanding its presence through HBM4, base dies, and broader semiconductor capabilities.

A base die sits beneath stacked HBM memory and manages connections between the memory layers and the processor package. Its design increasingly affects bandwidth, power efficiency, and customization.

Samsung can produce memory while also manufacturing logic chips through its foundry operation. That combination offers potential integration advantages for customized HBM. It also requires Samsung to execute across several technically difficult businesses.

SK hynix has emphasized memory specialization and customer alignment. Its position makes premium HBM demand a particularly important earnings driver. It also raises the consequences of qualification delays or changes in customer sourcing.

These differences explain why investors should avoid applying a single exchange-rate rule to both companies. A stronger won is a common headwind, but its final effect depends on each company’s business composition.

Their competitive relationship still matters. Samsung seeks to regain ground in premium AI memory, while SK hynix wants to protect its lead and diversify customers. Micron provides an additional global alternative for accelerator manufacturers.

However, that rivalry is supporting context for the current earnings issue. Currency translation is not evidence that one Korean manufacturer has taken business from the other.

The central question is how much operating momentum each company can produce before translation. Product pricing, HBM qualifications, yields, and shipments determine that momentum. Exchange rates determine how much appears in won.

This separation helps readers understand apparently conflicting headlines. One report can lower earnings forecasts because of the won. Another can raise a stock-price target because of HBM growth.

Both conclusions can be internally consistent. Earnings estimates measure expected financial outcomes under specific assumptions. Valuations also incorporate future growth, competitive position, cash generation, and the duration of the cycle.

The latest analyst revisions reflect that distinction. Mirae Asset raised Samsung’s target price from 370,000 won to 400,000 won. It increased its SK hynix target from 2.8 million won to 3.1 million won.

Target prices are opinions, not guarantees. Still, raising them while cutting profit estimates captures the paradox clearly. Currency lowered the near-term numbers, while memory fundamentals improved the longer-term assessment.

Strong Memory Pricing Cannot Remove Every Risk

The bullish case depends on price increases and HBM growth remaining stronger than currency pressure, capacity costs, and competitive responses.

Forecasts for Samsung SK hynix earnings currently rest on several moving assumptions. The exchange rate is only one of them. Memory prices, customer capital spending, manufacturing yields, and product qualifications can change quickly.

The first uncertainty is the won itself. Currency markets respond to interest rates, trade balances, capital flows, geopolitical events, and expectations about economic growth. A forecast of 1,400 won per dollar is not a fixed operating fact.

The average rate during the quarter matters more than one day’s closing level. A rapid movement late in the reporting period may have less impact than a sustained change across three months.

The second uncertainty is memory pricing. Analysts expect higher ASPs to offset much of the exchange-rate burden. That conclusion weakens if conventional DRAM or NAND prices rise more slowly than projected.

Memory remains a cyclical industry even when AI creates structural demand. Manufacturers eventually add capacity, customers adjust inventories, and high prices encourage alternative sourcing.

The current cycle differs from earlier smartphone and PC expansions because AI servers consume more valuable memory. Still, structural growth does not eliminate supply responses.

Samsung, SK hynix, and Micron are investing to meet demand. New fabrication facilities take years to construct, equip, qualify, and ramp. Once production arrives, it can alter the balance between supply and demand.

Capital spending also creates pressure before new capacity produces revenue. Manufacturers must fund buildings and equipment, then absorb depreciation as those assets enter service.

That spending is rational when customers need more memory. It becomes risky if cloud companies reduce AI infrastructure budgets or delay accelerator deployments.

Investors are already debating that possibility. Record profits do not automatically settle whether today’s investment levels will generate adequate returns throughout the next cycle.

HBM adds another layer of execution risk. Stacking more memory layers increases manufacturing complexity. Suppliers must maintain acceptable yields while meeting demanding thermal and performance targets.

A product can be technologically capable yet contribute less profit than expected if yields remain low. Qualification delays can also shift revenue between quarters without eliminating eventual demand.

Customer concentration raises the stakes. Nvidia remains a major center of the AI accelerator market, while cloud operators expand their own processors. Supplier diversification can reduce concentration, but each platform requires qualification work.

Mirae Asset’s positive case points to custom accelerators from companies including Meta, Microsoft, and OpenAI. Wider HBM adoption could enlarge the market and give suppliers more negotiating opportunities.

That outcome has not been fully secured by an industry forecast. Individual orders, shipment schedules, yields, and customer approvals will determine the financial benefit.

Samsung’s diversified structure presents separate risks. Its memory business can prosper while mobile or consumer divisions struggle with high component costs. The consolidated result may therefore lag the semiconductor division’s performance.

SK hynix carries greater exposure to a reversal in memory conditions. Its specialization strengthened the company during the HBM expansion, but concentration works in both directions.

Competition also remains active. Micron is expanding its HBM offerings, while Chinese memory manufacturers continue investing in conventional products. Increased competition could pressure prices at different levels of the market.

Export controls and trade restrictions can alter equipment access, customer demand, and regional supply. These policies affect manufacturers differently and can change with limited notice.

For those reasons, a stronger won should not become an all-purpose explanation for any earnings miss. Investors must separate translation pressure from weaker shipments, lower prices, higher costs, or execution problems.

The opposite warning also applies. A favorable product cycle should not lead readers to dismiss currency exposure. A five-percent earnings revision is meaningful even when the remaining forecast looks strong.

The sensible interpretation sits between those extremes. Currency is a measurable headwind inside a much larger semiconductor expansion. Its importance will vary with every quarter’s pricing and product mix.

The Currency Paradox Extends Across Korea’s Economy

Samsung and SK hynix show why a currency that helps consumers can simultaneously complicate the country’s most important export earnings.

A stronger currency can make imported energy, equipment, and materials less expensive in local terms. It can also improve the purchasing power of Korean households and companies buying foreign goods.

Exporters experience the other side. Dollar-denominated sales translate into fewer won, which can narrow reported margins when domestic costs do not fall at the same pace.

South Korea feels this tension acutely because semiconductors occupy an unusually large role in exports, industrial investment, and the stock market. Samsung and SK hynix are not isolated corporate cases.

The two companies together accounted for 36.9 percent of the Kospi’s weight in early February, according to a chip-cycle analysis. Their movements can therefore reshape index performance.

That concentration gives the exchange rate a second route into investor portfolios. It affects corporate earnings assumptions, then influences the valuations of companies that dominate the domestic benchmark.

The market may still reward those companies when product fundamentals improve. Samsung shares rose 5.7 percent and SK hynix advanced 8.1 percent during a September semiconductor rally, according to market coverage.

Such moves show that investors do not price currency exposure in isolation. AI demand, memory prices, interest rates, risk appetite, and capital-return policies all affect the stocks simultaneously.

The stronger won can even reflect confidence in Korea’s economic prospects or inflows into Korean assets. In that case, the same capital movement that supports the currency may accompany rising share prices.

This is another reason the situation looks paradoxical. The won can strengthen alongside optimism about Korean technology companies, then reduce the won value of those companies’ overseas revenue.

Currency translation also influences comparisons with foreign competitors. A Korean manufacturer reporting in won and a U.S. manufacturer reporting in dollars may show different growth rates even under similar market conditions.

Investors comparing Samsung, SK hynix, and Micron should therefore distinguish operational measures from reported currency outcomes. Shipment growth, ASPs, margins, and free cash flow each answer different questions.

For enterprise technology buyers, the implications are less direct but still relevant. Currency pressure alone does not imply lower global chip prices. International contracts can remain firm even while Korean reported earnings decline.

Buyers should focus on availability, qualification schedules, contract pricing, and delivery commitments. Those factors determine infrastructure costs more directly than the translation rate in a supplier’s consolidated accounts.

Developers and AI product teams should care because memory supply affects the cost and availability of computing capacity. HBM constraints can influence accelerator deployments, cloud pricing, and the speed of new AI services.

Knowledge workers will not feel a currency translation adjustment directly. They may feel its downstream effects through cloud budgets, hardware availability, and the pace of AI infrastructure expansion.

The wider lesson is simple. A national currency can improve one part of the economy while pressuring another. The result depends on who earns dollars, who spends them, and when each transaction reaches financial statements.

Three Signals Will Decide What Happens Next

The next earnings cycle must show whether product economics can stay ahead of currency translation.

The first signal is the average dollar-won exchange rate through the third quarter. A sustained level near the revised 1,400-won assumption would make the analysts’ lower forecasts more relevant.

A return toward the previous 1,553-won assumption would weaken the currency-headwind thesis. Further won appreciation would strengthen it and could prompt another round of estimate revisions.

Readers should not treat a single volatile trading session as decisive. The quarterly average, corporate hedging disclosures, and management commentary will provide a clearer picture.

The second signal is reported DRAM pricing and HBM product mix. Mirae Asset expects sequential DRAM ASP increases at both Samsung and SK hynix. Those gains form the main operating offset in its model.

If realized pricing meets or exceeds those forecasts, the memory cycle can absorb much of the currency pressure. If pricing disappoints, the stronger won becomes more damaging to the earnings outlook.

Product mix will matter as much as the headline ASP. Rising HBM shipments can lift the average even when conventional memory prices moderate.

Investors should look for HBM revenue growth, customer diversification, qualification progress, and manufacturing yields. These indicators reveal whether premium demand is converting into profitable volume.

The third signal is the companies’ third-quarter reporting, including operating profit and forward guidance. Samsung and SK hynix will have to explain the gap between operating performance and currency translation.

A modest forecast miss accompanied by strong shipments and pricing would support the current paradox thesis. The business would remain healthy while accounting translation reduced the reported result.

A miss caused by weaker orders, falling prices, or delayed qualifications would tell a different story. In that case, currency would be only one component of a broader slowdown.

Management commentary on capital spending will also matter. Continued expansion would signal confidence that AI memory demand remains durable. More cautious spending could indicate concern about future supply or customer budgets.

The strongest confirmation would combine three outcomes: sustained HBM growth, realized DRAM price increases, and results close to forecasts under the stronger-won assumption. That combination would show genuine operating resilience.

The thesis would weaken if the won stabilizes but earnings estimates continue falling. Such a pattern would point toward product, cost, or demand problems that currency translation cannot explain.

Samsung SK hynix earnings now sit at the intersection of two unusually strong forces. AI infrastructure is lifting memory demand, while the won is reducing the local value of overseas sales.

Neither force should be considered alone. The chip boom remains real, but reported profit is always the product of demand, pricing, costs, execution, and currency.

For readers following the AI hardware market, the next task is not predicting every daily exchange-rate move. Watch whether HBM growth and memory pricing keep outrunning the translation drag.

That comparison will reveal whether the stronger won merely softens a historic expansion or exposes a more vulnerable earnings structure.