Investing in Silver

On the morning of Thursday, January 29, 2026, silver’s spot price printed $121.67 an ounce, a record in nominal dollars and a gain of about 68% since New Year (SD Bullion; Finance Magnates). On our tape, which records the front-month COMEX future at the close, that day settled at $114.04. CME Group, the exchange that clears the contract, had raised the initial margin on a silver future from 9% to 11% of contract value the day before. The next morning President Trump nominated Kevin Warsh, a former Federal Reserve governor, to be the next Fed chair; the dollar jumped; and leveraged holders were liquidated.

By the close on Friday, January 30 the future was at $78.29, down 31.3% in one session, the largest single-day fall in the metal’s modern history. After the close CME raised the margin again, to 15%. Gold fell 9.8% the same day on the spot market, its worst session since 1983; the front-month gold future on our tape fell 11.4%.

What is remarkable about that Friday is how little of it was new. Silver’s peak in January 1980 ended when COMEX changed its margin rules and the Hunt brothers could not meet a call. Its peak in April 2011 ended when CME raised margins five times in two weeks and the price fell 33% in five sessions. Its peak in January 2026 ended the same way, at a higher price, with the same mechanism, in a market that had spent five years telling itself the physical deficit made a crash impossible. The deficit was real: 40.3 million ounces in 2025 on the Silver Institute’s count. It did not stop the price halving, because the price of silver is set in a futures market where leverage, not ounces, decides who has to sell.

This guide is about that split: supply that cannot respond to price because most silver is dug up by accident, solar demand that is now shrinking, the ratio that has timed the last two tops, the premiums and taxes that make physical silver the most expensive way to own any precious metal, and a worked $20,000 example in which the wrapper matters as much as the direction. The flagship guide on this hub, Investing in Precious Metals, covers gold, the platinum-group metals, dealers, storage and the full US tax map; where this guide touches the same ground it points there and moves on.

What silver has actually returned to a long-term holder, before the bull case is made, depends entirely on when you start counting; the two most common starting points, the 1980 top and the 2011 top, both produce a loss in real terms even after the 2025–2026 run.

Silver’s peak close in the Hunt corner was $49.45 on January 17, 1980, on APMEX’s and Britannica’s record of the London and COMEX prices, with an intraday high of $50.35 the next day. At $66.68 on September 8, 2026 the metal has gained 35% in nominal dollars over forty-six and a half years, about 0.65% a year. The US consumer price index went from 77.8 in January 1980 to 333.9 in July 2026 on the Bureau of Labor Statistics series, a factor of 4.29, so the 1980 buyer has lost roughly 68% of purchasing power in silver, before storage and before tax. The 1980 peak is worth about $212 in July 2026 dollars, which is why the $121 spot record of January 2026 was a nominal record and nothing more.

From the 2011 top the picture is milder but not good. The New York spot price peaked at $49.85 on April 28, 2011 (Nasdaq, May 2011); at $66.68 that is a 34% nominal gain in fifteen years and four months, about 1.9% a year, against a 48% rise in the CPI from April 2011 (224.906) to July 2026 (333.918). A 2011 buyer who held through the 2015 low and the 2026 record is still down about 10% in real terms. Over the same period the S&P 500 compounded at more than 11% a year nominal on officialdata.org’s 1971–2026 series, about 7% after inflation.

Silver’s defenders will say those are the two worst starting points in the record, and they are right: from the 2020 low near $12 to the 2026 peak the metal rose ten-fold. That is the point. Silver is not a compounding asset but a cyclical one whose long-run real return is close to zero, so the whole return depends on the entry. On our arithmetic, from its 1971 average of $1.55 an ounce (LBMA annual average, via metalcharts.org and SD Bullion) silver has risen 43-fold in nominal dollars to $66.68 and about five-fold in real terms, against roughly forty-fold in real terms for the S&P 500 on officialdata.org’s series; the 2025–2026 run improved the number without changing the shape. What silver offers is not growth but a very high beta to a monetary panic, meaning it moves several times as far as gold on the same news, and that is worth owning in a size that reflects how rarely those panics arrive.

Silver behaves like two different assets, and the smaller half sets the price. Gold’s demand is jewellery, central banks and investment, and its price is a pure expression of what people will pay to hold an asset that does nothing. Silver has a second life: it is the most electrically conductive metal there is, it does not corrode the way copper does, and it is cheap enough per gram to put into things that are thrown away.

On the Silver Institute’s World Silver Survey 2026, researched by Metals Focus and published April 15, 2026, industrial fabrication took 657.4 million ounces in 2025, down 3% after four years of growth. Total demand was 1,130.7 Moz by arithmetic from the survey’s supply and deficit figures, which makes industry about 58% of the market. The rest is jewellery, down 8% in 2025 with India down 20% as buyers baulked at the price; silverware, down 21%; and physical investment in coins and bars, up 14% with Indian bar buying up 33%. Exchange-traded products, which the Institute counts outside the balance, took in a net 68.3 Moz.

The industrial 58% is what the marketing calls the floor, and it is one in a narrow sense: a chipmaker does not stop buying silver because the price fell, and the metal is a small enough share of the finished product that a doubling barely registers in a panel’s cost. But it is a floor that erodes, because the same buyers spend the years after a spike engineering the metal out, a process the trade calls thrifting, and Section 5 shows the largest industrial user has already started.

The investment slice is the opposite: coin and bar buying, ETF creations and, above all, futures positioning can swing by more than a hundred million ounces from one year to the next, in a market whose annual surplus or deficit is measured in tens of millions. Because industry absorbs its 650 Moz whatever happens, the marginal ounce is always an investment ounce and the marginal investor is always a leveraged one. That is the mechanism behind every number in this guide.

Silver supply barely responds to price, which is why deficits persist and also why they matter less than they sound: most silver is a by-product. The Silver Institute’s analysis, summarised by goldsilver.com in its 2026 note on the primary-supply problem, puts the share of mine output that comes as a by-product at about 72%: 31% from lead-zinc mines, 25% from copper mines, 16% from gold mines, and only 28% from mines where silver is the main product. A copper miner in Peru does not open or close a pit because silver went from $30 to $100, and the primary silver miners that would are a minority of the total.

The 2025 numbers show what that means. Total supply rose 7% to 1,090.4 Moz, but almost none of it came from new mines. Mine production rose 3% to 846.6 Moz, mostly on base-metal output and on acquisitions by the large primary producers, and is forecast to slip 0.3% to 844.1 Moz in 2026. Recycling rose 2% to 197.6 Moz, its highest in more than a decade, as a price above $50 pulled old silverware, jewellery and industrial scrap out of drawers and off factory floors; the survey forecasts a further 7% rise to 211.3 Moz in 2026. Recycling is the one part of supply that responds to price, and it responds with a lag of about a year.

The producers you can buy are the exceptions to the by-product rule. Pan American Silver produced 22.8 Moz of silver in 2025, above its updated guidance, on its January 20, 2026 filing (SEC Form 6-K); Hecla Mining, the largest US primary producer, reported 17.0 Moz; First Majestic a record 15.4 Moz, up 84%, largely from its January 2025 acquisition of Gatos Silver. Together that is about 55 Moz, or 6.5% of world output, which is the scale of the whole listed primary-silver sector. Even they cannot bring on new ounces in less than the five to ten years a mine takes to permit and build, so a price spike produces recycling and thrifting long before it produces supply.

The phrase that has carried silver’s marketing for five years has a number attached, and the number did not prevent a 51% crash. The Silver Institute counts a deficit when fabrication and physical investment exceed mine and scrap supply, with ETF flows left out. On that definition 2025 was the fifth consecutive deficit, at 40.3 Moz, and 2026 is forecast to be the sixth at 46.3 Moz, taking the cumulative shortfall since 2021 to 762.1 Moz.

Two honesty notes belong here, because the number is quoted loosely. First, the deficit has shrunk: the 2022–2024 shortfalls were reported in earlier editions at well over 100 Moz a year, and on our precious-metals desk’s reading of the successive surveys the smaller current figures reflect both revised methodology and genuinely weaker industrial offtake. Second, the headline moved inside the year. The Institute’s own preliminary outlook of February 10, 2026, reported by Kitco that day and by Money Metals via Advisor Perspectives on February 15, put the 2026 deficit at 67 Moz and the 2025 shortfall at about 95 Moz; the April survey cut both, to 46.3 and 40.3 Moz. A larger number, a “record” 215 Moz for 2026, circulated on Canadian Mining Report attributed to the survey; it appears in none of the Institute’s own releases we could find, and we do not use it.

The reason a deficit does not empty the market is that silver, unlike copper or oil, has an enormous above-ground stock willing to sell at a price. The LBMA’s London vault data for end-August 2026 shows 28,431 tonnes of silver, about 914 Moz, valued at $64.2 billion in roughly 948,000 bars, up 0.77% on the month and near a three-year high as metal flowed back from the United States (LBMA via BullionVault, September 1, 2026). COMEX-approved warehouses held 338.7 Moz on September 4, 2026, of which 99.4 Moz was registered, meaning warranted and deliverable (CME data via goldsilver.ai and heavymetalstats). London and New York alone hold about 1,250 Moz, roughly thirteen months of total demand, before Shanghai and private vaults.

Much of the London figure backs exchange-traded products, iShares Silver Trust alone about 474 Moz by our arithmetic on its June 30, 2026 net assets, so it is not free float; but ETF holders are exactly the sellers who appear when the price falls, as they did between January and July. A deficit, then, is a rate of inventory drawdown, and 40 Moz a year against 1,250 Moz of visible stock is about 3% a year. It tightens the market and makes squeezes more likely, and it is a genuine reason silver’s floor rises over a decade. It is not a reason the price cannot halve in a quarter, which is what it did in the first half of 2026 with the deficit intact.

Silver’s largest growth market, the one the bull case of 2020–2025 was built on, was quietly taken away by the 2026 survey. A crystalline silicon solar cell collects its current through a grid of silver paste screen-printed onto the wafer; nothing else conducts as well, sinters as easily or lasts twenty-five years in the weather. As the world’s panel output went from tens of gigawatts a year to hundreds, photovoltaic silver demand rose from about 60 Moz in the mid-2010s to a record 197.5 Moz in 2024, roughly a fifth of all demand, and every forecast of a structural deficit had that curve at its centre.

The curve has turned. On the World Silver Survey 2026, as reported by Mercom India on publication, PV demand fell 6% to 186.6 Moz in 2025 and is forecast to fall a further 19% to 151 Moz in 2026. Installations kept growing; the silver in each cell shrank faster. The mechanism is thrifting, and the price did the engineering: at $30 an ounce silver paste was a nuisance, at $100 it was the largest non-silicon cost in the cell, and every manufacturer’s roadmap was pulled forward; the survey expects the declines to continue.

Two technologies drive it. The first is thinner printing: finer grid lines, less paste per cell, silver-coated copper in place of pure silver. Fraunhofer ISE, the German solar research institute, announced in April 2026 that it had cut silver consumption in a TOPCon cell, the mainstream architecture, from the industry’s 10–12 milligrams per watt to 1.1 milligrams, a factor of ten, on pilot production equipment, by plating nickel and copper and leaving only a trace of silver as corrosion protection. TOPCon cells use about 50% more silver than the older PERC design, which is why demand held up through 2024 even as loadings per cell fell.

The second is full copper metallisation, which plates copper contacts directly onto the wafer and removes silver from the cell altogether; LONGi, the largest module maker, said it would begin substituting base metals for silver in its cells in the second quarter of 2026 (pv magazine, January 6, 2026; Solar Power World, March 2026). Scenarios in 2026 trade reporting have copper metallisation reaching half of cell output by 2030 and PV silver demand falling toward 100 Moz; we could not trace that timetable to the Silver Institute’s own publications and treat it as an illustration of direction, not a forecast.

Whether copper wins on that timetable is an open engineering question; copper contacts have corrosion and adhesion problems silver never had, and Solar Power World’s March 2026 industry survey asked in its headline whether removing silver was wise. But the direction is not in doubt, and the asymmetry matters: the price spike accelerated the substitution, and the substitution does not reverse when the price falls. Electronics, contacts, brazing and grid wiring are growing from smaller bases at single-digit rates, which is why total industrial demand fell 3% in 2025 before the full solar decline landed. The honest reading of the industrial half is flat to down, with the bull case resting on investment demand, the half that produced the crash.

Silver’s three modern tops before 2026 each ended the same way, and the pattern is the most useful thing in the metal’s history. Silver does not top out because demand fades. It tops out because the leverage that took it up is withdrawn by the exchange that clears it.

1980: the Hunt corner

Nelson Bunker Hunt and William Herbert Hunt began buying silver in the early 1970s as an inflation hedge and by late 1979 held, on Britannica’s and APMEX’s accounts, more than 100 million ounces of bullion and futures, roughly a third of the world’s privately held supply. The price went from under $6 in early 1979 to a close of $49.45 on January 17, 1980. COMEX had already moved: on January 7, 1980 it adopted Silver Rule 7, restricting new positions on margin, and shortly afterwards went to liquidation-only, so the Hunts could sell but no longer buy. Deprived of new buyers the price fell for two months, and on Thursday, March 27, 1980, Silver Thursday, the brothers failed a margin call of about $100 million and the price collapsed from $21.62 to $10.80 in a session. The peak was not seen again in nominal terms for thirty-one years.

2011: five margin hikes in two weeks

The second top came out of the financial crisis, quantitative easing and the first ETF boom. Silver rose from about $18 in mid-2010 to a New York spot peak of $49.85 on April 28, 2011, just short of the 1980 close. Between April 25 and May 5, 2011 CME raised silver margins five times, an 84% increase in the cash needed to hold a contract, and the New York spot price fell as low as $34.95 on May 5, a 33% drop in five sessions (Bloomberg, May 5, 2011; Nasdaq, May 6, 2011; Investing News Network’s 2011 review). The metal then spent four years grinding down to about $14 while the deficit narrative slept.

2020–2021: the squeeze that was not

In the March 2020 liquidity panic silver fell to about $12 while gold held near $1,477, sending the gold/silver ratio to its all-time record in the mid-120s on March 18, 2020 (CPM Group; USAGOLD). Eleven months later, in the week after the GameStop squeeze, posts on the WallStreetBets forum urged retail investors to buy silver and SLV to squeeze the banks; on Friday, January 29, 2021 the iShares Silver Trust took in a record $944 million in a single day (TheStreet; Fortune), and on Monday, February 1 COMEX futures jumped as much as 13% to $30.35. It lasted three days: there was no short to squeeze, because the banks’ futures shorts hedged the metal in their vaults, and the price was back below $27 within a week. The squeeze did put a generation of retail buyers into coins at 30–50% premiums, the most expensive trade in this market.

Here is the fourth top, told from our own data rather than the press. Our tape is Invest Alternative’s live store: the daily close of the front-month COMEX silver future, the nearest-dated contract (Yahoo Finance, SI=F), stored since August 30, 2021 with a 500-session chart window that opens on September 12, 2024. It is a futures close, not the LBMA spot price, so it prints a little below the intraday records quoted in the press; every figure in this section is ours and dated.

The run began quietly. On September 12, 2024 the future closed at $29.74, with the gold/silver ratio at 85.8; it closed 2024 at $28.94, the low of the window. Through the first half of 2025 silver lagged gold badly, which is how the ratio reached 104.8 on April 21, 2025, and then it caught up all at once. The first close above $50 was $50.13 on October 13, 2025, above $75 on December 26, and above $100 at $100.93 on January 23, 2026. The single best day in the window was Monday, January 26, 2026, up 14.0% to the peak close of $115.08, and the future closed above $100 on exactly five sessions, January 23 through 29. Calendar 2025 was a gain of 142.3% on our closes, and January 2026 added another 64% to the peak.

Then the Friday. From $114.04 on January 29 the future closed at $78.29 on January 30, 2026, a fall of 31.3% in one session, worse than Silver Thursday measured from the prior close. The causes, in the order the market experienced them (Bloomberg, January 28 and 31; Finance Magnates and CNBC, January 30, 2026), begin with the margin. CME had switched silver from a fixed dollar margin to a percentage of contract value, 9%, on January 13; on January 28, with the price at records, it raised that to 11%. The Warsh nomination and a sharp dollar rally on the Friday morning then met a market that had been adding leverage all the way up, and forced liquidation ran across commodities as losses hit leveraged books.

After Friday’s close the exchange raised silver’s margin again, to 15% (16.5% for accounts flagged as heightened risk), effective at Monday’s close, and CNBC reported on February 2 that the increase extended the selling. On SD Bullion’s record spot traded just below $75 within thirty hours of the $121.67 print; the month-end spot close of $85.26 hides the fact that the futures close was already $78.

The crash was not one day. Apart from that Friday, the other five of the six worst sessions in our window fell between December 29, 2025 and May 15, 2026, each between 8.7% and 9.8%, and the rallies in between, to $92.68 by February 27, kept pulling buyers back before the next leg down. The trough close was $55.90 on July 16, 2026, three days after gold broke below $4,000 on Fed Governor Waller’s rate-hike remarks: a drawdown of 51.4% from January 26 in 171 days. Silver has since recovered 19.3% to $66.68, still 42.1% below its peak close but 61.0% above a year earlier ($41.43 on September 8, 2025). Our Precious Metals sub-index, an equal composite of the four metals’ futures based at 100 on September 2, 2025, stood at 139.05 on September 8, 2026, up 39.8% on the year; silver gave it more on the way up and took more back on the way down than any other metal.

The tape’s lesson is the asymmetry: the best six sessions in two years were between +7.3% and +14.0%, the worst between −8.7% and −31.3%. A holder who was right for sixteen months and late by one week gave back most of the trade. Silver rewards being early and punishes being late more than any asset on this hub, which is the argument for the rules in Section 8 and the sizing in Section 15.

The gold/silver ratio, the number of ounces of silver that buy one ounce of gold, is the one timing tool in precious metals with a mechanism behind it, and our tape shows it working twice. For most of monetary history it was fixed by law: the US Coinage Act of 1792 set it at 15 and the 1834 revision at 16. With both metals floating, the average depends on the window: about 47 for the whole twentieth century, and 63–65 for the floating-rate era since the 1970s (Britannica Money; USAGOLD’s 2025 guide; CPM Group’s long-run series). The extremes are what matter: a spike to about 100 in 1991, and the all-time record in the range of 123–127 on March 18, 2020, with gold near $1,477 and silver near $12.

The ratio means something because silver is a higher-beta version of gold’s trade: the same monetary anxiety moves silver two or three times as far, because silver’s investment demand is a small slice of a small market. When the ratio is very high, silver has been left behind by a gold move its own investors have not yet followed; when it is very low, silver has run further than gold on the same news and the beta is exhausted.

On our tape the ratio has done exactly what the theory says, twice. It stood at 85.8 when the window opened, climbed to a window high of 104.8 on April 21, 2025 as gold spiked on tariff fears and silver did not, then collapsed as silver caught up: 83 by the end of September 2025, 61.7 at year-end, and 44.1 on January 26, 2026, the day silver made its peak close. That was the lowest reading since 2011, and it marked the top to the day. By the July 16 trough the ratio had rebounded to 71.3; it is 66.7 as we write.

A rule that said “hold silver over gold above 90, hold gold over silver below 50” would have put you into silver in the spring of 2025 at $32 and out in the third week of January 2026 above $100. Use it as a switch between the two metals, not a signal to own either: the ratio says whether silver is expensive relative to gold and nothing about whether gold is expensive.

Neither the 2021 squeeze nor the 2026 crash makes sense without knowing where silver’s price is made and who takes the other side of your trade.

London is the physical hub. Trading is over the counter between LBMA members in 1,000-ounce Good Delivery bars, the LBMA’s standard for vault-quality metal, and the benchmark is the LBMA Silver Price, an electronic auction that replaced the century-old silver fix on August 15, 2014. The bullion banks that own or custody the London bars are the natural sellers into any rally: when silver rises they sell futures against their vault metal. That is the “short” the 2021 squeeze thought it could break, and it was a hedge of silver they already owned, not a bet against it.

New York is where the price is discovered minute to minute. The COMEX silver future is a contract for 5,000 troy ounces, deliverable from approved warehouses that held 338.7 Moz on September 4, 2026, of which 99.4 Moz was registered; a micro contract of 1,000 ounces serves smaller accounts. Because the future trades on margin, the exchange decides how much leverage the market carries, and the margin call is the instrument of every silver top since 1980. Since January 13, 2026 CME has set silver’s margin as a percentage of contract value rather than a fixed dollar sum: 9% at the switch, 11% from January 28 and 15% from February 2 (MINING.COM on the methodology change; Bloomberg, January 28 and 31, 2026), the sequence Section 7 tells in full.

In dollars, the contract’s notional, the full value of the 5,000 ounces it controls, had risen from about $150,000 to more than $550,000 in a year, so the move from 9% to 11% two days before the top asked traders who had been adding all the way up for roughly $11,000 more per contract, and the Friday morning’s 20% down move turned a correction into a liquidation. A percentage margin rises with the price automatically, which should make the next top less violent than one built on a fixed dollar margin that shrinks as a share of a rising contract.

Shanghai and India are the marginal physical buyers. The Shanghai Futures Exchange and the Shanghai Gold Exchange run their own contracts and warehouses. SHFE silver stocks fell through late 2025 to a reported low of about 318 tonnes in February 2026, when the exchange raised margins and tightened delivery rules (David Jensen’s tallies of the exchange’s warrant data; BloFin Research via Yahoo Finance), and had rebuilt to about 1,260 tonnes by August 2026 on CEIC’s series; the drawdown was the strongest evidence that the deficit was being met partly from Chinese inventory, and the refill says the price did its job. Indian physical investment rose 33% in 2025 while Indian jewellery fell 20%, and an Indian import surge was behind much of the 2024–2025 tightening in London.

The market has been policed. On September 29, 2020 JPMorgan agreed to pay about $920 million to the CFTC, the DOJ and the SEC to resolve charges that its precious-metals desk had spoofed gold and silver futures for eight years, placing orders it intended to cancel in order to move the price; two of its senior traders were convicted in 2022. The record found short-horizon cheating and no long-horizon suppression, which is the honest answer to the theory every silver bull market revives. The other side of your trade is a hedged bullion bank, a leveraged fund, an Indian household or a Chinese refiner, not a conspiracy.

Physical silver is the most expensive way to own any precious metal, for four reasons with numbers attached: the premium, the spread, the tax at the border and the bulk. The flagship guide covers dealers, testing equipment and vault choice for all four metals; this is silver’s own arithmetic.

The premium

A one-ounce American Silver Eagle is a US Mint coin of .999 fine silver, and the gap between its price and the metal in it is the fabrication charge, the Mint’s premium, the wholesaler’s margin and the dealer’s. On FindBullionPrices’ dealer comparison, which tracks the lowest national-dealer price daily, the Eagle premium averaged 8.48% in July 2026 and 6.87% in August, in a range of 6.36–7.90%; on September 8, 2026 the cheapest random-year Eagle was $69.94, $3.35 or 5.0% over melt, and the cheapest 2026-dated coin $71.61, about 7.5% over the same melt value.

Those are calm-market numbers. In the panic weeks of March 2020 and again in 2022–2023, on our precious-metals desk’s record, Eagle premiums went above 50%, and the 2021 squeeze buyers paid 30–50% over spot for coins they could later sell only at spot. Premiums are a percentage of a small number: a $5 fabrication charge is 17% at $30 silver and 5% at $100. Bars cost less: the cheapest 100-ounce bar on the same comparison was $6,607 on September 8, 2026, about 0.5% over its melt value at that morning’s $65.74 spot. A 1,000-ounce Good Delivery bar trades essentially at spot but can only be sold whole to a buyer who can assay it.

The spread

The premium is paid once, on the way in; what you recover is the dealer’s bid, which does not carry it. In the 2026 correction, on our desk’s record, national dealers bought Eagles back at spot or a few dollars over while selling 7–8% over, so a round trip costs about 7% before the metal has moved, against about 0.1% on a physical ETF; in a stressed market the bid drops below spot and the round trip widens past 10%.

The tax at the border

In the United States forty-five states exempt bullion from sales tax and five (Hawaii, Maine, New Mexico, Vermont and Washington) plus the District of Columbia do not; the flagship guide and Investing in Gold carry the 2026 map. Silver’s particular problem is Europe. Investment gold is VAT-exempt across the EU and the UK; silver is not. A British buyer pays 20% VAT on silver coins and bars, a German buyer 19%, a Swiss buyer 8.1% (BullionByPost’s and Swiss Gold Safe’s 2026 tax guides), and none of it comes back on resale; Germany at least exempts the gain after a year’s holding, while a British buyer pays capital gains tax on bars and foreign coins that started 20% underwater (UK legal-tender Britannias are exempt from CGT, though not from the VAT). For a European reader physical silver is a hobby and the ETF is the investment.

The bulk

Silver is about 10.5 grams per cubic centimetre and gold about 19.3, so at a ratio of 66.7 a dollar of silver weighs 66.7 times as much as a dollar of gold and takes up about 120 times the space. Twenty thousand dollars of silver at $66.68 is 300 ounces, 9.3 kilograms, roughly a litre of metal; the same money in gold is 4.5 ounces, 140 grams, smaller than a matchbox. Every storage cost scales with that: BullionVault charges 0.48% a year to vault silver against 0.12% for gold on its published tariff, a home safe that holds a lifetime’s gold holds a few thousand dollars of silver, and the 1,000-ounce bar, the only form that trades near spot, weighs about 31 kilograms.

Ranked by what they cost and what they are, these are the ways to own silver without taking delivery. The right wrapper for a silver view is almost never the coin, and among the paper routes the differences are large enough to change the outcome of a five-year trade, as Section 13 shows.

iShares Silver Trust (SLV)

The largest, oldest and most liquid: a grantor trust, meaning the tax code treats each shareholder as owning a slice of the metal directly, holding allocated silver in London, specific numbered bars in the trust’s name. The sponsor’s fee is 0.50% a year, paid by selling a little metal each month, and net assets were $28.19 billion at June 30, 2026 on its 10-Q, about 474 Moz at that day’s price by our arithmetic, more than half of everything in the London vaults. Its size buys the tightest spreads and the deepest options market; the fee is the price of that.

abrdn Physical Silver Shares (SIVR)

The same structure for 0.30%, a 0.45% sponsor’s fee under a voluntary waiver that abrdn can withdraw, with net assets of $3.95 billion at June 30, 2026 on its 10-Q and metal in London under the same allocated structure. It is the cheapest allocated silver on a US exchange and the default answer in this guide.

Sprott Physical Silver Trust (PSLV)

A Canadian closed-end trust whose units trade on the NYSE and Toronto at a premium or discount to net asset value that swings several percent with sentiment, and whose holders can redeem in physical 1,000-ounce bars above a minimum. Its management expense ratio was 0.51% annualised in the first quarter of 2026 (Sprott’s Q1 2026 report, SEC Form 6-K). Its attractions are that the metal sits at the Royal Canadian Mint rather than a bullion bank, and that its structure allows a US holder who files a timely qualified electing fund (QEF) election on IRS Form 8621, in the first year of ownership and every year after, to be taxed on a sale after one year at the 15% or 20% long-term rate rather than 28%, on Sprott’s published tax guide. The trust is a passive foreign investment company, and a holder who misses the election gets the PFIC regime instead, so confirm it with an accountant before relying on it.

Futures

Leverage, the best liquidity and the best tax treatment: a 5,000-ounce contract or a 1,000-ounce micro, margin set as a percentage of contract value since January 13, 2026 (9% at the switch, 15% after the crash; check CME’s page for today’s rate), and Section 1256 treatment at a blended top rate near 26.8%. The costs are the roll, a few percent a year of contango (the premium of later-dated contracts over nearer ones, which a holder pays each time the position is moved forward) at 2026 short rates, and the margin desk, which is the counterparty that ended every silver top on record.

Miners and streamers

Miners are operating leverage on the silver price with everything that can go wrong at a mine on top. Pan American Silver, Hecla and First Majestic (Section 3) each also mine gold, zinc and lead, so none is a pure silver bet. For 2026 Pan American guides to 25–27 Moz of silver at a silver-segment all-in sustaining cost of $15.75–18.25 an ounce, Hecla to 15.1–16.1 Moz at $15.00–16.25, and First Majestic to 14.6–15.5 Moz at $27.69–28.77 per silver-equivalent ounce (company guidance releases, January and July 2026); at $66 silver those are margins of $40–50 an ounce, which is why the shares move two to three times as far as the metal.

Wheaton Precious Metals is the streaming alternative: it finances base-metal miners in exchange for the right to buy their by-product silver at fixed prices far below spot, carries the price upside with no operating costs and 80–90% margins, and trades, as streamers do, at a premium to net asset value. Miners and streamers are equities taxed at 20% with qualified dividends, which is why a taxable investor may prefer them to the metal despite the operating risk.

The US tax treatment of each silver wrapper changes the ranking above. The flagship guide covers the full map for all four metals, the IRA rules, McNulty and the FBAR; here is what a silver holder needs.

Silver bullion and coins are collectibles under Internal Revenue Code §408(m), and §1(h)(4)–(5) caps the federal long-term capital-gains rate on a collectible at 28%, not the 20% that applies to shares. The 3.8% net investment income tax applies on top above the income thresholds, so the top federal rate on a silver gain held more than a year is 31.8% before state tax; a gain on metal held a year or less is ordinary income at up to 37%. Losses are capital losses and offset gains in the usual way. Storage and insurance are not deductible for an individual investor.

The physical ETFs are taxed as the metal. SLV and SIVR are grantor trusts; the IRS looks through the share to the silver, so a long-term gain on the shares is taxed at 28%, and the trust’s monthly sales of metal to pay its fee give every holder a small reportable gain or loss each year that the sponsor reports on a schedule most people never read. PSLV’s structure is the reason it exists for taxable investors: with the timely QEF election described in Section 11, its long-term gains are taxed at the 15% or 20% equity rate. Futures are Section 1256 contracts, 60% long-term and 40% short-term whatever the holding period, marked to market at year-end, with a blended top federal rate of about 26.8%, which is why the leveraged wrapper is taxed more gently than the coin. Miners and streamers are ordinary equities: 20% on long-term gains, qualified dividends.

An IRA may hold silver of at least .999 fineness, and American Silver Eagles by name, provided a trustee holds it at an approved depository under §408(m)(3); the Tax Court’s 2021 McNulty decision made a “home storage” IRA a taxable distribution of the whole account, and the IRS’s 2007 private letter ruling lets an IRA hold SLV or SIVR without triggering the collectibles rule, which for most people is the simpler route.

Dealers file Form 1099-B when you sell them silver bars of 1,000 ounces or more or $1,000 face value of pre-1965 90% silver coin; Silver Eagles are not on the reportable list in any quantity, which is a reporting matter and not an exemption, because the gain is taxable either way. A dealer receiving more than $10,000 in cash files Form 8300. Buying is not reported at all.

The same $20,000 goes through two wrappers so the costs become dollars. The assumptions are stated and simple: you buy on September 8, 2026 at the tape close of $66.68; you hold five years; you are in a state with no sales tax on bullion, in the 28% collectibles bracket, without the net investment income tax; and silver is either 50% higher at the end, $100, or exactly where it started. The premium and fee figures are the ones sourced in Sections 10 and 11.

Route A: Silver Eagles in a depository

At the August 2026 average premium of 6.87% an Eagle costs $71.26, so $20,000 buys 280 coins for $19,953. Storage at a depository’s 0.50% non-segregated rate on a value rising from $18,670 to $28,000 averages about $117 a year, $583 over five years.

You sell to a national dealer at spot, which is what the bid was through the 2026 correction: 280 × $100 = $28,000. The taxable gain is proceeds less basis, and the premium is part of the basis: $28,000 − $19,953 = $8,047, taxed at 28% = $2,253. Net cash after storage and tax: $28,000 − $19,953 − $583 − $2,253 = $5,211, a return of 26.1% on a metal that rose 50%. If silver is flat, the coins sell for 280 × $66.68 = $18,670, storage has cost $467, and you are down $1,750, or 8.8%, with a $1,283 capital loss to carry.

Route B: SIVR in a brokerage account

$20,000 buys the shares with no commission and a spread of a few basis points. At 0.30% a year the metal exposure decays to 98.5% of what it would have been: at +50% the position is worth $30,000 × 0.9851 = $29,553. Gain $9,553 at 28% = $2,675. Net $6,878, or 34.4%. Flat, the position is worth $19,702, a loss of 1.5%.

The gap between the routes is $1,667 in the good case and $1,452 in the flat case, and nearly all of it is the coin premium, paid in full on the way in and recovered nowhere on the way out. Add a 4–10% sales tax in one of the taxing states, or 20% VAT in Britain, and Route A loses money at +50%. The fee routes are not free either: the 28% rate takes more than a quarter of the gain in both, and a PSLV holder with a valid QEF election at 20% would keep about $500 more than the SIVR holder on the same move even after its 0.51% fee, and a futures trader about $115 more under 60/40, before the roll. Physical silver is for the reader who wants to hold metal for reasons that are not financial; for the return, use the fund.

There are four ways a silver investor loses everything rather than a percentage, and price is not one of them; a 51% drawdown is severe but survivable if the position was sized for it. These are the failures that are not.

Leverage into a margin change

Every silver top in this guide was ended by the clearing house raising the price of leverage at the moment the leveraged were most exposed, and each time the holders wiped out had been right about the metal for months and wrong about the exchange for a day. The Hunts held physical silver and were destroyed by a $100 million futures call. A futures position sized so that a 30% one-day move would not liquidate you is, at the 15% margin CME set in February 2026, one holding twice the posted margin in spare cash, and at the 9% the year began with, more than three times; most retail futures accounts are not run that way. Options on SLV are the same risk with a time limit.

Dealer storage that was never metal

Northwest Territorial Mint filed for bankruptcy in 2016 owing tens of millions to customers who had paid for metal that was never bought; Bullion Direct in Texas collapsed in 2015 with a vault that held a small fraction of the silver its customers believed was there. Both sold “storage” of allocated metal; both were the metal’s only custodian. The flagship’s test applies with more force to silver, because bulk makes dealer storage tempting: a bar list with serial numbers, a third-party vault operator, insurance in your name, and the right to take delivery. Without all four you own a promise from a coin dealer; unallocated and “pool” accounts are the same promise made openly.

Counterfeits and the broken chain

Fake silver is cheaper to make than fake gold and is made in volume: plated copper and lead-alloy Eagles, Maple Leafs and Morgan dollars from overseas workshops appear on marketplaces every year. The dealer’s stack of tests, weight and dimensions against the mint specification, a Sigma Metalytics resistivity pass, XRF for surface alloy, is described in the flagship guide; for silver the practical rule is that a coin bought outside a dealer’s invoice is worth what a refiner will pay after melting it.

Rule changes

The rules of the futures market changed on the Hunts in January 1980 and on everybody else in January 2026; the rules on retail leveraged metal trading changed in China in July 2026, when ICBC and other banks ended the product. Silver was nationalised once in the United States: Executive Order 6814 of August 9, 1934, under the Silver Purchase Act, required the delivery of silver bullion to the Mint at about 50 cents an ounce, exempting coins, and about 113 Moz was surrendered (The American Presidency Project; JM Bullion). We mention it only because a reader who keeps a portion of gold abroad has the same reason with silver, and the same FBAR paperwork.

The sequence, for a reader who has decided silver’s beta belongs in the portfolio and has read the costs, is short because the first step does most of the work.

1. Decide what the silver is for, and size it as a trade. If it is monetary insurance, buy gold instead (Investing in Gold works the wrappers, the storage arithmetic and the 28% rate for that metal); the flagship’s allocation of about 5% in gold stands, and silver on top is the leveraged expression of the same view. A silver position you would be content to see halve, which on our tape it did in 171 days, is the right size: for most readers 1–3% of financial assets, not 10%.
2. Choose the wrapper by tax status and geography. In a US IRA or Roth, SIVR at 0.30% with no collectibles problem. In a US taxable account, SIVR by default, PSLV if your accountant confirms the QEF election is worth the higher fee and the premium/discount risk, futures only if you can post several times the margin. In Britain or the EU, the ETF or a vault outside the VAT net; never physical at 19–20% VAT.
3. If you want metal in hand, buy the cheapest recognised unit and record the premium. Random-year Eagles or Maple Leafs at the lowest premium FindBullionPrices shows that day, or 10- and 100-ounce bars from a refiner you can verify; not 2026-dated coins at a 2- to 3-point surcharge, not rounds from unknown mints, not “collector” coins. Write the premium down: it is the number you have to recover before you make anything.
4. Store by value. Under about $5,000, a home safe and a discreet household. Above that, a third-party depository with a bar list and insurance in your name at a published rate near 0.50%, or a BullionVault-style allocated account at 0.48%, and accept that silver storage costs four times gold’s per dollar.
5. Buy on the ratio, not on the headline. Enter and exit on the Section 8 rule, above 90 in and below 50 out, adding in stages, with the exit set at the ratio and not at a price. Do not buy the day after a 10% up move: on our tape five of the six best sessions in two years came in the month before the top.
6. Keep the records the tax code wants. Purchase invoices with the premium shown, storage statements, and the sponsor’s annual gain schedule for SLV or SIVR. The 28% rate is unavoidable on the metal; the records make sure you pay it on the right number.

Silver’s story is told in a few numbers a year, most of them arriving on known dates; these are the readings that would change our view, with thresholds.

• The gold/silver ratio, daily on our tape. The two thresholds are 90 and 50, as in Section 8. It is 66.7 as we write; the readings that mattered last appeared on April 21, 2025 (104.8) and January 26, 2026 (44.1).
• The World Silver Survey, mid-April 2027. A photovoltaic demand print above 190 Moz would reopen the solar thesis; a print below the 151 Moz forecast confirms thrifting is running ahead of schedule. A deficit above 100 Moz on the survey’s own definition would be a genuine tightening; a “record deficit” headline from a different release is not, until you have checked the edition.
• London vault silver, monthly from the LBMA. 28,431 tonnes at end-August 2026. A fall below 25,000 tonnes would be the visible-stock warning that a deficit is finally biting; the current three-year high says it is not.
• COMEX registered stocks and margins. Registered silver of 99.4 Moz on September 4, 2026; below 60 Moz the deliverable market is thin enough for squeezes. Any CME margin change of more than 10% in a week is the signal that preceded every top in this guide, and the correct response is to cut leverage the same day, not to argue.
• Our tape. The trough close of $55.90 on July 16, 2026 is the line: a close below it says the bear phase that began January 30 is not over. A close above the February 27, 2026 rebound high of $92.68 would say the leveraged excess has been worked off and a new leg is under way, and we would expect the ratio near 55 when it happens.
• The Eagle premium. Monthly averages from FindBullionPrices: 6.87% in August 2026. A premium back above 20% means retail is chasing again and the coin market is the worst place to be a buyer; a premium below 4% with the bid at spot is the physical market at its cheapest, which is where a coin buyer should do all their buying.