In the 2026 HBM rally, the chipmakers and their suppliers have traced two very different price paths. SK hynix is up 230.3% year to date and Samsung Electronics 168.5%. Soulbrain (솔브레인, 357780.KQ, listed on the KOSDAQ), which supplies both companies with etchants and CMP slurry, is up 68.5% — six times the KOSDAQ index’s 11.2% over the same period, yet only about 30% of SK hynix’s move and about 40% of Samsung’s.
The gap is not the market overlooking materials, nor simple multiple suppression. It comes down to earnings elasticity: the HBM repricing amplified chipmaker profits while Soulbrain’s FY2025 profit actually fell — priced by usage versus priced by scarcity, showing up as two profit lines. Wet chemicals are settled by consumption and capture none of HBM’s ASP elasticity. The company’s most HBM-levered product line — HBM-dedicated CMP slurry — went from reported sole supply to dual supply within roughly five months. And the battery electrolyte business keeps pressing on company-wide margins through the EV downcycle. Financially, FY2025 delivered revenue growth without profit growth (revenue up 7.0%, operating profit down 20.5%), and then Q1 2026 produced record quarterly revenue with operating profit up 24.1% — a reversal signal and a structural problem sitting on the table at the same time.
Throughout, this piece uses the operating company Soulbrain (357780), not the holding company Soulbrain Holdings (036830).
From the start of 2026 to the June 12 close, Soulbrain rose from 262,000 won to 441,500 won, up 68.5%, for a market value of about 3.43 trillion won. Over the same window, the KOSDAQ index gained 11.2%, SK hynix 230.3%, and Samsung Electronics 168.5%.
The path carries more information than the endpoint. The stock made its year low of 277,000 won on January 12 and its year high of 530,000 won on January 30 — up roughly 91% in under three weeks, a stretch that coincided with a cluster of January sell-side reports raising 2026 earnings expectations. Then came more than four months of give-back and consolidation. On June 12 the stock jumped 24.4% in a single day, approaching the daily price limit, and closed at 441,500 won — still about 17% below the January high.
Fund flows produced a crossover worth recording on the spike day. On June 12, foreign investors net-sold 185,094 shares, roughly 2.4% of shares outstanding, taking the foreign ownership ratio from 26.28% the prior day down to 23.90%; institutions net-bought 266,651 shares the same day. Over the prior 20 sessions, foreigners had net-bought roughly 49,000 shares in aggregate while institutions net-sold roughly 40,000, with the foreign ratio climbing from 23.54% on May 14; on June 9, Fidelity Management & Research filed a large-holding report with DART related to a stake above 5%. Foreign money spent a month building the position, then monetized part of it into the surge. That sequence is itself a statement about what a materials stock is worth.
On valuation, 441,500 won works out to roughly 40.9 times trailing earnings on FY2025 net income of 84.0 billion won; on FY2024 net income of 119.6 billion won, the multiple is 28.7 times. The market is paying as if FY2025 earnings were a crater that 2026 will fill: sell-side reports (Shinhan Investment, Daishin Securities, and others, via media re-reporting) put 2026 at revenue of 1.089 to 1.10 trillion won (up 17 to 19%) and operating profit of 192.0 to 204.0 billion won (up 44 to 53%). One caveat needs stating: this piece did not pull comparable multiples for SK hynix or Samsung, and Soulbrain’s own 40.9 times cannot, by itself, support a claim that the market suppresses the materials layer’s valuation. The more direct explanation for the year-to-date gap sits in earnings: the HBM repricing amplified chipmaker profits while Soulbrain’s FY2025 net income fell. The three mechanisms in the next section unpack where that earnings-elasticity gap comes from.
Same supply chain, chipmakers up 230%, the core materials vendor up 68.5% — the difference is mechanism. In Korea’s Memory Industry: One Country, Two Cycles we discussed how HBM repriced the chipmakers from cyclical commodities into qualified-supply assets: customers locking volume, price, and roadmaps. The problem is that none of that repricing automatically passes through to the materials layer.
First, materials are priced by volume and have no ASP elasticity. Etchants, CMP slurry, and precursors are chemicals consumed per wafer start and per process step: chips got repriced on dollars per gigabyte, while chemicals are still settled on usage per step. HBM does raise per-unit consumption — more stacked layers, more etch and polishing steps — so materials vendors do capture volume growth. But the leg where chip ASPs double never shows up on a chemical price sheet. Public disclosure also shows no HBM-style multi-year volume-and-price lock-ins at the materials layer. The power structure is even written into the shareholder register: Samsung Electronics took a 4.8% stake in Soulbrain in November 2017 — roughly 20 months before Japan’s 2019 export controls — for 55.6 billion won (carried through the 2020 corporate split). A customer binding its supplier before the controls says this was a long-term supply-chain strategy; the 2019 controls amplified and accelerated it rather than created it. The structure stands: the customer leads, and localization remains a customer-led de-risking project, not a supplier pricing-power project.
Second, the most HBM-levered product line kept its sole-supply window for only about five months. Seoul Economic Daily reported exclusively on October 10, 2023 that Soulbrain solely supplied HBM-dedicated CMP slurry — a specialty material used to remove the copper layer in the HBM process — to both Samsung and SK. But according to The Elec (March 2024), Dongjin Semichem began supplying HBM CMP slurry to SK hynix from the first quarter of 2024. From reported sole supply to a competitor’s reported entry: about five months — and since qualification typically runs ahead of press coverage, the true exclusive window may have been shorter still. The multi-sourcing strategy chipmakers learned from their AI customers got copied straight down to the materials layer, and executed faster there. The window when HBM volume ramped hardest was exactly the window when the exclusive position got diluted.
Third, electrolytes are dragging at the other end. Soulbrain is not only a semiconductor materials company. Semiconductor materials are estimated at roughly 70% of revenue (the official segment split in DART filings is unconfirmed; this is an estimate), with the rest in display materials and secondary battery materials — electrolytes and lead tabs, including the subsidiary SunFluo System. Both struggled in 2025: display materials fell sharply year over year, and per Shinhan Investment’s segment estimates for the Q3 2025 reporting period (via re-reporting), semiconductor materials grew 20.7% year over year while display fell 51.9% and battery and others rose just 5.0%. Battery electrolytes are a low-margin business that offers little growth through the EV demand air pocket while pressing down the blended company margin; on top of that, the company booked impairments on investments in associates for a second consecutive year — 44.8 billion won in FY2024, another 29.5 billion won in FY2025. On April 21 the company also disclosed a decision to acquire shares in a subsidiary; the filing details remain unread, and whether that money goes toward batteries or semiconductors is worth watching.
Lay out four years of consolidated results (DART basis). FY2022: revenue of 1.02 trillion won, operating profit of 207.1 billion won, a 20.2% operating margin. FY2023: down to 844.0 billion won, 133.5 billion won, and 15.8%. FY2024: recovery to 863.4 billion won, 167.9 billion won, and 19.5%. FY2025: revenue of 923.4 billion won (up 7.0%), but operating profit down to 133.6 billion won (down 20.5%) and a 14.5% margin — the lowest of the four years.
The immediate cause of profitless growth sits in expenses: selling, general and administrative costs rose from 78.4 billion won in FY2024 to 99.0 billion won, up 26%. Net income fell harder, down 29.8% to 84.0 billion won — this is where the two consecutive years of associate impairments noted above land.
Then came Q1 2026 (quarterly report filed May 15): revenue of 263.8 billion won, up 25.9% year over year and a quarterly record; operating profit of 44.7 billion won, up 24.1%. Bulls read that as the first bar of the turn, reasonably. But one detail should not be skipped: the Q1 operating margin was 16.9%, actually down 0.3 percentage points year over year, and SG&A rose from 21.9 billion won to 27.7 billion won — still growing at around 26%, the same pace as FY2025. Daishin Securities’ view (via re-reporting) is that the FY2025 margin decline came from one-off costs rather than structural problems. The impairments may indeed be one-off, but expense growth showed no deceleration through Q1. The reversal, so far, exists in absolute profit, not in margins. Sell-side consensus puts the 2026 operating margin around 18.5% — above Q1’s 16.9%, still below FY2022’s 20.2%. The market is not even betting on a return to the prior peak.
The last time Soulbrain earned narrative-grade pricing was Japan’s 2019 export controls on Korea. Before the controls, the company mass-produced only 5N (99.999%) grade hydrogen fluoride. In January 2020, BusinessKorea reported that it had succeeded in mass-producing 12N (99.9999999999%, twelve nines) high-purity HF. After expansion, its capacity could cover roughly two-thirds of Korea’s semiconductor demand for high-purity HF, and its domestic share of high-purity HF-based etchants is estimated above 85% (a Samsung Securities estimate, via re-reporting).
In 2026 that history cuts both ways. The positive: the moat is real — no second domestic producer mass-produces at the 12N grade, and that revenue carries high certainty. The negative: the story is over. Above an 85% share, import substitution has no incremental option left to tell, and the valuation will not carry that line again. More subtly, the direction has begun to reverse: as Korea-Japan trade relations normalized after 2023, the Japanese high-purity HF that was once locked out could flow back. The localization premium has turned from an incremental story into an incumbent position that needs defending. That risk shows up in no financial line item today, but it explains why the market no longer pays for the chokepoint narrative.
Having argued the cautious side, the bull ledger deserves a fair accounting — 68.5% and 40.9 times trailing earnings are prices paid with real money.
The bull case runs roughly four points. One, volume growth is real and accelerating: the HBM3/HBM3E transition raises per-unit consumption of etchants and slurry, and Samsung’s 4nm/3nm GAA transition adds etch steps (noted in a Samsung Securities report, via re-reporting) — a structural rise in process step counts means materials revenue can persistently outgrow wafer starts. Two, 12N HF has no substitute supplier. Three, Q1 2026’s quarterly record is delivered results, not expectations, and the sell side projects 2026 operating profit up 44 to 53%. Four, volume-based pricing cuts both ways: you miss the ASP upside, but you also dodge the ASP downside — in the FY2023 memory downturn, revenue fell about 18% and the operating margin still held at 15.8%, far steadier than the chipmakers. In an industry that eventually mean-reverts, bulls would say, low beta is not a flaw.
Note the shape of the bull case: all four points are about volume and certainty, none about pricing power. Which confirms this piece’s conclusion from the other side — the best case for the materials layer is capturing volume growth in full, and half the rally is the structural ceiling. The gap between 68.5% and 230% is less the market deliberately marking down the materials layer than two earnings elasticities tracing their own paths: chips were repriced on scarcity and profits amplified; chemicals settle on usage and profits follow usage. One more explanation belongs on the table, and it is about flows rather than fundamentals: the foreign and institutional money chasing HBM concentrates in KOSPI large caps, and a KOSDAQ name with a market value around 3.4 trillion won cannot absorb flows of that order — part of the rally gap is fund-flow structure, not fundamental pricing. This remains the least glamorous layer of the supply chain, and the one whose revenue follows every wafer.
The official segment revenue figures in DART filings were not pulled this time; the segment structure cited here relies on sell-side estimates, and those estimates carry a labeling ambiguity in their period basis (the three segment figures sum closer to a single quarter than to a nine-month cumulative), so the text uses year-over-year direction only and drops the absolute amounts. The revenue contribution of HBM-dedicated CMP slurry inside Soulbrain is not publicly disclosed — which means the financial impact of the sole-to-dual supply shift, a key mechanism in this piece, cannot be quantified. The specific trigger for the 24.4% single-day move on June 12 is unconfirmed. The complete 52-week high-low range (including the second half of 2025) was not pulled. The revenue split between Samsung and SK hynix is, per DART convention, undisclosed. The foreign ownership series covers only the last 20 trading days.
This article should therefore be read as an observation about pricing structure — the materials layer structurally captures less of the chip layer’s windfall — not as a complete judgment on the company’s fundamentals. The mechanism is clear; the data to quantify it is not yet complete.
This report is an independent KSINQ market observation for informational purposes only. It is not investment advice. Data snapshot: June 13, 2026; prices as of the June 12, 2026 close.