JP Morgan Holds 39 Percent of COMEX Silver and Almost None Registered

Silver closed the week at $60.62 an ounce, down 5.82%, which takes out the $62 support this column identified last week and extends a decline that has now run for three consecutive weeks. The second marker set last week has not been hit: the London free float was to be watched for a break below 6,798 tonnes, and at 6,828.8 tonnes it sits 30.8 tonnes above that floor. So the price call was wrong and the escalation call has not triggered. What did happen was the reversal of the trend this column has been tracking for six weeks. COMEX silver stocks rose 7.0 million ounces week-on-week to 337,581,403 ounces, the second straight weekly build, while London’s vault total did not move by a single tonne for the third consecutive weekly print. The exchange that had been bleeding is refilling and the exchange that had been absorbing has gone completely static, and the depository-level data InProved published on 4 October explains why those two facts are less contradictory than they look.
The COMEX warehouse dashboard, stamped 1 October, shows a combined total of 337,581,403 troy ounces split 101.10 million registered, or 29.9%, against 236.48 million eligible, or 70.1%, with a weekly net flow of positive 4,658,818 ounces. Against the 330,555,907 ounces the global dashboard carried a week ago that is a gain of 7,025,496 ounces, or 2.13%. The registered figure is the one that matters most and it has crossed back above a threshold this column flagged three weeks ago: registered silver stood at 97.29 million ounces in the 19 September analysis, when the question posed was whether it would break below 90 million. It did not. It is now 101.10 million, 3.81 million ounces higher, which converts to 20,220 contracts of genuinely deliverable metal at the 5,000-ounce contract size. The daily flow reads zero against the prior session, so the build is a weekly phenomenon rather than a single dramatic delivery, and the net-flow panel shows green bars in each of the last two weeks after a run of red through late August and September.
This forces a revision to the framing this column has used since early September. The story through the back half of the quarter was COMEX losing share to London while the global pool shrank, and the inference drawn from it was that deliverable silver was becoming progressively scarcer in the one venue where delivery actually happens. The first half of that has reversed cleanly. On the global dashboard dated 1 October, total holdings across the five exchanges reached 41,530 tonnes, or 1,335.2 million ounces, up 302 tonnes from the 41,228 recorded a week earlier. The arithmetic of where that growth came from is unusually tidy: COMEX contributed 218.6 tonnes, China’s SHFE and SGE vaults 77.1 tonnes and India’s MCX system 6.6 tonnes, which sums to 302.3 against a reported total change of 302. London contributed exactly nothing, holding at 28,431.1 tonnes for the third straight weekly reading. The global pool now sits 46.2% of the way up its fifty-two-week range of 38,373 tonnes, set on 7 April 2026, to 45,211 tonnes on 3 October 2025, and remains 8.14% below that high.
The single most valuable chart InProved published this week breaks COMEX silver down by depository as of 1 October, and the concentration it reveals changes how the headline registered number should be read. JP Morgan Chase Bank holds 132.67 million ounces, 39.3% of all COMEX silver and by far the largest position. Of that, 6.87 million ounces are registered. That is 5.2% of its own holdings and just 6.8% of all registered silver on the exchange. The second-largest holder, Asahi Depository, holds 51.03 million ounces, a 15.1% share, but 38.14 million of it is registered, which is 74.7% of its own stock and 37.7% of every deliverable ounce on COMEX. Brink’s holds 42.22 million at 34.4% registered, Loomis 23.03 million at 35.3%, and HSBC 22.17 million at just 13.5%. The top five depositories account for 80.3% of the exchange’s silver and the top ten for 99.5%, so this is a market with ten participating vaults and effectively two that matter for delivery.
The lower panel adds the detail that closes the argument. Of JP Morgan’s 125.80 million eligible ounces, 70.17 million, or 55.8%, is iShares Silver Trust inventory, which InProved notes amounts to 29.7% of all eligible silver on COMEX. Strip the fund out and JP Morgan’s own non-ETF position across both categories is 62.50 million ounces. This is the depository-level confirmation of the free-float argument this column has been making from the London data for a month, and it is considerably sharper. Nearly a third of the eligible category at the exchange is not idle metal waiting to be upgraded to registered status if the price rises; it is fund inventory with a legal claim already attached. The change column makes the migration explicit. Since the end of December 2024 Asahi has added 23.06 million ounces and StoneX 9.75 million, both registered-heavy vaults, while JP Morgan has shed 2.86 million, Loomis 5.72 million, Brink’s 5.31 million and Delaware 3.26 million. Deliverable silver is consolidating into the vaults that specialise in it, which makes the exchange more efficient and the concentration risk more acute at the same time.
London’s frozen total deserves a section of its own because of what it implies mechanically. The LBMA figure has now printed 28,431.1 tonnes in three successive weekly snapshots. Over the same week, InProved’s London ETF monitor shows fund holdings rising from 21,577.0 to 21,602.4 tonnes, a gain of 25.4 tonnes, and the estimated free float falling from 6,854.2 to 6,828.8 tonnes, a decline of exactly 25.4 tonnes. That identity is not a coincidence and it is the clearest illustration available of why the free float matters. When the vault total is fixed, every tonne the exchange-traded funds acquire comes directly out of the unencumbered pool, one for one, with no new metal entering the system at all. Funds now hold 75.98% of everything in the London vaults, up from 75.89% a week ago, and the free float at 219.55 million ounces sits 30.8 tonnes above its three-month low of 6,798 tonnes and 853.2 tonnes below its three-month high of 7,682.
InProved’s own framing of the week was that price weakness is meeting institutional demand and accumulation on the dips, and the three-month series supports it: holdings have added 1,162 tonnes since early July, a gain of 5.68% from a low of 20,399 tonnes, while the silver price overlaid on the same panel fell from a high of $69.40 to $55.52 at its weakest before settling at $60.62. Funds lifted holdings 5.9% above their three-month low through a period in which the metal lost roughly a fifth of its value at the trough. That divergence is now seven weeks old and this column has been describing it for six of them without the price responding, which is long enough to require a different kind of honesty than a restatement of the thesis. Two readings remain live. Either the accumulation is informed and the price will eventually follow it, which is the case this column has argued and which has so far cost money; or the funds are mechanically absorbing creations driven by retail inflows that are themselves a response to the lower price, in which case the tonnage is an effect rather than a signal. Nothing in this week’s data distinguishes between them, and saying so is more useful than picking.
Silver’s forward curve told the same story as gold’s on the same day and for the same reason. The 2 October dashboard shows spot at $61.585, up 0.99% on the session, with the terminal marking the structure contango, flattening after the non-farm employment release. The December spread prints 34.3 cents, March $1.11, May $1.67 and July $2.21, and the intraday comparison shows December actually rose 5.5% from the prior session’s 32.5 cents, so again the compression is a week-scale rather than a day-scale event. Against the 55.0-cent December spread recorded in this column’s 19 September analysis the decline runs to 37.6%, and against roughly 92 cents in mid-August it comes to 62.7%. The step structure is worth noting precisely because it is uneven: December to March costs 76.7 cents, March to May 56 cents and May to July 54 cents, so the front of the curve is still the steepest segment rather than the flattest.
The analytical value of running gold and silver through the same lens on the same day is that it isolates what is common from what is specific. Both curves flattened after the same macro release, both carry the same contango-flattening label, and both saw their deferred spreads compress by a similar order, which means the bulk of the move in each is the same repricing of carry costs that followed the employment data. What remains after subtracting that common component is silver’s front-end steepness, which has persisted while the back of its curve flattened. A market with a cheap carry overall but a comparatively expensive first three months is saying that near-dated metal still commands something the deferred months do not. That is a far weaker claim than backwardation, which silver showed as recently as mid-September and no longer does, and it is a far smaller signal than the free-float and depository data this week. It is, however, directionally consistent with them, and in a week when the price fell 5.82% the consistency is most of what is available.
For bullion dealers, the depository breakdown should change how the headline registered figure is used. COMEX silver registered stock at 101.10 million ounces sounds comfortable and converts to 20,220 contracts, but 37.7% of it sits in a single vault at Asahi and only 6.8% at JP Morgan, which holds 39.3% of all the metal on the exchange. Availability is therefore far more dependent on one depository’s behaviour than the aggregate suggests, and dealers sourcing against COMEX should know which vault their metal is coming from. The constructive news is that the bleed has stopped: two consecutive weekly builds have added seven million ounces and registered has recovered to 101.10 million from the 97.29 million recorded three weeks ago, so the squeeze scenario that looked plausible in mid-September has receded. At $60.62 the metal is 5.82% cheaper than a week ago with more of it available, which is a straightforwardly better buying environment than the one this column described last week.
For conservative investors, the six-week divergence now requires a harder question than it has been given. London funds have added 1,162 tonnes over three months while the price fell from $69.40 to $60.62, and the free float has tightened to 6,828.8 tonnes with exchange-traded funds holding 75.98% of all London silver. This column has read that as informed accumulation for six consecutive weeks and the price has fallen in most of them. The competing reading, that fund tonnage is a mechanical consequence of retail inflows chasing a lower price rather than a signal about future scarcity, fits the same data equally well and nothing published this week separates them. Investors should size accordingly: a position justified by a thesis that cannot currently be distinguished from its opposite deserves less capital than one that can. A reasonable base case puts support at $58 with resistance at $64 and then $67, and a twelve-month range of $80 to $88 if the gold-silver ratio compresses from 68.52 toward 57 while gold reaches the $4,700 to $5,100 band set out in the companion analysis.
For active traders, the mechanical identity in the London data is the cleanest thing to trade around. With the LBMA total frozen at 28,431.1 tonnes for a third week, funds added 25.4 tonnes and the free float fell by exactly 25.4, which means that for as long as London’s vault stays static the free float is a pure function of fund flows and can be forecast directly from daily ETF creations. The level to watch is 6,798 tonnes, the three-month low, now 30.8 tonnes away; a break there with the LBMA total still unchanged would mean roughly four more days of current-pace fund buying and would be worth acting on. The forward curve offers less: December at 34.3 cents is down 37.6% from mid-September and the structure is flattening contango after the same employment release that flattened gold, so most of that move is a rates signal rather than a silver one. What remains silver-specific is the front-end steepness, with December to March at 76.7 cents against 56 and 54 cents for the two later steps, which is a modest point in favour of near-dated metal and not a squeeze.
Hugo Pascal’s observation about the AU9999 contract hitting a 10-week volume high underscores the increasing significance of physical gold trading on the Shanghai Gold Exchange. This trend not only highlights robust domestic demand in China but also reflects broader shifts in the global gold market toward physical-backed assets.
Latest articles
Tool and strategies modern teams need to help their companies grow.
Invite users to stay updated with exclusive insights and market trends by subscribing to the newsletter.
InProved Pte. Ltd. (“InProved”, UEN 201602269C). InProved is regulated by the Ministry of Law (“Minlaw”) and holds a Precious Stones and Precious Metals license for dealing in bullion products (PSPM License PS20190001819). For additional legal and privacy related information related to InProved, please visit are terms and conditions.
Our products and services are only available to Accredited Investors. Investing in bullion involves risk, and there is always the potential of losing money. Certain bullion products are not suitable for all investors. The rate of return on investments can vary widely over time, especially for long-term investments. Past performance is no guarantee of future results. Before investing, consider your investment objectives and any fees and expenses that may be charged by InProved and any third-party stakeholders. The content provided herein is for informational purposes only and is not investment or financial advice, tax or legal advice, an offer, solicitation of an offer, or advice to buy or sell or hold bullion products. This material has not been reviewed by the Minlaw.
Statements made are not facts, including statements regarding trends, market conditions and the experience or expertise of the author or quoted individual(s) are based on current expectations, estimates, opinions and/or beliefs. Opinions expressed by other members on InProved should not be viewed as investment recommendations from InProved. Endorsements were provided at the request of InProved. InProved is not affiliated with and does not purport to own or control any third-party content linked herein.
Copyright © 2026 InProved Pte Ltd (UEN 201616594C, PSPM license PS20190001819)