| | Kioxia (285A) | Micronics (6871) | |—|—|—| | What you own | World’s #3 NAND maker, AI enterprise SSD ramp | Probe card consumable, HBM-leveraged | | Layer of value chain | Fab (cyclical, capital-intensive) | Equipment/consumable (specialty, capital-light) | | Market cap | $129B…
Two ways to play the AI-driven memory cycle. Kioxia is the NAND fab; Micronics is the probe-card consumable. Same thesis, different layer of the value chain, very different risk profiles.
| Kioxia (285A) | Micronics (6871) | |
|---|---|---|
| What you own | World’s #3 NAND maker, AI enterprise SSD ramp | Probe card consumable, HBM-leveraged |
| Layer of value chain | Fab (cyclical, capital-intensive) | Equipment/consumable (specialty, capital-light) |
| Market cap | $129B (mega-cap) | $3.5B (small-cap) |
| TTM rev growth | +21% | +26% |
| TTM op margin | 26% | 20% (24% on FY25) |
| TTM gross margin | 26% | 48% |
| Balance sheet | Net debt ¥1.0T | Net cash ¥12B |
| Beta | n/a (recent IPO) | 1.47 |
| Cyclicality | Severe (NAND -80%/+100% peak-trough EPS) | Moderate (-30%/+50%) |
| Quality verdict (3-test) | Vulnerable | Durable specialty |
| Governance | PE overhang (Bain ~28-30%, exiting) | Clean, no overhang |
| Distance from 52w high | -3.8% | -2.8% |
| Analyst PT premium | +5% (mean ¥37,214) | -26% (mean ¥9,900) — stock above mean |
| Coverage | Broad (15-22 analysts) | Thin (3-4 analysts) |
| Verdict at current price | PASS | WATCH |
| Better entry | <¥18,000 (-49%) | <¥10,500 (-21%) |
| Catalyst | 2026-05-15 earnings | 2026-05-13 earnings |
Both stocks are AI-memory-cycle plays. The bull case for both rests on the same three legs: 1. AI inference creates structurally higher NAND/HBM demand (KV cache + dataset loading) 2. Hyperscaler enterprise SSD + HBM3E/HBM4 ramps drive a 2-3 year up-cycle 3. Industry discipline (Samsung restraint, China contained) sustains margins
Kioxia is leveraged to NAND ASPs. When ASPs are up, Kioxia rips. When ASPs revert, Kioxia compresses fast (highest pure-play NAND beta of any listed peer).
Micronics is leveraged to NAND/HBM volume × test complexity. Probe cards are a per-die consumable; rising chip complexity (more pins, more KGD touchdowns) means probe-card revenue can grow even if NAND ASPs flatten. More volume-sensitive, less ASP-sensitive.
This is the key distinction. If you believe in AI memory volume continuing through 2027 but worry about Samsung breaking pricing discipline, Micronics is the cleaner expression. If you believe in continued NAND ASP elevation through 2027, Kioxia gets you maximum operating leverage.
| Metric | Kioxia 285A.T | Micronics 6871.T | Notes |
|---|---|---|---|
| Price | ¥35,470 | ¥13,360 | Both near 52w high |
| 52w range | ¥1,805 – ¥36,870 | ¥2,922 – ¥13,750 | Kioxia 19.6× off lows; MJC 4.7× off lows |
| Market cap | ¥19.3T (~$129B) | ¥518B (~$3.5B) | 37× size difference |
| EV | ¥19.8T | ¥474B | MJC has net cash |
| TTM Revenue | ¥1.7T | ¥70B | 24× size difference |
| TTM Rev growth | +21% | +26% | MJC growing faster |
| Gross margin | 26% | 48% | MJC structurally higher |
| Op margin (TTM) | 26% | 20% (24% FY25) | Kioxia higher on cycle peak; MJC structural |
| Net income margin | 9.9% | 17.2% | MJC nearly 2× Kioxia |
| FCF (TTM) | n/a (¥108B per yfinance) | -¥13.5B (WC build) | Kioxia FCF positive; MJC negative on ramp |
| Net debt / cash | ¥1.0T net debt | ¥12B net cash | MJC structurally cleaner |
| P/E trailing | 124× | 43× | |
| P/E forward | 7.8× (on ¥490B guide) / 39× (peak EPS) | 61× | |
| P/Sales | 11.5× | 7.4× | |
| P/Book | 19.7× | 7.8× | Kioxia book v. extreme |
| Cycle position | Peak | Peak | Both at risk |
| Insider ownership | 17.8% | 18.5% | Similar |
| Analyst PT mean | ¥37,214 | ¥9,900 | Kioxia +5%, MJC -26% |
| Analyst PT high | ¥62,500 (+76%) | ¥14,000 (+5%) | |
| Analyst PT low | ¥17,000 (-52%) | ¥5,800 (-57%) | Both have meaningful downside in low PT |
| Coverage | 15-22 analysts | 3-4 analysts | Kioxia broad; MJC thin |
| Earnings date | 2026-05-15 | 2026-05-13 | Both within 18 days |
| Distance from 50dMA | n/a | +17% | MJC extended |
| Distance from 200dMA | +194% | +72% | Kioxia massively extended |
| Dimension | Kioxia | Micronics |
|---|---|---|
| Margin durability through cycle | FY23 GM -12%, op -23% | FY23 GM 45%, op 14% (still profitable!) |
| Capital intensity | Heavy (fab) | Light (specialty assembly) |
| Customer concentration | Top-5 ~50-60% | Top-2 ~30-40% |
| Switching costs | Low (NAND is interchangeable across qual) | High (2-4mo qual cycle per chip) |
| Pricing power | Price-taker | Modest (technical complexity) |
| Disruption threat | YMTC (China) | Korean domestic + Chinese probe makers |
| Balance sheet flexibility | Net debt, refinanceable | Net cash, fortress |
| 3-test quality verdict | Vulnerable | Durable specialty |
Read: Micronics is the higher-quality business. Cycle-through margin profile, balance sheet, and customer stickiness all favor MJC. Kioxia is the more leveraged play.
| Dimension | Kioxia | Micronics |
|---|---|---|
| Ownership clean? | No — Bain ~28-30% selling, Toshiba/JIP ~27%, SK Hynix ~14% | Yes — diffuse |
| Insider transactions | None reported (limited disclosure) | None reported |
| PE overhang | Yes — multi-year sell-down | No |
| Strategic competitor on cap table | Yes — SK Hynix with veto rights to 2028 | No |
| TSE float compliance | Below 35% target until ~2027 | Above 35% |
| Board independence | Bain + Toshiba/JIP appointees + minority independents | Standard TSE Prime |
| Mgmt DD verdict | Yellow | Neutral (insufficient depth, no flags) |
Read: Micronics has materially cleaner governance. Kioxia’s PE-backed-IPO governance is a real overhang that compresses any rally. The market’s “premium” on Kioxia is despite this, not because of it.
Both stocks are at peak-cycle valuations. The question is what does the next 18 months look like.
| Kioxia | Micronics | |
|---|---|---|
| Bull upside | +27 to +41% | +27 to +50% |
| Bear downside | -50 to -65% | -32 to -51% |
| Risk/reward | Poor (1:1.5 to 1:2 unfavorable) | Better but still mixed (1:1 to 1:1.2) |
Both are unfavorable at current price. Micronics is less unfavorable.
Pink already holds JEM (6855.T) — Japan Electronic Materials, #5 global probe card, $2.5B mcap, similar HBM exposure to Micronics. Adding Micronics to a portfolio that holds JEM is substantial duplication of the probe-card thesis.
| Comparison | JEM (6855.T, held) | Micronics (6871.T, candidate) |
|---|---|---|
| Probe-card focus | Memory + logic | Memory + logic |
| HBM exposure | Direct | Direct |
| Geographic mix | Japan-centric, Korean memory dominant | Same |
| Position in screen | #8 ranked, structurally weakest held (P<MA50, -35% off high, but +33% annual EPS) | #3 ranked unheld, near 52w high |
| Market cap | $2.5B | $3.5B |
| Valuation | Cheaper | More expensive |
Honest read for Pink: if you want MORE probe-card exposure, the right answer is probably to add to JEM on the structural weakness rather than buy Micronics at peak. The two are not differentiated enough to justify a 2-stock portfolio in the same niche.
Counter-argument for buying both: if you believe HBM volumes structurally double through 2026-27, owning the two largest Japanese probe makers as a basket diversifies single-name execution risk (one loses a Samsung qualification → other potentially gains). But this is a smaller hedge than just sizing JEM correctly.
Both report within 2 days of each other (MJC May 13, Kioxia May 15). The print sequence creates an information arbitrage opportunity:
Trading playbook for someone wanting exposure: wait for both prints, then size into the weakest reactor (which is most likely to be MJC given thin coverage = surprise risk).
| If your thesis is… | Best expression |
|---|---|
| AI-NAND ASP elevation through 2027 | SanDisk (SNDK) — same JV exposure as Kioxia, no PE overhang, US-listed |
| AI-memory volume + test complexity | Add to JEM (6855.T) — already held, cheaper, similar HBM beta |
| Highest beta / maximum cycle leverage | Kioxia (but only on a major pullback to <¥18,000) |
| Specialty quality with HBM optionality | Micronics (but only on pullback to <¥10,500) |
For Pink specifically: the cleanest action is none of the above today. Both stocks are at peak; Pink already holds the cheaper probe-card analog (JEM); the AI-NAND thesis has a cleaner expression (SNDK). The screen flagged both as “deserve a deep-dive” and the deep-dives reveal: the deep-dives confirm the thesis but disqualify the entry.
wiki/285A/285A.mdwiki/6871/6871.mdNeither is a buy today. The screen correctly flagged the quality of both, but the entry has been ruined by the 12-month rip. Watch both, prep size, deploy on the next cycle pause or post-print pullback.
If forced to pick one to enter: Micronics — better quality, better balance sheet, cleaner governance, less downside. But still better to wait for ¥10,500.