BofA Raises Samsung Electronics Price Target to KRW 550,000, Forecasting KRW 50.6 Trillion Profit for 2027
Alina Collins
Bank of America raised its Samsung Electronics target price to KRW 550,000 and lifted 2026–2028 EPS forecasts by roughly 20% — driven not by higher volumes or pricing, but by a sharp downward revision to DRAM and NAND cost assumptions, raising the profit floor across the board.
Why did profits jump so suddenly?
Samsung's 2Q26 preliminary operating profit came in at roughly KRW 89.4 trillion, up about 56% quarter-on-quarter — well above BofA's prior mid-KRW 70 trillion estimate and the Street consensus of mid-KRW 80 trillion.
This means → BofA's earlier assumptions on new-node manufacturing, R&D, and SG&A costs were too conservative. The beat came from the cost side, not a demand surprise.
BofA now expects quarterly profit to reach KRW 113.1 trillion in 3Q26 and KRW 121.4 trillion in 4Q26 — crossing the KRW 100 trillion mark for the first time.
What does an 80% DRAM margin really mean?
BofA's most aggressive call: even after special bonuses, 2Q26 DRAM margins likely approached 80%. The second half stays above 80%; the 2027 full-year figure is 79.8%.
In plain terms = for every $100 of DRAM sold, Samsung keeps nearly $80 as profit — an extraordinarily rare level in semiconductors.
The structural driver: server DDR5 and HBM — high-bandwidth memory, the fast-access storage built for AI chips — are crowding out advanced wafer capacity. Margins come from structural supply tightness, not volume expansion.
DRAM ASP assumptions: up 20% QoQ in 3Q, 8% in 4Q, then another 24% for full-year 2027.
What about NAND and valuation?
NAND ASPs surged 65% QoQ in 2Q26, with another 15% in 3Q. The 2026 margin is 63.3%, easing to 56.9% in 2027.
Demand is anchored to enterprise SSDs — high-capacity storage for data centres — and Samsung is not spreading capex evenly across all product lines.
The KRW 550,000 target is cross-checked three ways: roughly 10× average 2027–2028 EPS, a sum-of-the-parts valuation, and a DCF — all converge.
This reflects a valuation identity that is now overwhelmingly memory: DRAM and NAND together account for roughly 91% of segment fair value.
What holds up the 2028 "soft landing"?
BofA projects 2028 DRAM and NAND ASPs to decline roughly 10% and 12% year-on-year, yet EPS falls only about 4% — still roughly 26% above 2026.
The key assumption: new capacity is staggered. The P4 fab will mainly handle 1Cnm HBM and high-end DRAM; the P5 shell is expected to complete in 1H28, with meaningful new capacity arriving in late 2028 to 2029.
HBM revenue path: $26.9 billion in 2026 → $48.5 billion in 2027 → $62.7 billion in 2028. HBM4 — the fourth-generation high-bandwidth memory — rises from 32.4% of HBM shipments to 63.6% in 2027.
What could break the soft landing?
The falsification sequence is explicit: spot prices weaken → contract-price gains narrow → customer inventories rise → P4/P5 capacity ramps faster than planned.
This means → if those four signals appear in succession, BofA's 2028 assumptions of 76.4% DRAM margins and 52.5% NAND margins both need to come down.
In plain terms = this model is not "as long as HBM demand holds, we're fine." It rests on a precise supply-pacing assumption — once that rhythm breaks, the margin forecast cannot stand.
Are the non-memory businesses dragging on the group?
2Q26 memory operating profit was roughly KRW 91.1 trillion — actually higher than the group total of KRW 89.4 trillion. This means → the other divisions are collectively losing money.
Foundry and system LSI lost roughly KRW 2.1 trillion; mobile lost about KRW 360 billion; consumer electronics and displays were near breakeven or slightly negative.
Memory price increases cut both ways: the semiconductor division's external sales surge, but the mobile, TV, and appliance divisions absorb higher memory procurement costs. Mobile profit for 2026 is forecast at just KRW 1.2 trillion, down sharply from KRW 12.9 trillion in 2025.
Foundry losses are projected at KRW 6.6 trillion in 2026, KRW 3.7 trillion in 2027, reaching breakeven only in 2028. If US hyperscalers cut capex, HBM share disappoints, or new capacity arrives too fast, the case for KRW 550,000 at today's low multiple will become a renewed point of market debate.
Content is for reference only, not financial advice.