082920 — Vitzrocell Co., Ltd.
🔴 Live ₩29,800 (yf:082920.KS) · 2026-09-05 · research written 2026-06-29 your re-engage <₩27,000–₩30,000 (own-DD, on-pullback) — now in zone (-1%)
Thesis
First, the thing everyone gets wrong. Vitzrocell is not a supercapacitor company. It is the Korean leader in lithium thionyl-chloride (Li/SOCl2) primary batteries, with two fast-growing specialty legs in oil-and-gas downhole batteries and defense thermal/ampoule cells. The electric double-layer capacitor (EDLC) line that put it on the AI-datacenter watchlist sits inside a roughly 5%-of-revenue "EDLC plus new products plus other" bucket. Pure EDLC is low single digits. So before anything else: if you buy this for the supercap story, you are buying a low-single-digit slice of the business and a free option on a product that is not yet shipping.
What actually 3x'd the stock in 2026 (from about 20,000 in January to a 64,100 close on 11 May, now 36,950) is a real but discussion-stage catalyst. On 23 June 2026 CEO Jang Seung-guk confirmed Vitzrocell is in talks with Taiwan's Delta Electronics to supply lithium-ion capacitors (LIC, the energy-denser cousin of EDLC) for US Big Tech AI datacenters, with a decision "expected within 2026." His words, translated: when the contract closes, "the company's scale will fundamentally change." That is a documented, credible catalyst from a credible counterparty. It is also zero booked datacenter revenue, no signed contract, a single customer, no disclosed capacity, and a LIC product the company's own website still lists as "being prepared."
Here is the honest read, and it splits in two. As a business, Vitzrocell is genuinely good and getting better. Revenue has compounded about 21% a year over FY21–25, operating income about 41%, operating margin has risen roughly 1,300bps to 28.5%, net cash sits near 137B KRW (about 8% of market cap), and free-cash conversion is high. As an AI-datacenter supercap play, it is the third and least-proven Korean name on the theme, behind LS Materials (Vertiv MOU, Sept 2024) and Vinatech (Bloom Energy, May 2025), and its current supercap product is the EDLC, which is structurally the less suitable device for the datacenter buffering job than the LIC/HSC that Musashi sells. The datacenter-relevant product is the one Vitzrocell hasn't built yet.
So the cleaner way to own this is as a high-quality specialty-battery compounder carrying a cheap embedded option on AI power, not as a supercap pure-play. The trouble is the price already reflects the option: at about 29x trailing earnings and 5x book after a 3x run (now about 42% off the May high), the market is paying compounder-plus-option, not compounder-alone.
Verdict: PASS at 36,950 / WATCH on a deeper pullback. Conviction Medium on the business, Low on the AI-datacenter supercap angle as the reason to own. The business would justify a position at a more forgiving multiple; the supercap re-rating is the part to be skeptical of, not the part to chase.
Snapshot
Vitzrocell (KOSDAQ 082920), founded 1987, headquartered in South Korea, is a specialty primary-battery and electrochemical-storage maker. It sits in the Vitzro group under listed parent Vitzrotech (042370). Cross-linked to ai-infrastructure as a supercap/AI-power peer to 7220 Musashi, but its sector home is specialty batteries and defense-tech, not AI infrastructure. The supercapacitor technology grounding for this page lives in the supplied supercap primer (EDLC vs LIC physics).
Live market data as of the 2026-06-26 close (last print in yfinance; the 2026-06-29 session is not yet captured, and the Collyer Bridge chat noted Vitzrocell "up 5%" intraday on 29 June, so today's print is likely a touch higher, unconfirmed):
| Metric | Value |
|---|---|
| Last price | 36,950 KRW (2026-06-26) |
| Market cap | ~1.66T KRW ≈ US$1.08B |
| Enterprise value | ~1.52T KRW ≈ US$0.99B (net cash ~137B) |
| Shares outstanding | ~44.93M (end-2025); 45.34M issued (Jan-2026) |
| Trailing P/E | ~29x (FY25 EPS 1,267) |
| P/B | ~5.0x (FY25 equity 330.5B) |
| 52-week range (close) | 12,725 – 64,100 |
| 3-mo ADTV | ~743k sh / ~36.1B KRW (~US$23M/day) |
| Foreign ownership | 33.1% (Jan 2026) |
| Dividend yield | ~0.7% (FY25 DPS ~250–260, see data note) |
FX used throughout: ~1,543 KRW/USD (yfinance KRW=X close, 2026-06-29).
Business
Four legs, and the smallest one is the famous one
Vitzrocell makes batteries that have to work for ten or twenty years, or in places no rechargeable cell survives. That is the through-line across the portfolio. The revenue splits four ways, and the proportions matter enormously to the thesis because they tell you what you are actually buying.
| Segment | FY25 9M | FY24 FY | What it is |
|---|---|---|---|
| Li/SOCl2 primary cells | ~60% | ~70% | Non-rechargeable lithium cells: smart meters (gas/water/electricity, 10yr+ life), IoT, GPS/tracking, telecom |
| High-temperature batteries | ~20% | ~17% | Oil-and-gas downhole tools (>150°C MWD/LWD drilling) |
| Ampoule / thermal (defense) | ~15% | ~8% | Guided missiles, smart munitions, fuzes, torpedoes |
| EDLC / new products / other | ~5% | ~5% | Supercapacitors (EDLC) plus LIC development plus misc |
The two reads differ because defense and oil-and-gas grew fastest in 2025 and compressed the Li/SOCl2 share, but both pin the EDLC-plus-other bucket at about 5%. The decisive point for any supercap comparison is sitting right there in the table. The supercap line is, at most, 5% of revenue, and it shares a bucket with other new products, so pure EDLC is lower still. Anyone framing Vitzrocell as "a supercapacitor play" is buying the future LIC option, not the present revenue.
The chemistry, from first principles, and why a primary cell can be a great business
A primary battery is one you use once and discard. That sounds like a worse business than a rechargeable, and for most applications it is. But lithium thionyl-chloride is the exception that runs the table on a specific job: extremely long, low-current life. The cell pairs a lithium-metal anode with a liquid cathode of thionyl chloride (SOCl2), which is both the electrolyte solvent and the active cathode material. That construction gives the highest energy density of any practical primary chemistry (about 3.6V nominal, very flat discharge), and crucially a self-discharge rate low enough that the cell keeps useful charge for well over a decade.
Why that matters commercially: a gas, water, or electricity smart meter has to phone home for ten to twenty years on a battery nobody will ever change, often sealed behind a wall or underground. Replace it and you have sent a truck. The battery is a few dollars; the truck roll is the cost. So the meter OEM does not shop on price, it shops on whether the cell will still be alive at year fifteen, and it qualifies a supplier once and stays put. That dynamic gives Li/SOCl2 incumbents real switching costs and pricing power that the commodity-cell narrative misses. Vitzrocell claims roughly 90% domestic share in Korean Li primary cells and exports to 40-plus countries. That is a quietly excellent franchise.
The same long-life, harsh-environment logic extends to the other two legs. High-temperature batteries power measurement-while-drilling and logging-while-drilling tools two miles down an oil well, where ambient temperature exceeds 150°C and a conventional cell simply dies. Defense ampoule and thermal batteries sit inert for years inside a missile or a fuze and activate in milliseconds on launch. None of these are price-shopped commodities. All three are qualified-in, spec-locked, and tied to customers who care far more about reliability than about saving a few cents.
The supercap line, told straight
Now the EDLC. An electric double-layer capacitor stores charge electrostatically on the surface of activated-carbon electrodes, with no bulk chemical reaction. That buys it enormous power, half a million to a million cycles, and survival from −40°C to +85°C, at the cost of low energy density and high self-discharge. Vitzrocell's published EDLC line runs the VS (2.7V standard), VH (2.7V, to ~85°C), and VV (3.0V high-voltage) cell series plus module families, and the cells are cylindrical. Capacitance ranges, ESR, and form-factor dimensions are not published on the pages accessible for this work, so I will not attribute a specific farad range to Vitzrocell.
The EDLC's natural homes are exactly where small energy, high power, long cycle life, and wide temperature matter: IoT and wireless-sensor burst current, smart-meter "last-gasp" outage reporting and memory/clock backup, automotive stop-start and black-box hold-up, grid and renewables ride-through, industrial UPS and actuators, and defense. This is a broad, cost-driven, temperature-tolerant franchise tied to small-energy backup. It is a good fit for what Vitzrocell already sells into.
It is not, however, the right device for the AI-datacenter job, and this is the part the re-rating glosses over. The datacenter buffering task wants energy per litre (rack space is scarce and revenue-bearing), higher cell voltage (fewer cells in series), and the no-thermal-runaway safety of a hybrid cell, while the EDLC's one clear edge, extreme cold tolerance, is worthless in a climate-controlled hall. That set of constraints favors the lithium-ion capacitor (LIC), which is an asymmetric design with one activated-carbon electrode and one pre-lithiated faradaic carbon electrode. The pre-lithiation pins the negative electrode low and lifts the cell to roughly 2.2–3.8V. Energy scales with the square of voltage, so the 2.7V-to-3.8V jump is about a 2x gain on its own; the pre-lithiated faradaic electrode adds further capacitance, and the two together give the LIC roughly 3–5x the EDLC's energy in the same volume. That is the device Musashi sells as its Hybrid SuperCapacitor and the device the Flex/Musashi rack tray uses. It is also the device Vitzrocell is developing from its EDLC base and has not commercialized. Management targets LIC commercialization around 2026 "to fit AI and high-performance power-infrastructure demand," and the company website lists the LIC product as "being prepared."
So the precise statement is this. Vitzrocell's shipping supercap product (EDLC) is the wrong tool for the datacenter rack, and its right tool (LIC) is in preparation, behind Musashi's commercialized lead. The bull case is that Vitzrocell's mature cylindrical-cell manufacturing is an advantage: a subscriber technical thread (Collyer Bridge, 19 June) noted that LIC anodes swell during cycling because of the pre-lithiated lithium, and a cylindrical can distributes that swelling pressure more evenly than a prismatic one, so Vitzrocell's cylindrical-EDLC heritage could translate into a high-reliability cylindrical LIC. That is plausible and worth watching, but it is an argument about a product that does not yet exist at volume.
Where it sits in the chain, and who buys
Vitzrocell is a cell and pack maker. Upstream it buys lithium metal, thionyl chloride, activated carbon, and separators; none of those suppliers surfaced as a disclosed bottleneck, and the chemistry is mature. Downstream it sells to smart-meter OEMs, oil-and-gas drilling-tool makers, and defense primes (led by the Chunmoo guided multiple-rocket and smart-munitions supply chain, with ampoule/thermal exports rising to Türkiye, Israel, and India). The one named prospective customer on the supercap side is Delta Electronics, the world's largest power-supply systems integrator, for LIC into US AI datacenters. Specific named accounts beyond Delta were not disclosed in the sources read, which is normal for a defense-and-industrial supplier but means customer-concentration cannot be quantified precisely. The defense leg's growth from 1.6B KRW (FY21) to 35.9B (FY24) tells you that customer set is real and scaling.
Innova, the tariff hedge nobody is talking about
In October 2025 Vitzrocell closed the acquisition of Innova Power Solutions (Canada), a high-temperature battery-pack specialist, for about 33.6B KRW. It adds 20B-plus KRW/year of high-temp revenue and, more interestingly, a North American manufacturing base that hedges US tariff risk on the oil-and-gas leg. This is a sensible, on-strategy bolt-on into the highest-margin part of the business, not a diversification away from it.
Financials
The Core Four read
Strip the business to its four drivers and it scores well on three of them. Organic growth is strong and broad-based, about 21% revenue CAGR FY21–25 to 243.08B KRW, driven by the genuine demand pull in defense and oil-and-gas rather than by one product. Margins are expanding hard: operating margin climbed from 15.4% (FY21) to 28.5% (FY25), roughly 1,300bps in four years, as the mix shifted toward high-margin defense and high-temp cells. Capital intensity is modest and the model is cash-generative, with capex of 4–11B KRW a year, operating cash flow of 63.6B in FY25, and free cash flow about 52B. Capital deployment is the only soft spot, and only because the company is almost too conservative: it sits on 137B of net cash earning little, which drags reported returns (more on that below).
| KRW | FY22 | FY23 | FY24 | FY25 | Q1'26 |
|---|---|---|---|---|---|
| Revenue | 140.98B | 176.22B | 210.78B | 243.08B | 68.27B (+26.6% YoY) |
| Operating income | 28.83B | 37.69B | 51.91B | 69.27B | 20.28B (+34.5% YoY) |
| Op margin | 20.4% | 21.4% | 24.6% | 28.5% | 29.7% |
| Net income | 23.11B | 36.05B | 51.26B | 56.92B | 20.23B |
| EPS (real)* | ~510 | ~810 | ~1,130 | 1,267 | — |
*EPS computed on ~45M shares. Do not use yfinance's pre-2025 per-share figures (see data note).
One subtlety that flatters the headline and then un-flatters the growth. FY25 operating income grew 33% but net income only 11%, because FY24 pretax was lifted about 12.6B by one-off non-operating/financial income that did not repeat in FY25. So the FY24 net-income base was artificially high, and the underlying FY25 operating momentum is stronger than the 11% net-income growth implies. Net margin still landed at 23.4%.
The balance sheet is a fortress, and that is half a problem
Cash plus short-term financial assets ended 2025 at 140.96B KRW against total debt of just 3.62B, leaving net cash of about 137B (roughly US$89M, about 8% of market cap). One caveat on that figure: yfinance cannot independently confirm it, because its end-2025 "cash and cash equivalents" line reads just 24.375B; the ~137B net-cash position, and the resulting ~US$0.99B enterprise value, lean on roughly 116B of short-term financial assets that come from the Korean filings, not from yfinance. The 3.62B total debt does reconcile across both. Total equity is 330.5B; the company is effectively debt-free. That fortress is comforting and it funds the LIC build and the Innova deal without dilution. But it also depresses returns: operating ROIC (NOPAT over operating invested capital, stripping the cash) is excellent at roughly 28%, while return on equity is dragged down to about 17% by the idle cash pile. A more assertive capital-allocation posture (a buyback near the lows, or a larger dividend) would lift ROE without touching the operating business. That management hasn't done so is a minor knock, not a red flag.
Valuation in one breath
| Metric | Value (as of 2026-06-26) |
|---|---|
| Trailing P/E | ~29x |
| Forward P/E | ~24x (rough estimate, FY26 NI ~70B; labeled estimate, no clean consensus pulled) |
| P/B | ~5.0x |
| EV/sales | ~6.3x |
| EV/EBIT | ~22x |
| P/FCF | ~32x |
| FCF yield | ~3.1% (on market cap) |
| Net cash | ~137B KRW (~8% of cap) |
These are growth-company multiples on a business growing 20%-plus with rising margins, so they are not absurd on the fundamentals alone. They are rich once you remember that roughly nothing in current revenue is datacenter LIC, which is the part the re-rating is supposedly paying for.
Data quality warning (read before trusting any per-share figure)
yfinance will mislead you here in two ways, and the first is a ticker-suffix trap. Use 082920.KQ (KOSDAQ), which returns clean equity data (live price, market cap, shares). 082920.KS is a fund proxy that yfinance mislabels as a mutual fund, with a null market cap and a stale regularMarketPrice (~19,180) and stale 50/200-day MAs far below the live price; the earlier 06-26 pull leaned on that suffix's price history series, which happened to be right, but its info is unusable. Second, both suffixes carry a spurious "2.00 stock split dated 2025-12-10" that did not happen: the raw price series is continuous across that date (9 Dec 19,250, 10 Dec 17,820, a ~7% move, not a 50% gap), and no split is corroborated by Korean sources; share count is genuinely about 45M. Because of that phantom split the pre-Dec-2025 per-share history is halved (FY24 EPS shown as 575 vs a real ~1,130; FY24 dividend 230 vs a reported 460), so use the income statement, not the per-share fields. Dividend-per-share history remains murky across sources (FY24 cited at both 460 and 230; FY25 at 250–260) and possible bonus-issue mechanics are unresolved, so treat the ~0.7% yield as approximate. Update (2026-07-01): live 082920.KQ prints ~40,550, so the 36,950 / 2026-06-26 figures above are the dated basis of this 29 June analysis; the ~+10% move lifts the multiples to roughly 32x trailing and 5.5x book and does not change the verdict.
Industry landscape
This page links up to ai-infrastructure and defense-tech; the device-physics grounding is in the supplied supercapacitor primer. The short version of the sector thesis, because it is the reason the stock moved.
AI training racks swing power violently. NVIDIA documents rack utilization moving from about 30% idle to 100% and back in milliseconds as thousands of GPUs step in lockstep, which threatens voltage and frequency stability across the facility. The fix is a multi-timescale buffer that acts as a low-pass filter between chaotic GPU demand and a smooth grid draw, with ceramic and silicon caps at the die in nanoseconds, bulk caps at the board in microseconds, supercapacitors and lithium-ion capacitors at the rack in milliseconds to seconds, and battery storage at the facility for the slower ramps. The supercap/LIC layer is becoming mainstream specifically with the new 800V DC (or ±400V DC) rack architecture. Note the precise framing: the supercap buffer does not sit at 800V; the Flex/Musashi rack tray operates on the ~48V rail inside the 800V-fed rack. Do not write "800V supercap."
SemiAnalysis cross-check. SA's "AI Training Load Fluctuations at Gigawatt-scale" (Jeremie Eliahou Ontiveros, 25 June 2025) is the authoritative primary piece on this exact theme. It walks the power-quality problem from first principles and surveys the solutions from supercapacitors to UPS to BESS, and it explicitly identifies rack-level supercaps and li-ion caps as the right tool for millisecond peak-shaving, "on the verge to become mainstream with the rise of the new 800V DC architecture." SA validates the use case in full. It does not name Vitzrocell (or Musashi) as a winner, and it leans toward BESS for the longer second-to-minute fluctuations while reserving supercaps/LIC for the sub-second job. No contradiction with this write-up's conclusion; if anything SA reinforces the caution, because the structural demand is real but the specific equity winners are unsettled, and Vitzrocell is not among SA's identified beneficiaries.
Management
Who controls this, and who is selling the story
| Holder | Stake | Note |
|---|---|---|
| Vitzrotech (042370) | 34.22% (Q3'25) | Listed parent; ~37.8% with related parties |
| Chairman Jang Soon-sang | 41.40% of Vitzrotech | Top of the control chain |
| Treasury | ~1.8% | |
| Foreign | 33.1% | High for a KOSDAQ small-cap |
The control chain runs from Chairman Jang Soon-sang, through Vitzrotech, into Vitzrocell, with the family controlling roughly 38% of Vitzrocell. CEO Jang Seung-guk is the public face of the Delta/AI-datacenter narrative and holds a personal stake reported around 12.6B KRW. The shared surname and the structure strongly imply a family-succession arrangement between the Chairman and the CEO, but that relationship was not explicitly confirmed in the sources read, so I flag it as likely rather than asserted.
This is a founder/family-controlled group, which cuts both ways. The upside is patient ownership, the conservative balance sheet, and skin in the game. The consideration for a minority investor is the usual Korean small-cap governance setup, with a controlling parent, a related-party structure, and a free float around 60% where the family's interests set the agenda. Nothing in the sources points to asset-shuffling, shell-entity dealings, or related-party self-dealing, but the diligence here is limited by source access (no proxy-level related-party reading was possible), so treat the governance grade as "no red flags found, not a clean forensic bill of health." Capital allocation is conservative-good, marked by a sensible bolt-on (Innova), modest disciplined capex, a token dividend, and a large cash buffer that funds growth without dilution but sits idle.
Catalysts & risks
What could change the story
The near-term catalyst is binary and dated: a Delta LIC decision "expected within 2026." A signed contract with disclosed volume would be a genuine re-rating event and would validate the CEO's "scale will fundamentally change" framing; continued silence or a pass would deflate the supercap premium and leave the stock to trade on the (still good) battery business. Around that sit slower, more reliable drivers: LIC commercialization from the EDLC base targeted around 2026; the defense tailwind from rising global defense budgets and the K-defense export surge; the oil-and-gas tailwind from US shale drilling lifting high-temp demand; and the Innova integration adding North American high-temp revenue. The defense and oil-and-gas legs are the catalysts I would actually underwrite, because they are already showing up in the numbers. The Delta option is the one the market is paying for and the one least within the company's control.
Risks
| Risk | Likelihood | Mitigant / can it close? |
|---|---|---|
| Delta talks stall or yield no contract | Medium-High | The base business stands on its own; closes only when a contract is signed and disclosed. The supercap premium is at risk regardless. |
| LIC product is late or under-performs on energy density | Medium | Mature cylindrical-cell manufacturing helps; but the product is "in preparation" and qualification hinges on energy density, where it trails Musashi's shipping HSC. |
| Already priced in (valuation) | High | At ~29x trailing and 5x book after a 3x run, the option is in the price. Only closes via a deeper de-rate or earnings catching up to the multiple. |
| Trails LS Materials and Vinatech on datacenter traction | Medium-High | Real today; Vitzrocell is third on the theme. Its cash engine is better, but on pure datacenter traction it is behind. |
| Customer concentration on the supercap option | High (for that leg) | Single named customer (Delta). Structural to a discussion-stage relationship. |
| Korean small-cap governance / family control (~38%) | Low-Medium | No red flags found, but limited forensic visibility. Structural, not closeable. |
| Liquidity is real but thin for blocks | Medium | ADTV ~US$23M/day, up ~5x in 2026; fine for normal size, impactful for institutional blocks. |
| Defense export / geopolitical exposure (Türkiye, Israel, India) | Low-Medium | Diversified across customers and countries; structural. |
Bear case. Strip the AI option entirely and you have a very good specialty-battery business worth perhaps 18–22x earnings, which against ~29x trailing implies the supercap premium is roughly a quarter to a third of the current price. If Delta passes, the LIC slips, and the defense/oil cycle merely normalizes, the stock can de-rate to the high-20,000s on no change to the durable business. The thesis "break" is not a business failure, it is the option expiring worthless while you paid for it.
Valuation / DCF
No full DCF here (Register D keeps the model in a separate sheet). The framing is comparative, because the cleanest way to value Vitzrocell is against the name it is constantly compared to.
Vitzrocell versus Musashi, the comparison that actually matters
| Vitzrocell (082920) | Musashi (7220) | |
|---|---|---|
| Market cap | ~US$1.08B | ~US$1.7B (¥261.5B) |
| Supercap product | EDLC (shipping); LIC in preparation | LIC/HSC (shipping, Flex CESS design-win) |
| Supercap % of revenue | ~5% (EDLC+other) | <1–2% (not separately disclosed) |
| The other 95–99% | Specialty batteries, ~28% op margin, ~21% growth | Honda-tied auto driveline, ~5% EBIT margin, flat, BEV-exposed |
| Trailing / forward P/E | ~29x / ~24x est | 207x distorted / ~22x fwd |
| Price action 2026 | +220% then −42% off May high | +270% off Apr base to ¥10,550 peak (3 Jun), then −62% |
| Datacenter traction | Delta talks (discussion-stage) | Flex CESS, Mitsubishi Electric, UL certs (cell ramp delayed; ~200K cells 2026 vs 6.5M nameplate; no hyperscaler win named) |
This table holds the whole argument. On the supercap leg, Musashi is ahead, shipping the energy-denser LIC, with a named integrator design-win (Flex CESS) and UL-certified product, while Vitzrocell is in talks with one customer for a product it hasn't commercialized. But on the business underneath the option, Vitzrocell wins decisively. Its other 95% is a 28%-margin, 21%-growth specialty-battery franchise; Musashi's other 99% is a flat, 5%-margin, 50%-Honda-concentrated, BEV-exposed auto-parts shell. So with Vitzrocell you pay roughly compounder-plus-option; with Musashi you pay roughly auto-shell-plus-expensive-option that the tape already round-tripped. That is exactly why the source community has rotated.
Put plainly: Vitzrocell is the better business to own and the worse pure datacenter-supercap bet. If the only thing you want is exposure to the AI-rack-buffer theme, Vitzrocell is not even the most-exposed Korean name (LS Materials and Vinatech are further along). If you want a high-quality compounder that happens to carry a free option on that theme, Vitzrocell is a reasonable hold once the price gives the option back for free, which at 36,950 it does not quite do.
What is the business worth on its own? At ~21% revenue and ~30%-plus operating-income growth with a fortress balance sheet, the durable battery franchise supports something like 20–24x earnings without the option, call it the high-20,000s to mid-30,000s KRW depending on how much credit you give the defense/oil cycle. The option is worth whatever you think the probability-weighted Delta/LIC outcome is, which is genuinely uncertain and which the market is currently capitalizing fairly fully. The disconnect is not that the business is mispriced; it is that the option is priced as if it were closer to a contract than it is.
Decision log
2026-06-29 – Initial deep-dive (new idea). PASS at 36,950 / WATCH on pullback. Conviction Medium (business) / Low (AI-DC supercap angle). Vitzrocell is a high-quality Li/SOCl2 primary-battery compounder with strong defense and oil-and-gas legs (FY25 rev 243.08B, +15% YoY; op margin 28.5%; net cash ~137B; FCF ~52B), not a supercapacitor company (EDLC is ~5% of revenue, pure EDLC lower). The 2026 3x run was driven by a discussion-stage Delta LIC catalyst for AI datacenters, confirmed by CEO Jang Seung-guk on 23 June 2026, decision "within 2026," with zero booked datacenter revenue and the LIC product still "in preparation." Trails LS Materials and Vinatech on actual datacenter traction; its shipping EDLC is the wrong device for the rack and its LIC is behind Musashi's. Better standalone business than Musashi, worse pure supercap bet. At ~29x trailing / 5x book after a 3x run (now ~42% off May high), the option is largely in the price. Re-evaluate if a signed Delta contract with disclosed volume lands, if the stock de-rates toward the high-20,000s and hands the option back cheaply, or on LIC commercialization milestones and energy-density specs versus Musashi. Live price re-pull required before any action (yfinance data unreliable for this ticker; use history series, not info).
Sources
Source commentary (what the desks and the chat are saying)
The supercap/800V-DC buffering thesis is intact at the narrative level, but the two most-followed independent voices on it have turned cautious on Musashi specifically and rotated attention toward Vitzrocell as the cheaper, lower-execution-risk vehicle, caveated by Vitzrocell's LIC not yet shipping.
- Made in Japan (@InvestInJapan) published the Musashi HSC bull case on 4 April 2026 at 6.4x EV/EBITDA, sketching a ¥255bn HSC sum-of-the-parts against a then-~¥186bn total market cap. By 27 May, after the stock ran +172% in under two months, he un-paywalled the piece and downgraded the risk/reward, writing that "the only thing that's changed... has been the narrative," flagging underwhelming earnings and a repeated second-factory delay, and noting NVIDIA's Vera Rubin capacitor adoption as thesis confirmation. Stance: watchlist, not owned, "too many uncertainties." No Vitzrocell coverage from this author (he covers Japan).
- Collyer Bridge (@collyer, subscriber chat) is, as of late June, bearish/neutral on Musashi ("Nothing exciting," "bad setup," "down from 10k yen") and rotated to Vitzrocell as the preferred supercap vehicle, "waiting for Vitz to pull back more" before entering, consistent with his fast in/out discipline. His 19 June Vitzrocell thread established the key technical split via subscribers Bob and Min Htoo: Musashi = LIC/HSC (energy-denser, but pre-lithiated anode swells in prismatic form, hence the cylindrical sample requests); Vitzrocell = mature cylindrical EDLC (no swell, cheaper, longer cycle life, lower energy density, with a LIC product still "being prepared" and longer to build from scratch). The 29 June Musashi thread relayed a Morgan Stanley note: nothing new at the 26 June AGM, the Minami-Alps plant delayed to next year, and Musashi Energy Solutions' tangible fixed assets up to ¥13.8bn from ¥1.9bn (heavy capex already committed).
- Irrational Analysis was the proximate catalyst for the Musashi run (28 May supercap note); relayed the Delta-as-customer linkage for Musashi, which other sources could not independently confirm. Treat IA-asserted relationships as directional, not fact.
- Korea Invest Insights covers Vitzrocell (082920) directly in its "Korea Quality Re-Rating Watch" series (15/20/26/27 May 2026 editions). Not deeply mined here; a source to pull for a Korean-desk view.
- SemiAnalysis (cross-check above) validates the rack-level supercap/LIC use case under 800V DC but names neither Vitzrocell nor Musashi as a winner.
Key references
- Korea Economic Daily (한국경제), 23 June 2026 – CEO Jang Seung-guk confirms Delta LIC talks for US AI datacenters, decision within 2026.
- iM/IMFN broker report (Jan 2026) – segment mix, ownership, segment growth history, pre-run liquidity.
- yfinance 082920.KQ (clean equity data; the .KS suffix is a fund proxy) – price, share count, financial statements, cross-checked to Korean sources.
- Supplied supercapacitor technology primer – EDLC vs LIC/HSC physics, datacenter timescale hierarchy, 800V DC vs 48V-rail distinction.
- SemiAnalysis, "AI Training Load Fluctuations at Gigawatt-scale" (25 June 2025) – sector use-case primary source.
Appears in / comparisons
- True peer set (like-for-like Korean supercapacitor names): Vitzrocell vs Vinatech (126340.KQ) vs LS Materials (417200.KQ) — the genuinely comparable set, same country and same device family. Vinatech is the supercap pure-play (>90% of revenue) with the only booked datacenter-adjacent order (Bloom Energy), but loss-making and levered; LS Materials is ~74% aluminum with a minority EDLC slice on an unconverted Vertiv MOU. Vitzrocell is the best business of the three and the least supercap-pure.
- Narrative cross-reference, NOT a true peer: Vitzrocell vs Musashi (7220) — Musashi is a Japanese auto-parts shell with an HSC subsidiary; same technology conversation, different business entirely.
- Sector pages: ai-infrastructure, defense-tech.