Samsung Electronics Q2 Operating Profit Reaches KRW 89.4 Trillion as HBM4 Mass Production Begins
N.R. Finch
Samsung Electronics posted KRW 89.4 trillion in Q2 operating profit — exceeding last year's full-year total in a single quarter. But everything outside memory is still losing money, and whether HBM4 can actually close the gap with SK Hynix is the real question the market is pricing.
A single quarter bigger than all of last year — where did the money come from?
Quarterly revenue hit roughly KRW 171 trillion, with operating profit at about KRW 89.4 trillion — one quarter surpassing the prior year's full-year profit.
Bank of America estimates the memory division alone earned roughly KRW 91.1 trillion in operating profit. This means → foundry, mobile, and TV/appliance businesses combined are still in the red. Memory is carrying the entire company.
In plain terms = Samsung's profit structure is radically lopsided. Memory earnings are subsidizing losses everywhere else.
Profit forecasts doubled in six months — so why didn't the price target follow?
In January, HSBC and JPMorgan projected Samsung's full-year 2026 operating profit at KRW 169–173 trillion. By July, Citi and Bank of America had raised that to KRW 381–401 trillion — more than doubling in six months.
Yet valuation multiples actually contracted: Citi's memory-division multiple fell from 7.9× to 7.6×; Bank of America's corresponding P/E dropped from 11× to 10×.
This means → the analyst target prices of KRW 530,000–550,000 are driven purely by earnings delivery, not by multiple expansion. This reflects lingering doubt about whether Samsung's profits can last.
HBM4 is in mass production — but does "shipping" equal "leading"?
Samsung confirmed HBM4 mass-production sales for Nvidia's Vera Rubin platform, but disclosed no data on customer qualification status, market share, or product-level margins.
In plain terms = there is a long road between "started selling" and "selling well." Qualification progress, yield rates, and shipment share — none of the hard metrics are public yet.
HBM4 progress pulls on three business lines simultaneously: HBM itself, 4 nm base dies — the chips underneath each HBM stack — and advanced packaging. Internal orders do not prove external competitiveness.
The foundry is still losing money — how long can the vertical-integration story hold?
Samsung's foundry business remains a loss-making segment. The company's narrative: "Our own foundry makes base dies for our own HBM, creating a vertical-integration advantage."
This means → if HBM revenue scale, yield, or the 1c process node — Samsung's next-generation DRAM manufacturing step — stalls at any point, the synergy logic could reverse into a cycle of high capex and internal cost pressure.
What still needs proving, item by item: HBM revenue scale, customer breadth, yield, 1c process capability, and foundry external-client wins. Miss any one, and the story falls apart.
Can this profit surge survive a full cycle?
Bank of America projects 2027 operating profit at KRW 506.4 trillion and free cash flow at KRW 335.5 trillion — but its 2028 scenario already assumes average DRAM and NAND prices decline 10–12%.
In plain terms = the current profit explosion is fueled by strong conventional DRAM and NAND pricing, but prices always mean-revert. A blockbuster quarter does not guarantee the next cycle holds up.
This reflects where the real stress test lies: not today, but after 2027. Whether profits can endure a memory-price downturn is the precondition for Samsung's valuation multiples to expand again.
Content is for reference only, not financial advice.