VERDICT: NO POSITION AT CURRENT PRICE

The thesis is correct. The price already reflects it. No forced actors, no information edge, no alpha at current levels. Watchlist at A$17-18.

What SGM Is

World's largest publicly listed metal recycler. Three businesses:

1. Sims Metal (NAM + ANZ + Global Trading) — ferrous and non-ferrous scrap. Cyclical commodity business. ≈75% of revenue. Challenged by Chinese steel exports suppressing global scrap prices.

2. SA Recycling — 50% JV in US scrap. Hub-and-spoke model, 150 feed yards into 24 shredders. A$120M EBIT contribution in FY25. Well-run, growing through bolt-ons.

3. Sims Lifecycle Services (SLS) — IT asset disposition for hyperscale data centers. Decommissions servers, tests and repurposes DDR4 memory modules, resells into IoT/edge/automotive. ≈6% of revenue but increasingly dominant in earnings. This is the thesis.

Ownership: Mitsui & Co. 17.3% (strategic, board seat, since 2007). Orbis Investment 15-19% (value investor). Pinnacle Investment 5.01% (new substantial holder, Dec 2025).

The DDR4 Dynamic

All three major DRAM manufacturers — Samsung, SK Hynix, Micron — have issued end-of-life notices for DDR4 and exited or are exiting production by April 2026. Micron killed its entire Crucial consumer brand. Wafer capacity has been reallocated to DDR5 and HBM for AI workloads. No manufacturer is building new DDR4 capacity.

DDR4 remains the workhorse for IoT, edge computing, automotive, networking equipment, and legacy enterprise systems. Installed base is massive. Demand persists while new supply goes to zero.

Result: DDR4 spot prices +450% YoY as of H1 FY26. Sims repurposes DDR4 from decommissioned servers (tests, reprograms, resells) into this scarcity. SLS revenue +70% in H1 FY26. EBIT margin expanded 7.7 percentage points.

Cross-corpus validation from 6 companies in last 90 days:

The supply squeeze is not management spin. It's confirmed across the semiconductor supply chain by companies with no incentive to help Sims' narrative.

Duration estimate: Structural through 2029-2030, then collapses as DDR4 ecosystem dies. This is a 4-6 year scarcity window, not a 2-year spike. But terminal value of the DDR4 repurposing business is zero — every year there are fewer DDR4 modules in existence, and eventually DDR5 displaces legacy demand entirely.

Sources: TrendForce DDR4 exit timeline (Jul 2025), TrendForce memory exit strategy (Jan 2026), Micron investor relations (EOL notice), earnings transcripts via cross-corpus search.

Factor Decomposition

MODEL: SGM.AX ~ EWA + FXA + COPX + SLX + PICK
Period: Feb 2025 - Feb 2026 | 245 observations

Factor           Beta    Var Contribution   Edge?
─────────────────────────────────────────────────
PICK (metals)    +0.73   10.2%              NO
FXA (AUD/USD)    +0.31    1.6%              NO
EWA (ASX mkt)    +0.01    0.0%              NO
COPX (copper)    -0.26   -4.1%              NO
SLX (steel)      -0.11   -1.0%              NO
─────────────────────────────────────────────────
Idiosyncratic            94.1%
─────────────────────────────────────────────────
R² = 5.9%
α (backward) = 44.3% annualized
σ_idio = 30.5%
Idio Sharpe = 1.45

94% idiosyncratic across every factor specification tested. Stable result — not a modeling artifact. SGM's returns are almost entirely company-specific.

No edge on any factor. The 44% backward-looking alpha is the DDR4 windfall already captured in price. The question is forward alpha, not trailing.

What Current Price Requires

                    FY25 Actual    Market Implies (FY26)
────────────────────────────────────────────────────────
Underlying EBITDA    A$430M         A$550-600M
Underlying EBIT      A$175M         A$300-350M
SLS EBIT             ~A$33M         A$100-130M
Forward P/E          —              20x (on underlying)
EV/EBITDA            —              8-9x

The market is pricing continued SLS hypergrowth AND ferrous stability. Not heroic, but not conservative either. The DDR4 surprise has already been captured in a +62% 1Y move.

Statutory Earnings: The 25% Overstatement

This is the finding that most analysts gloss over.

FY25 SIGNIFICANT ITEMS (from Results Presentation, Slide 47):

                                        Pre-Tax    After-Tax
────────────────────────────────────────────────────────────
Underlying EBIT                          174.9M      83.1M
  Business transaction gains              (1.7)M     (2.5)M
  Non-qualifying hedges                  (14.3)M    (14.3)M
  Closure of SRR (plasma gasification)   (25.1)M    (17.6)M
  Redundancies and restructuring         (23.5)M    (17.3)M
  Asset impairments                      (45.1)M    (34.0)M
────────────────────────────────────────────────────────────
Statutory EBIT                            68.6M       2.4M

Of the A$107M gap between underlying and statutory:

• True one-offs (A$27M): SRR closure + business transactions. Won't recur.
• Recurring pattern (A$83M): Hedges (mark-to-market every period), restructuring (FY24: A$10.9M → FY25: A$23.5M, trending UP under "Simplify Sims"), impairments (FY24: A$64.2M → FY25: A$45.1M, including ongoing UK Metal/Unimetals receivable losses with a further GBP 30M provision taken in H1 FY26).

True sustainable EBIT is ~A$135-140M, not A$175M. The market values the stock on "underlying" earnings that overstate reality by ≈25%.

At true EBIT of ~A$135M, forward P/E is ≈40x, not ≈20x.

Source: FY25 Results Presentation Slide 47 (significant items reconciliation), H1 FY26 transcript (additional GBP 30M Unimetals provision).

Positioning: Nobody Is Forced to Do Anything

Short interest:      0.69% (down from 2.3% in March 2025 — squeeze already happened)
Days to cover:       1.6 days
Listed options:      NONE (no ASX ETOs on SGM)
ADR options:         NONE (SMSMY $62K avg daily dollar volume — untradable)

Insider activity:

• CEO Mikkelsen: Bought 9,425 shares at A$14.85 in May 2025 (A$140K). No purchases at current levels. No selling.
• CFO Ranson: Net seller (Sep 2025, amount not confirmed from primary source).
• Board: Token only.

Substantial holders:

• Mitsui 17.3%: Strategic anchor. Won't sell, won't buy more. Creates floor.
• Orbis ≈15-19%: Value investor. No recent Form 604 filing found to confirm reduction. Earlier secondary-source claim of reduction UNVERIFIED.
• Pinnacle 5.01%: NEW substantial holder as of Dec 22, 2025. Bought at ~A$20-21. Mildly positive.

Forced actors if thesis plays out (stock → A$28): None. Shorts already covered. No gamma. No index trigger. Repricing happens through analyst upgrades (5 of 13 at sell) over weeks to months. Grind, not gap.

Forced actors if thesis fails (stock → A$17): Momentum funds on 200DMA breach. But Mitsui's 17.3% won't sell, creating a floor around A$18-19.

Variable Templates

SLS Revenue Growth FY26

Consensus:     +40-50%
Our estimate:  +65-80%
Delta:         +20-30ppts
Evidence:      H1 +70% (transcript). CEO: "expect second half materially better
               than first half." Ireland ramp contributing from June/July.
Base rate:     ITAD market 8-9% CAGR. SLS outperformance is DDR4 price-driven
               (+450% pricing vs +18% volume = revenue is pricing, not growth).

DDR4 Pricing Duration

Consensus:     Strong through FY26, normalization FY27
Our estimate:  Structural through 2029-2030
Delta:         +3-4 years
Evidence:      All three manufacturers exited DDR4 by April 2026. Zero new
               capacity being built. IoT/auto/edge demand persists. 6 companies
               confirm constraints in last 90 days.
Base rate:     Memory transitions (DDR3→DDR4) took 4-5 years for prior gen
               to become truly scarce. DDR4→DDR5 started 2024.

SLS EBIT Margin

Consensus:     ≈10-12%
Our estimate:  14-17% (H1 was ≈15.7%)
Delta:         +3-5ppts
Evidence:      H1 margin expanded 7.7ppts YoY. Automation investment. Ireland
               adds fixed cost initially but volume leverage in H2.
Base rate:     ITAD industry 8-15%. SLS at high end driven by DDR4 scarcity
               premium. Margin compresses when DDR4 normalizes (post-2030).

True Earnings vs Underlying

Consensus:     FY26 underlying EBIT A$300-350M
Our estimate:  True sustainable EBIT A$250-280M (after recurring significant items)
Delta:         -15-20%
Evidence:      FY25 had A$83M in recurring "significant items." Restructuring
               trending UP. UK Metal/Unimetals receivable still impaired with
               additional provisions taken in H1 FY26.
Base rate:     Companies with persistent "one-off" adjustments over 3+ years:
               the adjustments ARE the business. (FY24: A$75M items. FY25: A$107M.)

Competitive Landscape

SK Tes (SK Group subsidiary): 36,000 sq ft Ireland facility opened mid-2025 — 10 months ahead of Sims' 120,000 sq ft facility opening April 2026. 40+ facilities in 20 countries vs Sims' ≈15. Backed by SK Group ($150B+ conglomerate). Parent company SK Hynix manufactures the DDR4/DDR5 chips — vertical integration Sims can never replicate.

Key unknown: Whether SK Tes does DDR4 memory repurposing specifically, or focuses on data destruction/hardware recycling. Sims' moat is the testing/reprogramming/resale of DDR4 modules, not just shredding. Cannot confirm from primary sources whether SK Tes competes on this specific capability.

Iron Mountain (IRM): Mentioned by SLS President Ingrid Sinclair as known competitor. Sinclair: "Don't see doing data centers." IRM's own transcripts don't mention Sims or data center ITAD as a growth area. Not a material threat.

Source: SK Tes press releases (Ireland, Virginia facilities), H1 FY26 earnings transcript (Sinclair commentary).

Catalysts and Repricing

DATE              EVENT                        WHAT MUST HAPPEN TO REPRICE
──────────────────────────────────────────────────────────────────────────
Mid-Mar 2026      Nashville SLS Investor Day   SLS EBIT guidance >A$130M, OR
                                               named hyperscaler contract, OR
                                               multi-year facility rollout plan.
                                               Without one of these: non-event.

Apr 2026          Ireland facility opens       Already guided. Operational
                                               milestone, not financial. No EBIT
                                               until June/July per CEO.

Aug 2026          FY26 Full Year Results       6 months out. Too far to position.
──────────────────────────────────────────────────────────────────────────

DDR4 spot pricing is UNOBSERVABLE to outside investors. No public index,
no futures market. Sims has 2-3 month visibility; you have semi-annual
(earnings releases). Information asymmetry runs AGAINST retail.

Repricing speed: Slow. No forced actors. No options gamma. No short squeeze. Analyst upgrade cycle takes 2-8 weeks post-event. Institutional reallocation is quarterly.

Edge Audit

Market cap:        A$4.4B (~US$2.8B)         Not small enough for retail edge
Analyst coverage:  13 (4 buy, 4 hold, 5 sell) Fully covered
1Y momentum:       +62%                       Already discovered
Insider buying:    Last at A$14.85 (May '25)  53% below current price
DDR4 narrative:    Public knowledge            TrendForce, Tom's Hardware, every semi analyst
Cross-corpus work: Confirmatory, not novel     Any Bloomberg terminal does this

Specific edge identified: NONE
Every claim in this memo is derivable from the H1 transcript and public data.
A sell-side analyst would not disagree with any finding presented here.

Kill Conditions

All four triggered simultaneously:

1. No delta. Probability-weighted target A$23 vs current A$22.66. Expected return +1.5% before adjusting for true earnings overstatement.

2. No factor edge. 94% idio is clean. But no edge on the idiosyncratic component — it's DDR4 pricing, which is unobservable and consensus.

3. No forced actor. 0.69% short interest (shorts already covered from 2.3%). No listed options. No index trigger. No mandate-driven buying threshold.

4. No informational advantage. DDR4 pricing: 2-3 month lag vs insiders. 13 analysts all heard the same transcript. Cross-corpus validation is routine sell-side work.

Recommendation

NO POSITION at A$22.66.

Correct thesis, fair price, no mechanism to extract alpha.

SLS Harvesting Model: What's the DDR4 Window Worth?

DDR4 structural scarcity runs 2025-2030. Terminal value of the repurposing business is zero. The question is how much cash Sims extracts in the window.

SLS CASH FLOW MODEL (A$M)
─────────────────────────────────────────────────────────────────────────
                 FY25A   FY26E   FY27E   FY28E   FY29E   FY30E   TOTAL
─────────────────────────────────────────────────────────────────────────
Revenue           427     730     900     950     800     500
  Growth           —     +71%    +23%     +6%    -16%    -38%
  Driver          base   DDR4    Ireland  peak   decline  runoff
                          price   + vol   plateau

EBIT margin        8%     15%     16%     15%     12%      8%
EBIT               33     110     144     143      96      40     566
Tax @ 25%          (8)    (28)    (36)    (36)    (24)    (10)   (142)
After-tax EBIT     25      82     108     107      72      30     424
─────────────────────────────────────────────────────────────────────────
Cumulative NOPAT                                                  A$424M

Assumptions:

• FY26: H1 was +70%, H2 "materially better" (CEO). Ireland contributes from June.
• FY27: Full year Ireland (1M units), peak DDR4 pricing, automation benefits.
• FY28: DDR4 volume plateau — peak decommission wave, but pricing stabilizes.
• FY29: DDR4 ecosystem shrinking. Fewer servers to decommission. IoT migration to DDR5 begins.
• FY30: Rapid decline. DDR5 repurposing may exist but unproven and won't carry DDR4-era margins.
• Margins: Peak at 16% (FY27) driven by pricing + automation. Compress as DDR4 scarcity eases post-2028.

Present value of SLS harvesting window at 12% discount rate: ~A$340M.

At 10%: ~A$360M. At 15%: ~A$310M.

This is what the DDR4 window is worth to an equity holder — roughly A$310-360M in present value of after-tax earnings over the scarcity period, then zero.

Sum-of-Parts: What SGM Is Actually Worth

COMPONENT              METRIC         MULTIPLE    VALUE     BASIS
──────────────────────────────────────────────────────────────────
Metal (NAM+ANZ+GTO)    True EBIT      6.0x EV     A$600M   A$100M true EBIT
                       A$100M                               (A$142M underlying
                                                            minus A$42M recurring
                                                            restructuring + impairments)

SA Recycling (50%)     EBIT A$130M    7.0x EV     A$910M   Consistent performer,
                                                            growing through bolt-ons

SLS: Going-concern     EBIT A$110M    10x EV     A$1,100M  FY26 run-rate, ITAD
     (ex-DDR4 premium)                                      multiple on sustainable
                                                            base (not DDR4 peak)

SLS: DDR4 harvesting   PV of A$424M   1.0x        A$340M   5-year NPV at 12%,
     surplus           cumulative                           terminal = 0

TCT acquisition        EBITDA A$25M   5.0x EV     A$125M   Management guided,
                                                            post-synergy
──────────────────────────────────────────────────────────────────
ENTERPRISE VALUE                                  A$3,075M
Less: Net debt (est.)                             (A$400M)
Plus: Property portfolio (Houston land + others)   A$200M
──────────────────────────────────────────────────────────────────
EQUITY VALUE                                      A$2,875M
Shares outstanding                                   194M
──────────────────────────────────────────────────────────────────
SOTP FAIR VALUE:                                  A$14.82
CURRENT PRICE:                                    A$22.66
PREMIUM TO SOTP:                                     +53%

Key SOTP decisions:

1. Metal true EBIT = A$100M, not A$142M underlying. FY25 metal underlying EBIT was ~A$142M (NAM A$80M + ANZ A$72M - central costs + other). But restructuring (A$24M) and impairments (A$45M) are recurring metal business costs. Netting ~A$42M of recurring charges against metal, true EBIT = ~A$100M. At 6x (cyclical industrial multiple, depressed ferrous), this is generous.

2. SLS split into going-concern + harvesting surplus. The going-concern SLS business (ITAD services, hardware decommissioning, data destruction) would exist without DDR4 scarcity. Pre-DDR4-spike EBIT was ~A$33M (FY25). Growing with data center buildout at 15-20% annually, sustainable FY26 base EBIT ex-DDR4 premium is ~A$50-60M. At 10x, that's A$500-600M. But current margins (15%+) include the DDR4 windfall. I've used A$110M at 10x, splitting the difference — acknowledging SLS is a better business than pre-DDR4 but not permanently a 15%-margin business. The DDR4 surplus above this base is modeled separately as a finite cash flow stream with PV of A$340M.

3. SA Recycling at 7x. Best-performing segment. Consistent, growing, well-managed. Deserves a premium to core metal.

4. Property at A$200M. CEO guided "excess USD 100M" for Houston land alone. Group-wide property portfolio likely higher but unverified.

Sensitivity:

                    Bear        Base        Bull
─────────────────────────────────────────────────
Metal multiple      5.0x        6.0x        7.0x
SLS multiple        8.0x       10.0x       12.0x
DDR4 PV          A$280M      A$340M      A$400M
SAR multiple        6.0x        7.0x        8.0x
─────────────────────────────────────────────────
Equity value     A$2,175M    A$2,875M    A$3,725M
Per share          A$11.21     A$14.82     A$19.20
─────────────────────────────────────────────────
vs current        -50.5%      -34.6%      -15.3%

Current price A$22.66 is above even the bull SOTP case (A$19.20). The market is either: (a) assigning SLS a 15x+ multiple (tech premium for a recycler), or (b) capitalizing DDR4-era earnings as permanent, or (c) both.

Corrected Watchlist Entry

The original memo said A$17-18 based on sloppy math. Here's what actual numbers say:

SOTP fair value:         A$14.82
Bull SOTP:               A$19.20
30% margin of safety:    A$14.82 × 0.70 = A$10.37

ENTRY PRICE LOGIC:
  At A$15:  1% premium to base SOTP. No margin of safety.
            DDR4 optionality is roughly free.
            CEO bought at A$14.85 — you're at his price.

  At A$12:  19% discount to base SOTP.
            Real margin of safety. DDR4 is free optionality.
            Mitsui floor (≈17% holder) likely holds here.

  At A$19:  28% premium to base SOTP, at bull case.
            Paying for DDR4 continuation. No margin of safety.
            This is where the stock is interesting ONLY IF
            Nashville reveals binding hyperscaler contract.

Revised watchlist: A$14-15. Not A$17-18. The original was too generous because it used "underlying" EBIT — the same metric I criticized the market for relying on.

At A$14-15:

• True P/E on FY26 true EBIT A$250M at 25% tax = A$188M NPAT → P/E = 14-15x. Reasonable for a cyclical industrial with a finite growth kicker.
• DDR4 harvesting value (A$340M PV) is free optionality.
• Aligns with CEO's own money (A$14.85 open market purchase, May 2025).

Corrected Orbis Treatment

The original memo said "cannot confirm Orbis sold — remove from bear case." The review correctly identified this as flawed reasoning. Absence of a Form 604 filing does not mean absence of selling — ASX disclosure triggers at 1% incremental changes, and gradual selling stays below threshold.

Corrected framing: Orbis reduction from ≈19% to ≈15% is probable based on secondary sources (ainvest.com, MarketScreener) but unverifiable in exact magnitude from primary filings. No Form 604 found in 2025-2026 to confirm or deny.

Weight as: mild negative signal, not confirmed, not removed. The most sophisticated active holder in the register appears to be reducing — but we can't prove the pace or current level. Meanwhile, Pinnacle entering at 5.01% (Dec 2025, ~A$20-21) is routine portfolio construction for an Australian fund manager, not a conviction signal that offsets a potential Orbis exit.

Recommendation

NO POSITION at A$22.66. Stock trades 53% above base SOTP and 18% above bull SOTP.

Watchlist entry: A$14-15. At this level:

• True P/E: 14-15x on FY26 sustainable earnings (after stripping recurring "significant items")
• DDR4 harvesting window (A$340M PV over 5 years) is free optionality
• Aligns with CEO's open-market purchase price (A$14.85, May 2025)
• Margin of safety covers: SK Tes competition (unresolvable), statutory gap persistence, DDR4 duration uncertainty

What would change this at higher prices:

• Nashville reveals binding multi-year hyperscaler contract (converts SLS from "relationship" to "contracted revenue" — justifies 15x+ ITAD multiple)
• CEO buys >A$500K at current prices (insider conviction at A$22, not A$15)
• Observable DDR4 supply shock beyond consensus (manufacturer bankruptcy, fab fire)
• Verifiable SK Tes failure in DDR4 repurposing specifically (confirms Sims moat)

LR Signal: 0.85 — Bearish. Market is overpricing SGM by relying on "underlying" earnings that mask A$83M/yr in recurring charges, and by capitalizing DDR4-era SLS margins as if they're permanent. SOTP analysis using true sustainable earnings shows A$14.82 fair value vs A$22.66 market price. The DDR4 thesis is correct and the scarcity window is real, but it's finite (2025-2030), consensus, and already reflected in a +62% move.