AngloGold Ashanti AU

Reference table

1. The disclosure gap — read this before anything else

Normally a gold-miner dive opens with all-in sustaining cost per ounce, reserve life, and production guidance. This dataset contains none of them.

What is missing, explicitly:

• No all-in sustaining cost (AISC). Not per ounce, not by mine, not in aggregate.
• No reserve or resource statement. No ounces in the ground, no mine life, no grade.
• No production volumes. No ounces produced, in any period.
• No cost guidance of any kind.
• seg_geo is four years stale. The most recent geographic revenue breakdown is FY2022 — before the FY2023-FY2025 period in which revenue more than doubled.
• seg_prod is nearly useless. It contains one line — FY2025 "Spot Revenue: $9.610B" — which is genuinely informative for one purpose (see section 3) and tells you nothing about mines or products.

The crude proxy we can build, clearly labelled as a proxy and NOT as AISC. FY2025 cost of revenue was revenue minus gross profit: $9.893B - $4.604B = $5.289B. Add FY2025 capex of $1.607B and you get $6.896B of combined operating and capital outflow against $9.893B of revenue — 69.7%, implying an all-in margin on revenue of roughly 30.3%.

That number is not AISC. AISC is a per-ounce industry-standard measure that includes sustaining capital, corporate general and administrative costs, exploration and reclamation, and excludes growth capital. Our proxy lumps growth capital in with sustaining, uses revenue rather than ounces as the denominator, and cannot be compared to any peer's published AISC. Use it only as a directional sanity check that the business is currently generating a wide all-in margin, and do not compare it to anything.

Why this matters for the verdict. A tracked claim from Money of Mine dated 2026-07-10 at conviction 62 states: "Consensus gold price forecasts run 10-20% above spot for FY27 and costs are undercooked; expects cost guidance in the $3,000+/oz range and a disappointing quarterly season for gold." That is a specific, dated warning about exactly the number we cannot see. We cannot check it against AngloGold's own disclosure from this dataset. That unresolved tension is the primary reason position size is capped.

2. The FY2025 operating-leverage event

The arithmetic that defines this business. From FY2023 to FY2025, revenue rose 116% and operating income rose 677%. Operating margin went from 12.5% to 45.1% — a 3,260 basis-point expansion in two years. That is not a management achievement in any ordinary sense; it is a commodity price meeting a fixed cost base.

Two honest asterisks on the headline growth. First, weighted diluted shares rose 17.9% from 430.92M (FY2024) to 507.90M (FY2025), so per-share growth is meaningfully below aggregate growth. The dataset does not explain the increase — no acquisition is identified in the FY2025 cash-flow statement (the acquisitionsNet line is a positive $91M, i.e. an inflow) — so we flag the share expansion and do not attribute a cause. Second, minority interest is $1.825B, 22.6% of total equity, meaning a substantial slice of the asset base is shared with partners.

3. The 97% spot problem — no hedge book, no cushion

The single most useful line in the segment data: FY2025 "Spot Revenue: $9.610B" against total revenue of $9.893B. That is 97.1% of revenue sold at spot. The FY2024 and FY2023 figures were 92.5% and 92.0% respectively — so the spot share has been rising.

What that means, plainly: AngloGold has essentially no hedge book. It does not lock in forward gold prices. There is no cushion when the gold price falls and no cap when it rises.

This is the correct posture for a miner whose shareholders want gold-price exposure, and it is why the stock behaves as a leveraged proxy for the metal. It also means every earnings estimate in this dive is really a gold-price estimate wearing a disguise, and that the 8.1x FY2027E multiple is only "cheap" conditional on a gold price the dataset does not contain.

For contrast, the knowledge base carries a claim about a peer whose hedge book repayment is itself expected to drive a re-rating (Money of Mine, 2026-07-10, conviction 70, on Bellevue Gold) — a reminder that hedging status is a first-order variable in this sector and that AngloGold sits at the fully-unhedged extreme.

4. The reversal is already visible — Q2 2026

This is the most important recent data in the file and it cuts against the bull case.

A 4.1% revenue decline produced a 22.8% EPS decline. That is a 5.5x amplification, and it is the clearest possible demonstration of what a 45%-plus operating margin does when the price moves against you. Anyone underwriting this stock on the FY2025 margin should model what happens if revenue falls 15% rather than 4%.

Year-over-year the same quarter still looks excellent — Q2 2026 revenue was up 27.0%, operating income up 58.5%, EPS up 48.5% versus Q2 2025. First-half 2026 revenue is $6.340B versus $4.408B, plus 43.8%, and first-half EPS is $4.50 versus $2.20, plus 104.5%. Both readings are true. The sequential one is the leading indicator.

And the execution record is poor. Against the earnings calendar's consensus figures, AngloGold has missed EPS in four of its last five reported quarters:

Revenue also came in below consensus in each of the four most recent reports for which both figures exist. (Note: the earnings-calendar EPS of $1.98 for Q2 2026 differs slightly from the income statement's diluted EPS of $1.96, a basis difference we have not reconciled; the comparison above uses the calendar's own figures throughout for consistency.)

5. The balance sheet and cash generation — the genuine strength

• Cash $2.928B (2025-12-31) versus total debt $2.436B → NET CASH of $492M. The reported enterprise value of $40.283B is below the $41.274B market cap, which is the mechanical confirmation.
• TTM net debt to EBITDA: -0.139x. Interest coverage 30.1x. Current ratio 2.71x. Debt-to-equity 0.20x.
• FY2025 cash flow: operating cash flow $4.712B, capex -$1.607B, free cash flow $3.105B — a 117.9% conversion of $2.636B of net income. Free cash flow yield 10.3%.
• Property, plant and equipment $8.702B (from $4.561B at FY2023). Goodwill zero. Intangibles $106M. This is a hard-asset balance sheet with essentially no acquired-intangible padding — the opposite profile to the serial acquirers elsewhere in this batch.
• Retained earnings turned positive — from negative $2.148B (FY2023) to negative $1.316B (FY2024) to positive $76M (FY2025). The company has only just worked off its accumulated deficit.
• Returns (TTM): ROE 45.8%, ROIC 31.3%, ROCE 42.9%, return on assets 24.5%. Extraordinary — and cyclical-peak figures that should not be extrapolated.

The dividend, and why the headline yield is misleading. FY2025 dividends paid were $1.871B, up 7.7x from FY2024's $244M. Trailing dividend per share is $4.60, giving a 5.64% yield at $81.61. The trailing payout ratio is 81.0% of net income (and 60.3% of free cash flow).

An 81% payout on peak-cycle earnings from an unhedged commodity producer is not a floor — it is a distribution of a windfall. If the gold price falls materially, that dividend falls with it. Do not underwrite the 5.6% yield as income. It is a variable return of a variable profit, and treating it as a bond-like coupon is the most common way retail investors get hurt in this sector.

6. Valuation — priced in or room?

At $81.61 (market cap $41.274B, net cash $492M, EV $40.283B):

Other trailing marks: EV/EBITDA 5.67x, EV/Sales 3.41x, P/B 4.63x, P/S 3.49x, P/FCF 9.71x, earnings yield 9.17%, free cash flow yield 10.3%.

Three data-quality warnings on the estimate block, and they matter:

1. Coverage is dangerously thin. FY2026E EPS rests on two analysts; FY2029E and FY2030E on one each. These are not consensus figures in any meaningful sense. Only the revenue lines (5-8 analysts) carry reasonable breadth.

2. The dispersion is enormous. FY2027E EPS ranges $7.11 to $11.75 — a 65% spread. FY2028E revenue ranges $10.649B to $20.239B.

3. The consensus EBITDA and EBIT lines are internally inconsistent and should not be used. FY2026E consensus shows net income of $4.805B exceeding EBITDA of $4.654B, which is arithmetically impossible for a taxpaying company with $1.015B of annual depreciation. Separately, the implied TTM EBITDA from the EV/EBITDA ratio is roughly $7.10B, making the FY2026E consensus EBITDA of $4.654B some 34% below the trailing run-rate — which does not square with consensus revenue rising 32%. We have used only the revenue and EPS lines from this block and discarded EBITDA and EBIT entirely.

Street position: consensus PT $95.20 (+16.7%), median $110 (+34.8%), high $128, low $42. That $42-to-$128 span is a three-fold range and is itself the most honest description of this security: the analyst community does not agree on what it is worth within a factor of three, because they do not agree on the gold price. Ratings are 9 buy, 3 hold, 2 sell — the only sell ratings anywhere in this batch. FMP's quantitative rating is A-, overall 4, with ROE and ROA sub-scores of 5.

Synthos fair values:

• Bear ~$70 — 7x FY2028E consensus EPS ($9.999). Gets there if gold mean-reverts, if the Money of Mine cost warning proves right and cost guidance lands in the $3,000-plus per ounce range, and if the operating margin keeps compressing at the Q2 2026 rate. -14.2%.
• Base ~$91 — 9x FY2027E consensus EPS ($10.114) = $91.02. Nine times sits below the 10.9x trailing multiple, which is the appropriate discount for a business whose own consensus has revenue peaking in FY2028. Two independent cross-checks land in the same place: the 200-day average is $91.20 and the Street consensus target is $95.20. Three anchors within 4.6% of each other is unusually tight for a commodity name. +11.5%.
• Bull ~$121 — 12x FY2027E consensus EPS ($10.114) = $121.36. Requires gold to hold near the roughly $4,000 level that tracked voices reference, the second half of 2026 to deliver, and a re-rating toward the multiple a net-cash 45%-margin producer arguably deserves. Brackets the Street median of $110 and approaches the high of $128 and the 52-week high of $128.26. +48.7%.

6a. What today's price assumes (the inversion)

At $81.61 (8.5x FY2026E, 8.1x FY2027E consensus EPS), today's price assumes roughly:

• Second-half 2026 revenue of $6.762B, up 6.7% on the $6.340B first half, and second-half EPS of $5.09, up 13.0% on the $4.50 first half. (Falsifiable at the 2026-11-10 Q3 print, consensus $2.922B and $2.00.) This is the single most fragile assumption in the price. Q2 2026 EPS just fell 22.8% sequentially, and the Q3 consensus itself models revenue below Q2's $3.104B — so the required second-half acceleration must land almost entirely in Q4. Consensus-derived, and note only two analysts carry the FY2026 EPS figure.
• The gold price holds near current levels through FY2028. Consensus revenue of $13.9B-$14.4B in FY2027-FY2028 against $9.893B in FY2025 is largely a price assumption, since the dataset gives no production growth data. (Falsifiable directly in the gold price — which this dataset does not contain, a limitation we are flagging rather than filling.) Money of Mine, 2026-07-10, conviction 62 — "Consensus gold price forecasts run 10-20% above spot for FY27."
• The operating margin stabilises somewhere near 45-50% rather than continuing the Q1-to-Q2 slide from 56.8% to 51.0%. (Falsifiable: quarterly operating margin.)
• Costs do not inflate materially. (Falsifiable — but NOT from this dataset, which contains no AISC. This assumption is embedded in the price and unverifiable here.)
• The market keeps paying roughly 8x forward earnings and does not de-rate toward the 7x that a business with consensus revenue declining after FY2028 might otherwise attract. (Falsifiable in the price.)
• The FY2029-FY2030 consensus decline is a gold-price assumption, not depletion. If it is depletion, the terminal value is far lower than 8x forward earnings implies. (NOT falsifiable from this dataset — there is no reserve statement. This is the deepest unresolved question in the entire dive.)

6b. The return bridge (why the multiple moves)

expected return ≈ EPS growth + multiple drift + shareholder yield

The decomposition for a commodity producer is honest only if you say plainly that the "EPS growth" term is really a price forecast. Over a two-year horizon to FY2027E, consensus EPS goes from $9.59 to $10.11 — just 5.4% total, or 2.7% a year — because consensus already assumes the gold-price windfall is banked and does not repeat. The trailing dividend yield is 5.64%, but at an 81.0% payout on peak earnings it is not a durable component and we discount it heavily.

So our base case assumes essentially NO earnings growth and a modest multiple RE-RATING — from today's 8.1x FY2027E to 9x. We want to be explicit about why, because it is unusual for us to lean on multiple expansion:

• The stock is 36.4% below its high and below both moving averages, having already de-rated. The 10.9x trailing multiple is the pre-drawdown mark.
• A producer with net cash, 30x interest coverage, 45% operating margins and 118% free-cash-flow conversion arguably deserves better than 8x, and the market paid better than that within the last year.
• But we cap the re-rating at 9x, not 11x or 12x, precisely because consensus has revenue declining after FY2028 and because we cannot see the reserve data that would justify a longer duration.

The bridge to $91 is therefore: roughly 0% earnings growth, plus about 11% of multiple re-rating, plus a discounted dividend. That makes multiple expansion the fragile leg of this thesis, and we are flagging it as such. If the multiple simply holds at 8.1x, the stock is worth about $82 — which is where it trades.

6c. Variant perception (where we differ, what would surprise)

• We are more constructive than the two sell ratings and considerably less constructive than the $110 median target. Our $91 base sits below the Street's $95.20 consensus and well below its $110 median. Where we differ from the bulls: we do not believe the second-half 2026 acceleration consensus requires, because Q2 already fell 22.8% sequentially and the Q3 consensus itself models further decline. Where we differ from the bears: a net-cash producer generating a 10.3% free cash flow yield at a 36% drawdown is genuinely cheap, and the structural debasement lane in our knowledge base is deep, cross-corroborated and dated — Michael Howell at conviction 80 (fair value roughly $4,000), Luke Gromen at conviction 70, Ray Dalio at conviction 75, Raoul Pal at conviction 85 on the liquidity mechanism.
• Our genuine variant perception is about what the market is NOT pricing: the jurisdiction question. The last disclosed geographic mix is FY2022 and puts Tanzania ($920M), Ghana ($874M) and Guinea ($591M) at 54.4% of disclosed revenue. Against that, Money of Mine on 2026-08-01 at conviction 72 states: "Political/security risk has risen sharply; Mali is a no-fly zone, exited Burkina Faso and Ghana, left Mexico (cartels, riots) — hunker down in safest jurisdictions." A tracked mining specialist has exited Ghana. AngloGold's exposure there is, on four-year-old data, 19.4% of revenue. We think that risk is underweighted in an 8x multiple — and we also acknowledge that we cannot size it, because the disclosure is stale.
• Positive surprise that would force a repricing higher. A Q3 or Q4 2026 print showing operating margin stabilising above 50% with revenue at or above consensus, or any disclosure of AISC materially below the $3,000-per-ounce level the tracked caution anticipates, or reserve-life confirmation that the FY2029-FY2030 consensus decline is a price assumption rather than depletion. Any of the three would justify moving from 9x toward the 12x bull case and lifting the position from hedge-sleeve to core.
• Negative surprise that would force a repricing lower. A second consecutive 500-plus basis point sequential margin decline, cost guidance confirming the $3,000-plus warning, a fifth consecutive EPS miss, or a jurisdiction event in the African assets. Any of these opens the $70 bear case and, given Money of Mine's 2026-07-25 warning at conviction 65 about extraordinary volatility and "low-IQ new capital walking out the door" driving downside, likely overshoots it.

7. Knowledge base — a deep theme, an almost-empty name, and cautions from the same source

This is the most nuanced knowledge-base picture in the batch and it needs to be stated carefully.

At the NAME level: exactly ONE claim. kb_claim_count: 1, breadth 1.

• Money of Mine, 2026-05-15, conviction 35, bullish, thesis-horizon: "AngloGold's Nevada development project looks phenomenal; it needs to come into production before Anglo will part with Australian assets like Tropicana due to tier-one-jurisdiction credit-rating constraints."

That is it. One claim, at conviction 35 — low, about a development project rather than the current business, from a single source. This is not a conviction lane. It is a passing mention. We are not going to inflate it.

At the THEME level: one of the deepest lanes in the entire base. The gold and monetary-debasement theme is extensively covered, recent, and cross-corroborated across multiple independent tracked voices. Selected, dated, verbatim:

• Michael Howell, 2026-06-12, conviction 80: "Long-term gold goes substantially higher because world debt forces money printing; fair value ~$4,000/oz — picking it up substantially below $4,000 is doing well."
• Luke Gromen, 2026-07-23, conviction 70: "Base case for coming years: S&P up in dollar terms but down in gold terms — gold rises more than stocks; stocks already down in gold terms since 2000 and 2022."
• Ray Dalio, 2026-07-30, conviction 75: "Gold is a very effective diversifier — tends to do well when stocks, bonds and houses do badly; recommends 'hard money' at 5-15% of portfolio and prefers gold over Bitcoin."
• Raoul Pal, 2026-07-23, conviction 78: "Global liquidity is the dominant macro force; it rises ~8% a year to pay debt interest, forcing scarce assets up as the currency denominator is debased."
• Money of Mine, 2026-06-12, conviction 80: "Gold miners are being hit 30-40%, a level really outside fundamentals; planning to pull hedge profits and redeploy across favorite gold players — this metals/miners bull has a long way to go."
• Michael Howell, 2026-07-27, conviction 70: "At the liquidity-cycle peak you want much bigger commodity weightings; commodities should best most asset classes as money rotates from financial markets into the real economy."
• Crux Investor, 2026-07-29, conviction 35 (fact-tagged): "Host notes gold has pulled back to around $4,000, still a strong environment compared with one or two years ago."

Important framing. That last claim is the closest thing to a current gold price in our sources, and it is a tracked thinker's statement dated 2026-07-29, not a verified market quote. This dataset contains no commodity price data. Every gold-price reference in this dive is a citation of what a tracked voice said, labelled as such.

And now the part that most write-ups would omit: the cautions, from the very same source that named the company.

• Money of Mine, 2026-07-10, conviction 62, BEARISH: "Consensus gold price forecasts run 10-20% above spot for FY27 and costs are undercooked; expects cost guidance in the $3,000+/oz range and a disappointing quarterly season for gold."
• Money of Mine, 2026-08-01, conviction 72, BEARISH: "Political/security risk has risen sharply; Mali is a no-fly zone, exited Burkina Faso and Ghana, left Mexico (cartels, riots) — hunker down in safest jurisdictions; only West Africa holding is Teranga in Cote d'Ivoire."
• Money of Mine, 2026-07-25, conviction 65, BEARISH: "Propensity for violence is extraordinarily high — most volatile market he's seen; low-IQ new capital walking out the door will drive an incredible amount of downside violence even as quality stays cheap."
• Michael Howell, 2026-06-12, conviction 48, NEUTRAL: "Gold miners were attractive but shine comes off as gold falls; rising energy costs and a falling gold-oil ratio make them not a slam dunk — be cautious."
• Lyn Alden, 2026-06-21, conviction 58: "Gold had a powerful 2-year run and got over its skis near $5,000/oz; no longer cheap but not drastically overvalued."
• Money of Mine, 2026-07-10, conviction 30: "No strong view on the gold price; sees both the debasement/debt bull case and a more-hawkish-Fed bear case toward $2,000, but $4,000/oz remains an incredible margin level for producers."

How we read the whole lane. The metal is strongly supported at conviction 70-85 across four independent voices. The miners are not — the same specialists warn about undercooked costs, extreme equity volatility, and jurisdiction risk that specifically includes Ghana, where AngloGold's last disclosed mix carries 19.4% of revenue. And the single claim that names AngloGold does so at conviction 35.

That combination is exactly what a Low-Moderate conviction rating means, and it is exactly why the verdict is a small Stage-In rather than a large one. Own the theme; respect the specialists' caution on the vehicle.

8. Technicals — a broken trend without an exhaustion signal

• Price $81.61 (2026-08-03 close, +2.89% that session from $79.32). 52-week range $50.20-$129.14 (quote) / $48.76-$128.26 (tech).
• -36.4% from the 52-week high. The max_dd_from_peak reading matches at -36.4% — this is the full extent of the drawdown, and it is deep.
• Below both averages, both falling. 50-day $84.32 (price is 3.2% below); 200-day $91.20 (price is 10.5% below). The 50-DMA is below the 200-DMA. MACD -1.12. The intermediate and primary trends are both broken.
• RSI 52.1 — and this is the most important technical fact in the dive. The stock is not oversold. There is no downside-exhaustion signal to lean on. A washed-out reading in the 25-30 range would make this a far better tactical entry; a neutral 52 after a 36% decline means the selling has been orderly and can continue without any technical resistance.
• Relative performance: -12.0% (3-month), -12.1% (6-month), +76.5% (12-month) against SPY at +5.1% / +9.5% / +19.9%. So AngloGold massively outperformed over the year and has been giving it back for six months. It is still +67.4% above its 52-week low.

Read: this is a broken uptrend in the middle of its correction, not a bottom. That is precisely why the entry is staged in three tranches rather than taken at once — the 200-DMA at $91.20 is now overhead resistance, and there is a very long way down to the 52-week low at $48.76 if the correction extends.

Insider activity is uninformative: six director share awards on 2026-05-11 (1,580-2,371 units each, at zero cost) and one small director sale of 400 shares at $89.09 on 2026-05-19. Nothing to read.

9. Verdict, kill-criteria and the flip conditions

Stage-In — small, hedge-sleeve sizing (1-2% maximum), built in three tranches.

The case for buying: AngloGold is a net-cash senior gold producer with 30.1x interest coverage, a 45.1% operating margin, 118% free-cash-flow conversion, a 10.3% free cash flow yield, and a 45.8% return on equity, trading at 8.1x FY2027E consensus earnings 36.4% below its 52-week high. Three independent anchors — 9x FY2027E consensus EPS at $91.02, the 200-day average at $91.20, and the Street consensus target at $95.20 — converge within 4.6% of each other at roughly $91, about 11.5% above spot. And the structural debasement thesis behind gold is the deepest, most cross-corroborated macro lane in our knowledge base.

The case for keeping it small is threefold and each part is disqualifying at full size:

1. The reversal has already started. Q2 2026 revenue fell 4.1% sequentially and EPS fell 22.8%; the operating margin lost 583 basis points in one quarter; consensus for Q3 models revenue and EPS below Q2; and the company has missed EPS in four of its last five quarters.

2. We cannot see the numbers that decide a mining call. No AISC, no reserves, no production ounces, and a geographic disclosure that is four years stale. A tracked specialist has warned at conviction 62 that industry cost guidance is undercooked — and we cannot check it.

3. The knowledge base does not underwrite this name. One claim, at conviction 35, about a development project. The same source supplies conviction-62 and conviction-72 cautions that cut directly against the company, one of them naming Ghana specifically.

Staged entry (build a 1-2% hedge-sleeve position in three tranches):

1. Tranche 1 — now, roughly one-third, at ~$81.61. A 36% drawdown, net cash, a 10.3% free cash flow yield, and 8.1x forward earnings is enough to start. Underwrite the balance sheet and the macro lane, not the second-half consensus.

2. Tranche 2 — on either a genuinely oversold reading (RSI below 35) OR the 2026-11-10 Q3 print showing the operating margin stabilising above 50%. One is a better price, the other is better information; take whichever arrives first.

3. Tranche 3 — only after AISC and reserve data are in hand, or on an overshoot toward the $60-$70 zone with the balance sheet intact.

Pre-registered kill / do-not-add criteria:

• A second consecutive 500-plus basis point sequential decline in operating margin. The Q2 compression was the start, not an aberration.
• Cost guidance landing at or above $3,000 per ounce, confirming the Money of Mine caution at conviction 62 — at which point the entire margin structure is re-based and the 8x multiple is not cheap.
• A fifth consecutive EPS miss at the 2026-11-10 print. Four in five is a pattern; five in six is a management problem.
• Any jurisdiction event in the African asset base — Tanzania, Ghana or Guinea — given a conviction-72 tracked claim that a specialist has already exited Ghana.
• The dividend being cut without a corresponding gold-price move, which would mean cash generation is weaker than the reported free cash flow suggests.
• Evidence that the FY2029-FY2030 consensus revenue decline reflects depletion rather than a gold-price assumption. That changes the terminal value fundamentally and would make 8x forward earnings expensive rather than cheap.

Pre-registered flip to a core position (upsize toward 3-4%):

• Disclosure of AISC materially below the $3,000-per-ounce warning level, plus a reserve statement supporting a mine life that contradicts the FY2029-FY2030 consensus decline. That single combination removes the largest cap on size in this dive.
• AND operating margin stabilising above 50% across two consecutive quarters with revenue at or above consensus.
• AND a refreshed geographic disclosure showing jurisdiction concentration has fallen from the FY2022 mix, or the Nevada project — the subject of the only claim naming this company — advancing toward production.

Where AU fits in the Synthos Framework Portfolio. The macro-hedge / hard-asset sleeve, alongside rather than inside the equity book. The correct mental model is insurance with a positive carry: a leveraged, cash-generative claim on the debasement thesis that several independent tracked voices underwrite at conviction 70-85, sized the way you size insurance (1-2%) rather than the way you size a conviction equity. Ray Dalio's conviction-75 framing — hard money at 5-15% of a portfolio — is the right sleeve-level guide, with AngloGold being only one, higher-beta, way to fill part of it. Explicitly do not size this on the dividend. Logged as a tracked Synthos call (Stage-In, hedge-sleeve) as of 2026-08-04 at $81.61.

Single biggest risk: operating leverage running in reverse. A 45% operating margin means a 10% fall in the gold price does far more than 10% of damage to earnings — Q2 2026 already showed a 4.1% revenue decline producing a 22.8% EPS decline, a 5.5x amplification. With 97.1% of revenue sold at spot and no hedge book, there is nothing between the metal price and the income statement.

Single most fragile assumption in the price: that the second half of 2026 delivers revenue of $6.762B (up 6.7% on the first half) and EPS of $5.09 (up 13.0%), as consensus requires — when the most recent quarter fell 22.8% sequentially and the Street's own Q3 estimate ($2.922B revenue, $2.00 EPS) is below what Q2 actually printed. The entire required acceleration has to arrive in a single quarter, on an EPS consensus carried by two analysts.

Provenance and disclosures

• Traceability: Exactly ONE knowledge-base claim names AngloGold Ashanti — Money of Mine, 2026-05-15, conviction 35, bullish, on the Nevada development project and tier-one-jurisdiction credit-rating constraints. kb_claim_count: 1, breadth 1. That is not a conviction lane and this dive does not treat it as one. The gold and debasement theme lane is deep and cross-corroborated (Michael Howell 2026-06-12 conviction 80; Luke Gromen 2026-07-23 conviction 70; Ray Dalio 2026-07-30 conviction 75; Raoul Pal 2026-07-23 conviction 78; Money of Mine 2026-06-12 conviction 80; Michael Howell 2026-07-27 conviction 70) and is cited throughout as theme-level, not name-level. Four cautionary claims are given equal prominence in section 7 — Money of Mine 2026-07-10 conviction 62 (undercooked costs, $3,000-plus guidance), Money of Mine 2026-08-01 conviction 72 (West African jurisdiction risk, exited Ghana), Money of Mine 2026-07-25 conviction 65 (extreme miner volatility), Michael Howell 2026-06-12 conviction 48 (miners not a slam dunk).
• Data as-of: prices and quote 2026-08-03 close (FMP quote timestamp 2026-08-03 20:00 UTC) · income statement through Q2 2026 (2026-06-30, filed 2026-07-31) · annual balance sheet and cash flow FY2025 (2025-12-31, filed 2026-03-26) · product-segment data FY2025 · geographic-segment data FY2022 — four years stale · analyst estimates, price targets, grades and earnings calendar last updated 2026-08-04 · knowledge-base search 2026-08-04.
• CRITICAL disclosure gap — read before using this dive: the dataset contains no all-in sustaining cost (AISC), no reserve or resource statement, no production ounces, and no cost guidance. For a mining company these are the three or four figures that normally decide the call. Section 1 constructs a crude all-in-margin-on-revenue proxy (30.3% for FY2025) and explicitly labels it as not AISC and not peer-comparable. Position size in this dive is capped principally because of this gap. The dataset also contains no gold price; every gold-price reference here is a citation of a dated tracked-thinker statement, labelled as such.
• Stale geographic disclosure: the most recent seg_geo row is FY2022, showing Australia $967M, Tanzania $920M, Ghana $874M, Brazil $717M, Guinea $591M and Argentina $319M against $4.501B of FY2022 revenue. On that mix, Tanzania plus Ghana plus Guinea was 54.4% of disclosed geographic revenue and Ghana alone 19.4%. FY2025 revenue is 2.2x the FY2022 level, so the current mix is unknown and all jurisdiction commentary here is explicitly based on four-year-old data.
• Estimate-block integrity warning: the consensus EBITDA and EBIT lines in this dataset are internally inconsistent and have been discarded. FY2026E shows net income of $4.805B exceeding EBITDA of $4.654B, which is impossible for a taxpaying company with $1.015B of annual depreciation; and the FY2026E EBITDA figure sits roughly 34% below the trailing run-rate implied by the TTM EV/EBITDA ratio while revenue is modelled up 32%. Only the revenue and EPS lines have been used. Coverage is also thin: 2 EPS analysts on FY2026E, 4 on FY2027E, 3 on FY2028E, 1 each on FY2029E and FY2030E. Published price targets span $42 to $128.
• Share-count note: weighted diluted shares rose 17.9% from 430.92M (FY2024) to 507.90M (FY2025). The dataset does not explain the increase — the FY2025 acquisitionsNet line is a positive $91M inflow and commonStockIssuance is zero — so we have flagged the dilution and not attributed a cause.
• Minor reconciliation note: the earnings calendar reports Q2 2026 actual EPS of $1.98 while the income statement reports diluted EPS of $1.96. The beat/miss table in section 4 uses the calendar's own figures on both sides for internal consistency; the difference is not material to any conclusion.
• Dividend caveat, stated in the body and repeated here: the 5.64% trailing yield rests on FY2025 dividends of $1.871B, up 7.7x year over year, at an 81.0% payout of peak-cycle net income. It is a variable distribution of a commodity windfall, not a contracted coupon, and this dive explicitly instructs readers not to size the position on it.
• Fair-value method: scenario multiples on consensus EPS — bear 7x FY2028E ($9.999), base 9x FY2027E ($10.114), bull 12x FY2027E ($10.114). Stated arithmetic, not a discounted cash flow. The base is cross-checked against two independent anchors (the $91.20 200-day average and the $95.20 Street consensus target).
• Not investment advice. Independent research, educational and informational only, never personalised.
• Version: 2026-08-04-full.