Two broker forecasts agree that Samsung Electronics is having an extraordinary 2026. Their 2027 DRAM assumptions tell very different stories about what comes next.
Samsung Electronics’ 2026 memory boom is not the controversial part. The harder question is what happens after it.
Two recent Korean broker reports are strikingly close on Samsung’s 2026 operating profit, yet sharply divided on 2027. That gap is useful because it isolates the core debate around the memory cycle: whether the extraordinary pricing environment of 2026 fades as supply expands and demand elasticity weakens, or whether HBM and higher-value memory can keep both pricing and volume rising for another year.
The disagreement begins with DRAM pricing
BNK Investment & Securities forecasts Samsung’s DRAM average selling price to fall 8% year on year in 2027 after a 274% increase in 2026. Mirae Asset Securities takes the opposite view, forecasting another 22.6% increase in 2027 after a 277.1% rise in 2026.
The 2026 numbers are unusual and heavily influenced by a sharp comparison base. The more important signal is the 2027 direction. BNK treats 2026 as a peak-like pricing year that gives way to normalization. Mirae Asset assumes the cycle remains strong enough for pricing to keep rising.
Because the two firms use different product-mix and forecasting assumptions, their ASP numbers are not perfectly apples-to-apples. But they still frame the central question clearly: does the market absorb new supply without breaking pricing?
Similar 2026 profit, very different 2027 outcome
The earnings forecasts show how much that one assumption matters. BNK forecasts Samsung’s operating profit at about $269.4 billion in 2026 and $280.6 billion in 2027. Mirae Asset is at about $266.6 billion for 2026, but jumps to $385.9 billion in 2027.
That means the two forecasts are almost aligned on 2026, then diverge by roughly $105.3 billion in 2027. The spread is not a probability range; it is simply the difference between two separate broker models. Still, it illustrates how quickly the earnings picture changes when the memory-price path changes.
BNK’s concern is demand elasticity meeting new supply
BNK argues that system-level memory costs are reaching levels that weaken demand elasticity. Its report says this pressure is no longer limited to traditional IT products and is increasingly relevant in servers as well. At the same time, memory producers are still adding capacity based on a more optimistic view of long-term demand.
That combination produces a progressively weaker 2027 price path in BNK’s model. It forecasts Samsung’s DRAM ASP to fall 1% sequentially in the first quarter of 2027, then 8% in the second quarter, 15% in the third and 30% in the fourth.
This is a broker scenario, not company guidance. But the shape of the forecast matters: the downside case is not based on an abrupt collapse in AI demand. It is based on the idea that prices eventually rise enough to slow incremental demand just as supply becomes less constrained.
Mirae Asset assumes HBM keeps the mix stronger for longer
Mirae Asset’s model is built on a more supportive combination of volume and pricing. It expects DRAM bit growth of 25.3% in 2026 and 21.0% in 2027 while keeping DRAM ASP growth positive in both years.
That framework implicitly gives more weight to HBM and other higher-value memory categories absorbing capacity and supporting pricing. In that case, Samsung’s 2026 earnings surge is not simply a cyclical spike. It becomes the first leg of a longer profit expansion.
2027 is where the thesis gets tested
The useful point is not that one broker must be right and the other wrong. It is that 2027 exposes the variable that matters most for Samsung’s earnings after the 2026 surge: the balance between memory supply growth and the ability of AI-driven demand, especially HBM demand, to keep absorbing it at favorable prices.
If demand remains mix-rich and supply stays disciplined, the more bullish earnings path becomes easier to sustain. If demand elasticity weakens while capacity additions accelerate, the 2026 boom could still be followed by a much flatter 2027. The selected reports do not provide company guidance for either outcome, so the split should be read as a scenario map rather than a forecast consensus.
