Gold Streaming Companies
Business Model Explained

A streaming deal has three numbers that matter: how much the streaming company pays upfront, how much it pays per ounce on each delivery, and what percentage of the mine’s output it gets to buy.

Wheaton Precious Metals’ agreement on HudBay Minerals’ Constancia copper mine, signed in 2013, is a good one to look at because the parameters are clean and public. Wheaton paid $750 million in installments for the right to buy byproduct silver at about $5.90 per ounce and byproduct gold at about $400 per ounce. In Wheaton’s 2024 annual report, the company disclosed its average per-ounce payments across the portfolio: $440 for gold, $4.98 for silver. Those are the ongoing costs. With gold averaging above $2,400 for most of 2024 and silver above $28, the spread is enormous.

Contract duration usually binds to the life of the mine, not to some fixed term. This matters more than people realize. If the mine lasts longer than originally estimated, the streaming company keeps buying at the original terms. The upfront payment, which was calculated based on an expected mine life and reserve base, gets diluted across more ounces. Per-unit cost drops without any additional outlay.

The Constancia deal wasn’t a deal for a gold mine. It was a deal for the byproduct precious metals coming out of a copper mine. This distinction explains more about streaming company economics than anything else in the model.

A copper mine’s CFO models revenue around copper. Gold and silver that come out in the process are supplementary. The byproduct precious metals might account for 10% to 15% of total mine revenue, sometimes less. In a five-year capital allocation plan, the price of byproduct gold doesn’t move the needle on whether the project gets greenlit or whether the mine stays open. Copper does that.

When a streaming company shows up offering hundreds of millions of dollars for rights to those byproduct metals, the copper company’s incentive structure pushes toward a deal. The metals aren’t central to the business plan. Giving them up at a steep discount in exchange for immediate development capital is an easy sell to the board. The streaming company ends up with a gold and silver purchase agreement priced as though the metals were an afterthought, because to the seller, they were.

Wheaton’s entire portfolio leans into this dynamic. A large share of its production comes from streams on mines operated by Glencore, Vale, Newmont, and similar operators where the primary commodity is something other than gold or silver. These are among the largest, best-capitalized mining companies on the planet. Counterparty risk is about as low as it gets in the mining sector.

But here’s the part that most valuation models handle badly. When an analyst runs a gold price sensitivity test on a streaming portfolio, the standard approach is to cut the assumed gold price by 20% or 30% and see what happens to cash flows. The problem is that cutting gold prices by 30% affects different assets in the portfolio in fundamentally different ways.

A pure gold mine has a breakeven point. If gold drops below the mine’s all-in sustaining cost, the operator may curtail or shut down production. Deliveries to the streaming company go to zero. The upfront payment is impaired.

A byproduct stream on a copper mine doesn’t work like that. Gold production at a copper mine is incidental to copper production. If gold drops 30% but copper holds steady, the mine stays open, gold keeps being extracted as a byproduct, and deliveries to the streaming company continue on schedule. The streaming company makes less money per ounce, sure, but it still gets its ounces. The floor on deliveries is set by the copper price, not the gold price. The two commodities don’t even have to be correlated.

This cross-commodity hedge is probably the most structurally important feature of the business model, and I’m not confident it’s priced correctly. Sell-side NAV models typically assign the same production sensitivity to byproduct assets and primary assets. The ones I’ve seen don’t bifurcate the analysis.

Franco-Nevada had 46 full-time employees at the end of 2024. Forty-six. The company generated $1.11 billion in revenue that year and deployed over $1.3 billion in new acquisitions. Revenue per employee was north of $24 million.

The cost structure of a streaming company is unlike anything else in the resource sector. There is no mine to operate. No processing plant. No tailings dam to monitor. No fleet of haul trucks. No permitting battles with local regulators. No workforce of thousands. The headquarters is an office in downtown Vancouver or Toronto with geologists, lawyers, accountants, and investor relations people. That’s the whole operation.

New deals don’t require new infrastructure or new headcount. The marginal cost of adding a streaming asset to the portfolio is close to zero on the operating side. This is why operating margins for the top companies run so high and why the margins expand as the portfolio grows. Wheaton’s G&A of $40.7 million is going up to an estimated $50 to $55 million in 2025, but that’s still negligible against a revenue base that’s growing faster.

A royalty holder gets a percentage of mine revenue or a percentage of the value of metal produced, and that’s it. No ongoing payment per ounce. Pure upside. An NSR (net smelter return) royalty of 2% means the holder receives 2% of what the mine earns after deducting smelting and refining costs, with no further obligation.

A streaming company pays an upfront deposit and then pays again on every delivery. The ongoing payment is well below market price, but it’s still a cost. Streaming involves higher capital outlay upfront and lower margins per ounce relative to a royalty, but because of the ongoing payment structure, mining companies can raise significantly more capital through a stream than through a royalty on the same asset. For a mid-tier miner trying to raise $500 million to build a mine, the difference between what a royalty and a stream can deliver may determine whether the project gets financed at all.

Mining projects have a structural timing problem. Discovery to production: seven to fifteen years. During that entire period, the company is spending money, every dollar of exploration, feasibility studies, environmental assessment, road building, power line construction, processing plant design, and community relations, against zero revenue. Bank project finance requires demonstration of cash flow to service debt. A mine under construction doesn’t have cash flow. High-yield bonds are an option for companies with the credit profile to issue them, but they come loaded with financial covenants that restrict operational flexibility, and the interest rates for sub-investment-grade mining companies can be punishing. Equity issuance dilutes existing shareholders and is functionally impossible during sector downturns when mining stocks are already beaten down.

A streaming upfront payment doesn’t show up as debt on the balance sheet. No scheduled repayments. No covenants. No dilution. It’s expensive capital, if you back out the implied cost over the life of the mine it can dwarf conventional loan rates, but it’s capital that’s available when nothing else is.

This is where it gets interesting.

Franco-Nevada routes its streaming contracts through Franco-Nevada (Barbados) Corporation. The metal gets purchased and sold through this entity. Profits are booked in Barbados, not Canada, and are taxed at Barbados rates. This isn’t hidden. It’s in the annual report, in the SEC filings, in the investor presentations if you read the footnotes.

The settlement came in December 2018. Wheaton agreed to increase the service fee markup charged to its international subsidiary from 20% to 30%, which increased the taxable income attributed to Canada going forward. Transfer pricing penalties were reversed. No additional cash taxes for the 2005–2010 years after applying loss carryforwards. The stock jumped 14% on the announcement.

Here’s the part that matters for investors today: Wheaton’s 2024 annual report discloses a $35 million expense related to the global minimum tax (GMT), which is the OECD’s Pillar Two framework starting to take effect. This is new. It wasn’t in the cost base two years ago. And it’s going to get bigger as more jurisdictions adopt the 15% minimum. Franco-Nevada is also under CRA audit for its Mexican revenue arrangements. The pattern across the industry is clear: the tax efficiency that has been a structural contributor to streaming company margins is under pressure, and the direction of travel is toward higher effective rates, not lower ones.

Most sell-side models I’ve seen extrapolate historical effective tax rates into their forward estimates. Almost nobody stress-tests a scenario where the effective rate rises by 500 or 800 basis points. Given what’s happening at the OECD level, this strikes me as a gap.

Most streaming contracts specify a fixed percentage of production over the life of the mine. Not a fixed number of ounces. If the mine’s reserves grow, production increases, and the streaming company’s entitled volume grows proportionally. The price terms don’t change.

The mining company funds the exploration. The mining company bears the geological risk. If the exploration works and reserves expand, the streaming company’s upfront payment is now spread across a larger production base without any additional outlay.

Franco-Nevada’s early history is the most extreme example of this. Pierre Lassonde and Seymour Schulich acquired royalty interests on Nevada gold belt properties in the 1980s at negligible cost. Over the following three decades, other companies spent hundreds of millions on exploration in the district. The covered resource base expanded dramatically. Franco-Nevada contributed nothing to that exploration but captured the entire upside proportionally. The company was later absorbed by Newmont and re-IPO’d in 2007. Since the re-listing, it has outperformed gold, the GDX gold miners index, and for long stretches, the S&P 500.

When streaming companies evaluate new deals, the geological expansion potential of the district carries considerable weight, maybe more weight than the currently defined reserves. A mine with 15 years of reserves in a geologically promising belt, where brownfield exploration has a reasonable probability of extending mine life to 25 or 30 years, embeds an option value that doesn’t show up in a standard DCF but absolutely shows up in the returns a decade later.

Franco-Nevada, Wheaton Precious Metals, Royal Gold. There’s no regulatory barrier to entry. No license required. A sovereign wealth fund with $10 billion in deployable capital could theoretically go to a mining company tomorrow and offer to do a streaming deal. None of them have, at least not at scale.

The evaluation capability is one reason. Pricing a streaming deal requires a team that can independently assess geological models, metallurgical recoveries, cost assumptions, political risk, environmental liabilities, and contract enforceability across dozens of jurisdictions. This is a specialized skill set that takes a long time to assemble. You can hire individual geologists and mining engineers, but institutional knowledge about how streaming deals behave over full mine cycles takes decades to accumulate.

The bigger barrier is trust. A streaming deal is often the most consequential financing decision a mining company will ever make. The amounts are large, the terms are permanent, and the timeline is tight because mining projects have construction windows that can’t slip without cascading cost overruns. The mining company’s management team needs to be confident that the streaming counterparty will close. If the deal falls apart at the last stage because of committee indecision or a change of heart at the fund level, the mining company may lose its construction window entirely. Franco-Nevada and Wheaton have closed hundreds of these transactions. Their reputation for executing is a competitive asset that no amount of capital can instantly replicate.

The 2013–2015 gold bear market made this worse for potential new entrants. Gold went from near $1,700 to around $1,100. Mining sector financing dried up almost completely. The only buyers with balance sheet capacity to do large deals during that window were the established streaming companies. Franco-Nevada, which has maintained a zero-debt balance sheet for its entire post-IPO history, was able to deploy aggressively. The deals signed at the bottom of that cycle turned out to be the highest-returning assets in the portfolio. A new entrant who wasn’t already established by 2012 missed the best vintage in the industry’s history.

Whether this oligopoly structure is permanent is debatable. A prolonged period of cheap capital and high gold prices could bring in competition. The evaluation and relationship moats would still hold, but the pricing advantage might compress. It hasn’t happened yet.

The Cobre Panama situation is the clearest current illustration of what can go wrong. Franco-Nevada holds a gold and silver stream on First Quantum Minerals’ Cobre Panama copper mine, one of the largest copper mines in the world. In November 2023, Panama’s Supreme Court ruled the mine’s operating contract unconstitutional following large-scale public protests. The mine went into preservation and safe management, which is a polite way of saying it stopped producing. Franco-Nevada’s stream deliveries went to zero. The company has filed for international arbitration. As of the most recent filings, Panama’s new president has indicated willingness to discuss the situation, but there is no resolution and no timeline.

This is a single asset, and it hit Franco-Nevada’s production hard. The company reported that Q4 2024 revenue was up 30% year-over-year excluding Cobre Panama. In other words, the rest of the portfolio performed well, but the Cobre Panama shutdown was large enough to be a standalone disclosure item. For a company that prides itself on diversification and low risk, losing a top-tier asset to a political event that was essentially unforeseeable in its timing and severity is a meaningful data point about the limits of the model.

Tanzania’s 2017 action against Acacia Mining is another instructive case. The Tanzanian government presented a $190 billion tax claim, a number so absurd it was clearly a negotiating tactic, but one that demonstrated how far a host government is willing to go when it decides the terms of resource extraction are no longer acceptable. The claim was eventually negotiated down to a restructuring of the ownership and economics of the mine, but the process destroyed enormous shareholder value along the way. Streaming companies with exposure to Tanzanian assets had to mark time while the political situation resolved.

Delivery risk is more mundane. Mines shut down because of geological problems, tailings failures, labor disputes, water shortages, or simply because the ore body didn’t behave the way the feasibility study predicted. Make-whole provisions in contracts are supposed to provide compensation when production falls short, but a mining company facing a major operational failure is usually also facing financial stress. The paper obligation and the ability to pay may not coincide.

On the accounting: streaming assets are amortized using the unit-of-production method. This creates large non-cash charges that depress GAAP net income. Both companies push investors toward adjusted cash flow and adjusted earnings, which strip out the amortization. The definitions of these non-GAAP metrics are not standardized across companies. Franco-Nevada and Wheaton may exclude different items from their respective “adjusted” numbers. Comparing them at face value without reading the footnotes is unreliable.

NAV is the standard. Model each streaming and royalty asset as a standalone DCF, sum the net present values, add cash and undeployed capital. It’s a workable first approximation.

The weakness is that it’s linear. A DCF with a single gold price path can’t capture the convexity of a streaming portfolio. The return profile is asymmetric: gold going up by $500 expands margins by more than gold going down by $500 compresses them, because the ongoing payment is fixed. In a higher-volatility environment, that asymmetry has more value. There’s no volatility input in a NAV model.

Both companies trade at persistent premiums to NAV. Sell-side analysts call this a management premium or platform premium, which is more of a label than an explanation. The premium probably reflects some combination of expected future deal-making success, an imprecise market attempt to price the convexity, and simple scarcity, there being very few investable pure-play streaming companies globally. I’m not sure these components can be cleanly separated.

One thing that is analytically tractable: the distinction between organic growth and acquisition-dependent growth. Organic growth comes from production ramp-ups at existing mines, reserve expansion, and new mine starts on assets already in the portfolio. The marginal cost is near zero. All of it flows to free cash flow. Acquisition growth requires new upfront capital. At the same headline revenue growth rate, a portfolio tilted toward organic sources has materially higher free cash flow quality.

Bullish on gold: buy the metal or an ETF. Want operating leverage: buy gold miners. Streaming companies offer leveraged gold exposure plus volume growth from mine production increases, without the operational risk that comes with actually running a mine. That combination doesn’t exist elsewhere in a single equity.

They sit in an odd spot for institutional allocators. Precious metals bucket, so they benefit from sector inflows during gold rallies. High margin, stable cash flow, minimal debt, so they also attract quality-growth fund managers who would never own a cyclical mining stock. The overlap in demand from different investor categories supports both liquidity and valuation.

Large mining companies sometimes hold streaming and royalty shares as a portfolio hedge. Gold up: the miner benefits directly and streaming stocks appreciate. Gold down: the miner takes a hit but streaming stocks fall less due to lower operating leverage, partially cushioning the decline. This isn’t widely discussed in the institutional ownership breakdowns but it’s a real source of demand.

The 2013–2015 period is the reference cycle. Gold in sustained decline, mining sector capital markets frozen. Franco-Nevada and Wheaton accelerated deal-making instead of retrenching. Contracts signed at the bottom carried the most favorable terms in the industry’s history: lowest upfront-to-production ratios, deepest ongoing payment discounts. Those assets became the highest-returning portions of the portfolio from 2016 onward.

Doing this requires balance sheet capacity and management conviction. Franco-Nevada’s zero-debt policy is maintained for exactly this purpose. Whether any given management team will actually pull the trigger during the next downturn, when sentiment is terrible and every headline says gold is finished, is a separate question. Having dry powder and being willing to use it are not the same thing.

Copper, cobalt, and lithium streaming deals have started appearing. The economics are structurally thinner. Gold has a safe-haven bid and a credible long-term purchasing power argument. Copper and lithium are industrial commodities driven by demand cycles. Ongoing payment prices for base metal streams have to be set at a higher percentage of market price, leaving less room. The crypto-adjacent mining finance experiments of the past few years suggest there’s interest in alternative funding structures, even if nothing has scaled yet. Tokenized mineral rights, mining-specific REITs, direct sovereign wealth fund participation in project finance, any of these gaining traction would compress the pricing advantage streaming companies currently enjoy.

The model’s durability comes down to whether the structural funding gap in mining construction finance persists. As long as banks won’t lend to pre-revenue mining projects and equity markets won’t fund them during downturns, streaming fills a gap that nothing else can. If that gap closes, the terms get worse. No clear signal of it closing yet.