South Korea's July memory export data provides the first set of leading indicators for Samsung and SK Hynix's third-quarter HBM business.
Bernstein has long tracked South Korea's multi-chip memory exports to Taiwan and Malaysia. The report believes these exports have a strong correlation with the quarterly HBM revenue of Samsung and SK Hynix: Taiwan is a key location for CoWoS advanced packaging, while Malaysia hosts Intel's EMIB packaging facilities.
In July, South Korea's multi-chip memory export value to Taiwan and Malaysia fell 32% from June's historical high. However, the report primarily attributes this change to intra-quarter seasonality. Compared to the first month of the previous quarter, April, July's export value still grew 13% and was up 64% year-over-year, indicating overall HBM demand has not significantly weakened.
While overall volume remains strong, the export trends of the two South Korean memory makers have diverged: Samsung's related exports continue to rise, while SK Hynix's related data is significantly below Bernstein's expectations.
Samsung's Q3 HBM Revenue Could Exceed Expectations by 30%
Bernstein uses exports from Chungcheongnam-do as a proxy indicator for Samsung's HBM shipments, as Samsung's related back-end production and packaging facilities are primarily located in this region.
In July, Chungcheongnam-do's multi-chip memory export value to Taiwan and Malaysia reached $2.2 billion. Although this represents a 35% decline from June's seasonal high, it is up 122% compared to April. Samsung's related export value has also exceeded SK Hynix's for two consecutive months.
Based on the regression relationship between historical export data and HBM revenue, Bernstein calculates that Samsung's Q3 HBM revenue may reach approximately $12 billion, representing roughly 80% quarter-on-quarter growth and about 30% higher than its original forecast of around $9.3 billion.
If only recent data since Q1 2025 is used for regression, the forecast value further rises to $12.6 billion, corresponding to approximately 90% quarter-on-quarter growth.

South Korea's monthly multi-chip memory export value (billions USD) to Taiwan and Malaysia and year-over-year growth rate. Although July export value declined month-over-month due to seasonal factors, it still grew 64% year-over-year, indicating strong overall HBM demand.
However, this figure remains a model estimate, not company guidance. Samsung's exports are typically concentrated in the latter two months of a quarter. Since 2025, the first month of a quarter has on average accounted for less than 20% of the quarter's total export volume. Therefore, whether Q3 revenue ultimately reaches the model's forecast will depend on actual shipments in August and September.
Bernstein believes the July data is at least consistent with the direction Samsung previously indicated: the company expects Q3 HBM4 sales to grow more than threefold quarter-on-quarter, and for HBM4 to account for over 60% of HBM sales in the second half of 2026.
Unit Value Doubles, HBM4 Becomes Samsung's Growth Driver
In addition to export volume, the unit value of Samsung's related product exports has also risen significantly.


Left: Chungcheongnam-do (location of Samsung HBM packaging) multi-chip memory export value (millions USD) to Taiwan and Malaysia compared to Samsung's quarterly average HBM revenue. July export value reached $2.2 billion, up 122% from April, exceeding SK Hynix for the second consecutive month. Right: Regression analysis based on historical data shows July export data suggests Samsung's Q3 2026 HBM revenue could reach $12 billion, up ~80% QoQ, ~30% higher than Bernstein's original forecast.
In July, the unit value of multi-chip memory exports from Chungcheongnam-do increased 21% month-over-month, more than doubling from April levels and nearing four times that of SK Hynix's related regions. Given HBM4's higher capacity, stack layers, and technical complexity, Bernstein believes this change likely reflects the rapid increase in the proportion of Samsung's HBM4 shipments.
This does not mean the price of Samsung's same-spec HBM doubled within a few months. Export unit value is also influenced by product mix, capacity specifications, and packaging combinations, and can only serve as a directional indicator for average selling prices. A more reasonable explanation is: Samsung is replacing some HBM3 and HBM3E with higher-priced HBM4, thereby lifting overall export value.
This is also key to Samsung potentially regaining market share. Previously, SK Hynix held a leading position with HBM3E and closer supply relationships with Nvidia; as the cycle moves into HBM4, the competitive focus is shifting to who can complete verification earlier, improve yields, and achieve volume delivery.
SK Hynix Exports Weaken, But Single-Month Data Doesn't Directly Equate to Share Reversal
In contrast to Samsung, SK Hynix's related exports weakened noticeably in July.
Bernstein uses exports from Chungcheongbuk-do and Icheon as proxy indicators for SK Hynix's HBM shipments. In July, export value from these two regions fell 28% from June and also declined about 27% compared to April.
The base regression model calculates that based on this, SK Hynix's Q3 HBM revenue might only be $5.6 billion, down about 20% quarter-on-quarter, 55% lower than Bernstein's original forecast. If this materializes, Samsung's Q3 HBM revenue would significantly overtake SK Hynix's.
However, the uncertainty of this prediction is much higher than for Samsung. The report also points out that if using SK Hynix's historical seasonal pattern where shipments are most concentrated in the latter part of the quarter, its Q3 HBM revenue could still reach $12 billion, close to Bernstein's original forecast.
In other words, the same set of July data, under different seasonal assumptions, could correspond to a vast range from $5.6 billion to $12 billion. At this stage, a more prudent judgment is not that "SK Hynix has already lost its lead," but that its Q3 start is weaker than expected, and its ability to catch up in August and September will determine the final share.
Bernstein speculates the weakness may stem from delays in Rubin-related HBM4 shipments. SK Hynix previously stated that HBM4 began mass production in Q2 and will ramp up fully in the second half of the year. However, attributing the July export decline directly to Nvidia's Rubin progress remains an analyst inference based on industry cadence and has not been explicitly confirmed by the related companies.
HBM Not Following Traditional DRAM Price Hikes Synchronously
The report also notes that HBM and traditional memory price trends are diverging.
Since Q3 2025, traditional memory prices have risen approximately fivefold cumulatively, but the export unit value of SK Hynix's related products has remained generally flat. Samsung's unit value roughly doubled, also primarily due to the increased proportion of HBM4, rather than significant simultaneous price hikes for existing products like HBM3E.
This means HBM's contract pricing, supply relationships, and product iteration cycles are partially decoupling it from traditional DRAM spot price fluctuations. Short-term price increases in traditional DRAM do not necessarily transmit proportionally to HBM, whose profit changes rely more on the proportion of new-generation products and annual customer contracts.
Bernstein believes suppliers and customers have already begun negotiating 2027 HBM contracts, and expected price increases are likely to drive further upward revisions to next year's profit forecasts for Samsung and SK Hynix. Compared to share changes in a single quarter like Q3, this may be a more important pricing variable for the memory sector in the next stage.
Malaysia Export Surge Remains an Unresolved Variable in the Supply Chain
In July, South Korea's multi-chip memory export value to Malaysia increased 20% month-over-month to approximately $1.3 billion, continuing the rapid growth of the past few quarters. Samsung still accounts for most of this, but SK Hynix's exports also increased noticeably.
Bernstein speculates these HBMs may be destined for Intel's EMIB advanced packaging facilities in Malaysia, for use in server chips equipped with HBM. However, the report also acknowledges it is currently difficult to explain why related exports are growing so early, well before product mass production.
Therefore, exports to Malaysia are better suited as a subsequent supply chain clue rather than being directly equated to Intel customer orders or specific product ramp-up.
Overall, the July data strengthens the logic of Samsung catching up in the HBM4 cycle, but is not yet sufficient to confirm a long-term market share reversal has occurred. What truly needs observation next is whether Samsung can sustain export growth in August and September, and whether SK Hynix can recover the shipment gap from the start of Q3 as HBM4 ramps up.
Share Reversal Still Awaits August and September Data
The July export data strengthens the logic of Samsung accelerating its catch-up in the HBM4 cycle: related export volume grew significantly, unit value rose rapidly, and it corroborates the company's HBM4 sales guidance.
However, this data is still insufficient to confirm Samsung has regained HBM leadership on a full-year basis.
First, Bernstein uses a regression model between regional exports and company revenue, not actual corporate order data. Second, if HBM packaging moves outside South Korea, or some products undergo further packaging within South Korea, customs data may not fully capture the activity. Finally, single-month exports can also be affected by customer acceptance, shipping schedules, and quarterly seasonality.
Three variables need to be observed next: whether Samsung's August and September exports can continue growing, whether SK Hynix's exports and unit value recover as HBM4 ramps, and whether the two companies' 2027 HBM contract prices can truly be raised.
If Samsung's exports continue to grow and SK Hynix fails to catch up, HBM4 could drive more substantive changes in market share; if SK Hynix concentrates deliveries in the latter two months of the quarter, July's divergence is more likely just a timing mismatch in shipments.
Therefore, what the current data truly changes is not that the HBM competition already has a result, but that the speed at which Samsung is re-entering the share race may be faster than the market previously expected.






