Ticker correction (2026-06-18): every "ALMU" in this file means Almonty Industries (the tungsten Layer-1 / Sangdong pure-play). That is the correct company — but the ticker is wrong: ALMU is Aeluma Inc., an unrelated compound-semiconductor / QD-laser-on-silicon photonics firm. Almonty's real tickers are ALM (NASDAQ) / ALMTF (OTC) / AII.TO. The analysis and conclusions here (Sangdong off-take, Plansee/GTP partners, no KMT link) are about Almonty and stand; only the symbol is mislabeled. Root cause + full trace in research/tasks/TASKS-RESEARCH.md (ALMU/Aeluma conflation, closed 2026-06-18).

Question: Two follow-ups to the 2026-05-25 KMT deep-dive: (93) Does KMT's gross margin actually respond to tungsten APT price cycles — does the cemented-carbide tooling franchise pass through (or even widen spread), or is KMT a price-taker? (95) Does ALMU's Sangdong restart have any disclosed off-take partner that is KMT or a KMT subsidiary — i.e. is there a signed Layer-1-to-Layer-3 link?

Verdict: Mixed. (93) Passes-through, does not widen-spread — KMT gross margin barely moves against large APT swings; the franchise insulates margin rather than amplifying it. (95) Not confirmed — no KMT off-take exists; the disclosed Layer-1 link runs to a KMT competitor (GTP/Plansee), not to KMT.

What we're asking

The 2026-05-25 KMT deep-dive (2026-05-25-kmt-deep-dive-tungsten-carbide-toll-booth) asserted a "Layer-3 toll-booth" thesis: KMT sits downstream of tungsten chemistry and either passes the tungsten-carbide (WC) input cost through to customers or eats compression. The same-day Financials addendum showed the current (FY26) cycle inflection with gross margin expanding 32.1%→35.1% YoY, and read that as "the pass-through working." But a single up-cycle quarter doesn't distinguish pass-through (margin holds/expands because ASP discipline offsets input cost) from demand-driven mix (margin expands because volumes/fixed-cost absorption recovered, regardless of tungsten). The clean test is the historical gross-margin RESPONSE to known tungsten APT cycles — the 2014-2016 collapse and the 2021-2023 recovery — which this investigation assembles.

Question (95) is the structural payoff check: if ALMU (Layer-1 Sangdong pure-play) signed an off-take with KMT (the Layer-3 toll-booth), that would be a direct evidentiary link between the two ends of the thesis chain we track in supply-chain-traces.

What we found

(93) Tungsten APT price history (external market data — web sources)

Granular date-stamped APT Rotterdam prints are paywalled (Argus PA00143630000, Fastmarkets, Asian Metal, SMM — all subscription). Argus and the USGS PDF tables would not render via fetch (403 / 503 / binary-encoded). What is freely citable bounds the cycles well enough to test direction and rough magnitude:

Bounded magnitudes: the 2014→2015/16 collapse was roughly -45% to -59% (from a ~$350-440 top to ~$180). The 2021→2023 recovery was a milder +20-30% off the COVID-era low into the low-$300s. (Note: the 2025-2026 China-export-control spike to ~$3,000+/mtu is a separate, post-window regime and is NOT what these two test windows cover — that move post-dates KMT's available filings.)

(93) KMT annual gross margin for the same windows (filings truth — SEC 10-Ks)

KMT fiscal year ends June 30. Annual revenue + gross profit pulled from the FY2014, FY2016, FY2021, FY2022, FY2023 10-Ks (via SEC EDGAR / 8-K exhibits). Gross margin computed (code, not prose):

Collapse window (FY2014→FY2016): APT fell ~-45% to -59%. KMT gross margin fell 31.6% → 29.4% = -226bp (and only -120bp by FY2015). Revenue fell -26% over the same span (oil & gas + mining + FX + divestiture), so the bulk of the pain was volume / fixed-cost absorption / mix, not raw tungsten cost. A ~50% input-commodity collapse produced barely a 2-point gross-margin move.

Recovery window (FY2021→FY2023): APT rose ~+20-30%. KMT gross margin rose 30.0% → 32.3% (FY22) → 31.1% (FY23) = +228bp then back to +113bp net. Again the magnitude of the margin move (~1-2 points) is small relative to the commodity move and is better explained by post-COVID volume recovery and pricing actions than by tungsten pass-through specifically.

Read: across BOTH a -50% collapse and a +25% recovery in tungsten APT, KMT's gross margin moved only ~1-2 percentage points. The margin line is far more sensitive to volume / fixed-cost absorption (the industrial cycle) than to the tungsten commodity. That is the signature of a franchise that insulates margin — it passes the WC input cost through (margin doesn't collapse when tungsten halves, and doesn't melt up when tungsten doubles) but it does not widen spread on a tungsten up-move. ASP discipline holds the floor; it does not turn tungsten into operating leverage.

Caveat on the FY26 addendum: the deep-dive's same-day addendum read FY26 Q3's 32.1%→35.1% YoY gross-margin jump as "pass-through working." This historical study reframes that: a ~300bp jump is large by KMT standards and is more consistent with volume-driven fixed-cost absorption in a recovery (revenue +21.8% YoY) than with tungsten pass-through — because in the two prior full cycles, even ±50% tungsten moves never moved gross margin more than ~2 points. The FY26 margin lift is the industrial cycle inflecting, with tungsten a secondary (insulated) input. The toll-booth thesis survives but as a margin-stability story, not a margin-amplification story.

(95) ALMU-Sangdong off-take — disclosed counterparties

From Almonty press releases / investor materials (almonty.com, Plansee Group, mining-technology, Northern Miner, Korea Times via search):

Named off-take / supply counterparties for Sangdong:

1. Plansee Group — 15-year off-take covering Phase-I production, floor price US$235/mtu. Plansee is also Almonty's largest shareholder (~14%).
2. Global Tungsten & Powders (GTP) — the US processor that takes the concentrate; GTP is a Plansee subsidiary (Sangdong ore → processed at GTP in the USA). Bundled into the Plansee 15-year agreement.
3. Tungsten Parts Wyoming, Inc. (TPW) — US defense contractor; binding off-take for min. 40 t/month tungsten oxide for US defense applications.
4. Metal Tech (MT) — Israeli tungsten processor converting the oxide to tungsten metal powder (paired with the TPW defense deal).
5. SeAH Group — molybdenum off-take (100% of Sangdong moly output, floor ~US$19/lb). Moly, not tungsten — co-product.

Does KMT appear? No. Kennametal is not named in any disclosed Sangdong off-take, supply agreement, or investor material reviewed. The almonty.com defense-off-take release explicitly names only TPW and Metal Tech; the long-term tungsten off-take runs to Plansee / GTP.

Thesis-relevant negative: the disclosed Layer-1 link does NOT confirm the KMT toll-booth — it points the other way. GTP (Global Tungsten & Powders) is itself a WC-powder producer and a direct competitor to KMT's Infrastructure segment. So the cleanest western tungsten Layer-1 output (Sangdong) is contractually committed to a KMT competitor's vertically-integrated chain (Plansee/GTP), plus a US-defense channel (TPW/Metal Tech) — not to KMT. This is a (modest) datapoint AGAINST the "KMT will source the new western tungsten" sub-narrative, and a reminder that Plansee/GTP is the vertically-integrated rival that already locked the marquee non-China supply.

Verdict + reasoning

• (93) Pass-through verdict: PASSES-THROUGH, does not widen-spread (price-insulated, not price-amplified). Across a ~-50% APT collapse (FY14→16: GM 31.6%→29.4%, -226bp) and a ~+25% APT recovery (FY21→23: GM 30.0%→31.1%, +113bp net / +228bp peak FY22), KMT gross margin moved only ~1-2 points each way. The franchise demonstrably absorbs/passes the tungsten input cost — margin neither collapses on a tungsten crash nor melts up on a tungsten spike. KMT is not a tungsten price-taker at the margin line, but it is also not a spread-widener: tungsten is an insulated input, and the dominant driver of KMT gross margin is the industrial volume cycle (fixed-cost absorption), not the APT print. The deep-dive's "toll-booth that widens spread on tungsten up-moves" framing is too strong; the correct framing is "toll-booth that holds margin stable through tungsten cycles."

• (95) ALMU off-take: NO KMT link — thesis sub-claim UNPROVEN / mildly NEGATIVE. Named off-takers: Plansee (15yr, $235/mtu floor), Global Tungsten & Powders (Plansee sub, US processing), Tungsten Parts Wyoming + Metal Tech (US-defense oxide), SeAH (moly co-product). KMT appears nowhere. The disclosed Layer-1 tungsten output is locked to Plansee/GTP — a KMT competitor — not to KMT. No signed Layer-1→Layer-3-via-KMT link exists in public disclosure.

Data I could NOT source (bounded transparency)

• Granular date-stamped APT Rotterdam monthly prints are paywalled (Argus / Fastmarkets / Asian Metal / SMM). I bounded the cycles with free secondary sources (INN, Core Consultants, USGS, businessanalytiq, NAI500) — directionally solid, magnitudes approximate (collapse -45/-59%, recovery +20/30%). The exact 2014 peak and exact 2015-16 trough monthly prints would need a paid Argus/Fastmarkets history.
• KMT segment-level gross margin (Metal Cutting vs Infrastructure) for these windows — only consolidated gross margin is used here. Infrastructure (the more tungsten-exposed segment) might show a larger tungsten response masked at the consolidated line; a segment-level pull from the 10-K MD&A would refine this. The consolidated read is sufficient for the price-taker-vs-pass-through question but not for isolating the Infrastructure-specific tungsten beta.
• Whether KMT has an UNdisclosed Sangdong relationship — only public disclosure was checked; absence of a named off-take is not proof of zero commercial relationship, but a material off-take would normally be disclosed by ALMU (it discloses TPW/MT/Plansee/SeAH by name).

Sources

External market data (tungsten APT — event/price truth, web):

• INN — "Tungsten Outlook: Does 2016 Have Potential?" and "When Will the Tungsten Price Go Up?" (2014-2016 collapse to $180-190/mtu)
• Core Consultants — "Tungsten Market Outlook 2025-2030" (2023 $312/mtu; 2011-2015 collapse reference)
• USGS 2018 Minerals Yearbook — Tungsten (2017 $245 / 2018 $326 concentrate proxy)
• businessanalytiq / NAI500 — long-run ~$300/mtu average; 2021-2023 recovery framing
• Argus Tungsten Monthly Outlook 2023-07 (European APT $325-330/mtu mid-2023)

KMT financials (filings truth — SEC EDGAR, CIK 0000055242):

• KMT FY2014 10-K (rev $2,837.2M / GP $897.0M / GM 31.6%)
• KMT FY2015 10-K (rev $2,647.2M / GP $806.0M / GM 30.4%)
• KMT FY2016 10-K (rev $2,098.4M / GP $616.1M / GM 29.4%)
• KMT FY2021 10-K (rev $1,843M / GP $552.5M / GM 30.0%)
• KMT FY2022 10-K (rev $2,009M / GP $648.0M / GM 32.3%)
• KMT FY2023 10-K (rev $2,078M / GP $646.4M / GM 31.1%)
• stonks/data/stocks/KMT/massive/income-statements.json (FY24 Q4–FY26 Q3, recent gross-margin series; pulled 2026-05-25) — used in the deep-dive addendum, cross-referenced here

ALMU off-take (event truth — ALMU investor materials):

• almonty.com — "Almonty Enters into Binding Offtake Agreement to Supply Tungsten Oxide Solely for U.S. Defense Applications" (TPW + Metal Tech named; KMT not named)
• Plansee Group — "Stable and independent supply of tungsten" (15yr Plansee/GTP off-take; GTP is Plansee subsidiary; Plansee ~14% ALMU holder)
• mining-technology.com — Almonty/SeAH molybdenum off-take
• Northern Miner / Korea Times / discoveryalert — Sangdong off-take structure ($235/mtu floor)

Cross-references

• Parent deep-dive: research/deep-dives/2026-05-25-kmt-deep-dive-tungsten-carbide-toll-booth.md (addendum added 2026-05-29 folding this summary)
• Backlog rows: TASKS-RESEARCH.md [@claude-bg-kmt-tungsten] (rows 93 + 95)
• Watchlist: stonks/watchlists/supply-chain-traces.json (KMT, ALMU)