The USA's Silver Shadow Inventory

It seems like bullion banks built a shadow inventory of silver stock outside of the COMEX system. A quick timeline recap from my notes:

August 2025 - USA announces consideration of silver for critical materials list
October 2025 - Silver refineries curtail buying less than .999 fine silver from PM dealers; LBMA runs dry of liquid free float silver vault stock
November 2025 - USA adds silver to the critical materials list

From September to November and again from December through March, the COMEX drains silver inventory massively as bullion banks ship the silver to London to support the LBMA in satisfying demand for physical silver from India, China et al. However, from February through March (and possibly April and May but we don't have the data yet), the USA exported ~500 metric tons more silver than what was being withdrawn from the COMEX.

I posit that this mysterious excess silver export stock is coming from silver bars being produced from LATAM silver dore by USA silver refiners and the bullion banks are vaulting the new bars, but not depositing it into the COMEX system. This could partially explain how COMEX vault custodians magically "adjusted" 78 metric tons of silver into existence in the COMEX vaults in April and May.

So how much shadow inventory could there be? If I'm correct that USA refineries are still producing silver bars and they just aren't being deposited into the COMEX, we should be able to see a decline in COMEX silver deposits. So let's look at the data*:

In the chart above, we can see the total silver deposits for each month along with the four year average of average monthly silver deposits from 2021-2024. I did not include 2025 in the average because the tariff scare and consequent draining of LBMA vault stock skewed the monthly totals dramatically. For the sake of this exercise, we want to see what the normal COMEX silver inflows looks like.

COMEX silver deposits fell dramatically in February 2026 versus the four year monthly average for the month of February. COMEX deposits across February and March 2026 were about one third (33%) of the four year monthly average ("normal"). April was slightly over half of "normal". It looks like COMEX silver deposits normalized in May.

The total deficit from February through April (versus "normal") was 19.7M ozt which works out to ~613 metric tons. That perfectly explains the ~500 metric tons exported in February and March that were not withdrawn from the COMEX. We have to wait a bit longer for the USA's silver export data, but it would not surprise me at all to see another ~35 metric tons of silver beyond COMEX withdrawals left the shore:

613t "shadow inventory" - 500t already exported - 78t "adjusted" into COMEX vaults = 35t

The vault flows of physical silver aren't very transparent, but there are clues available for us to make inferences.

*Special thanks to @AGgregator for database wizardry assistance in putting the COMEX deposit data together.

 

[nickndfl]

 

[pmbug]

Nick making off some shadow inventory of his own.

 

[Voodoo]

I forget who's idea it was that originally purported that theory... That of the US refiners going "off-line" to process ore we are getting from Latin America. It seemed quite logical and consistent with our Venezuelan oil tactic.

Almost all of that was probably going to China for refining before, so that would coincide with the critical minerals thing being the excuse.

 

[pmbug]

Vince Lanci & Josh Philip Phair (Scottsdale Mint CEO/Founder) were the guys I remember first talking about it.

 

[Voodoo]

I think there was an interview with a big wholesale dealer in the Florida area who mentioned it as well. He finally got a big shipment into his refiner right before the interview. I think it was this guy Brian Kuzmar... Commercial Rare Coins. This isnt the interview I remember but might be an update.

 

[Voodoo]

He talks about Metalor a lot and those guys came straight out of a rigging operation. So you can be sure they are doing the bidding of the bankers in the end.

 

[nickndfl]

I was in Lauderdale-by-the-sea last week where he is headquartered. In fact, I had a route business there in the 90s and have been in most of the offices and condos in the area. It is too busy for me to live there now.

I was just up the road in Pompano Beach in a condo office when 911 went down. I watched it on tv with the client and knew I better finish up and get off the road. The travel agent in the area who sold some of the plane tickets to the hijackers was a client too. He closed the business shortly after.

 

[KnowWhy]

Dealers stopped buying silver because it was too high. $120/oz is nuts

 

[pmbug]

Dealers stopped buying silver because it was too high. $120/oz is nuts

Dealers don't care what the price is. They make money off of sale volume (what they buy from the public, they sell to refineries - vig between retail and wholesale market). When refineries stopped buying, it threw a monkey wrench into the wholesale market.

 

[Voodoo]

Dealers stopped buying silver because it was too high. $120/oz is nuts

Ya, that's a pretty silly argument. They still buy gold at $5,000 / oz right...

 

[pmbug]

Last week, @Sorenthek reported that GS claimed the US had 5 years worth of silver reserves. Knowing that the COMEX silver vault stock is down ~18% YTD, I got curious.

.@Sorenthek indicated that GS was referring to private vault stock - what I have been calling the USA's shadow inventory (because it's not publicly disclosed like COMEX vault stock). Included in the GS report is a graph:

If I'm reading that correctly, GS is claiming:

• US known silver reserves are on the order of ~23,000 metric tons (~739.5M ozt)
• US consumes about 6,200 metric tons of silver annually
• US imports about 4,200 metric tons annually

It's not clear from what I can see if that number is supposed to include COMEX vault stock or not. Assuming the conservative case that it does include the COMEX vault stock, we can subtract it from GS' total to calculate the total asserted shadow inventory. The COMEX currently has ~332.6M ozt in the vaults and ~70.2M ozt of that is allocated to the SLV ETF, the COMEX effectively has an upper bound of ~262.4M ozt free float (available) silver. The conservative conclusion is that Goldman is claiming the USA's shadow inventory is on the order of 477.1M ozt (or ~14,800 metric tons).

I checked Nick Laird's Gold Charts R Us site for USA import/export data and see that the annual net import is more or less on par with Goldman's 4,200 metric ton graph - at least it was circa 2008:

I mention this as it provides some insight into how accurate GS is being with their numbers/graph (not terribly precise, but "good enough for government work").

I decided to take a closer look at the issue. To start, let me update and extend my chart on COMEX silver deposits to cover the period since Jan 2025 through the present day (September isn't complete yet, but a minor incomplete data point for the month isn't going to matter for our purposes here) compared to the month-over-month average of deposits across 2021 through 2024:

As you can see in the chart, COMEX silver deposits ran unusually high during Jan-Apr 2025 as bullion banks were pulling silver from the LBMA for risk mitigation against uncertainty over USA tariff policy. That policy got publicly clarified in April and the LBMA draining waned and COMEX deposits returned to a normal flow for a few months.

September 2025 was an anomaly though and the huge inflow coincided with heavy buying from India. That caused stress in London which by mid-October was having trouble satisfying physical delivery obligations. This sparked the great reversal as COMEX withdrawals ran hot and silver flowed back to London in late October.

Curiously though, COMEX support for the LBMA slowed down in November and December. It looked like the bullion banks didn't want to return the entire silver ocean back to the LBMA. Unfortunately for the LBMA, India continued buying silver heavily and investors were piling into ETFs and by January, the LBMA's liquid free float was once again getting stressed. It looked to me like another LBMA technical default was imminent.

Fortunately for the LBMA, the silver market got monkey hammered in late January and the ~40% price drop flushed investors out of ETFs which returned vault stock back to the LBMA right when they needed it most. Very convenient!

Over the next couple of months (from February through March), the USA exported much more silver than what was withdrawn from the COMEX (to the cumulative tune of 540 metric tons). Then in April and May, the COMEX vault stock reports started reporting silver being magically "adjusted" into existence (ie. not deposited) to the tune of 78 metric tons which led me to compose my initial look into the shadow inventory issue posted up above.

It looked like COMEX deposits were normalizing in May with respect to the 2021-2024 MoM average and I was lax in keeping an eye on this issue. However, looking at it now, we can see that COMEX deposits continued to run less than the four year MoM averages through August. The cumulative total deficit of COMEX deposits to the four year MoM average from Jan though August has been 33.6M ozt or 1,046.4 metric tons. My assumption was that the difference was essentially being held as shadow inventory.

But that isn't a complete picture. Consider the following graph:

In this graph, the red line represents the four year MoM average less actual COMEX deposits as just discussed.

The blue line represents the monthly USA silver exports less actual COMEX withdrawals (NOT "deliveries" - actual, physical vault stock withdrawals). When the blue line is positive, the USA exported more silver than what was withdrawn from the COMEX and I assume the difference was pulled from the shadow inventory.

The green line represents the shadow inventory's monthly gain or loss when combining the month's inflow (red line) and outflow (blue line).

From January through July, the cumulative shadow inventory gain that we can see from public data (considering both inflows and outflows) appears to be on the order of 15.788M ozt or 491 metric tons. That's not nearly sufficient to support Goldman's 14K metric ton minimum case claim. Let's dig a bit deeper.

.@Sorenthek et al have previously posited that the USA has essentially commandeered silver mining output from Latin America and the proceeds are being held outside of the COMEX system in shadow inventory. But how much silver is this though?

I asked Grok to estimate LATAM silver mining output YTD considering everything that is known about ~15% reduced silver mining output for the region. I paired Grok's monthly estimates with my shadow inventory gain/loss analysis in the following chart:

The blue line is Grok's estimate for LATAM (Mexico, Peru and Chile) monthly silver mining output which runs from 22M ozt to 27M ozt per month.

The red line is the shadow inventory gain/loss I calculated in my big picture graph (the green line in my big picture graph).

The green line is the percentage of the LATAM production represented by the observed shadow inventory gain/loss.

We don't know how much of LATAM production has actually been commandeered and flowing into the USA's shadow inventory vaults, but it would seem that somewhere in the neighborhood of 15-20% would be sufficient to explain the observed shadow inventory flows. If the USA is commandeering a much greater percentage of LATAM mining output, the USA's shadow inventory could be pretty significant indeed.

We can estimate a theoretical maximum if we consider that the USA commandeered 100% of LATAM silver mining output (not really realistic IMO, but this is a thought exercise). Just considering January through July, the shadow inventory gain from LATAM mining inflow (~183.2M ozt) minus USA export leakage (beyond COMEX withdrawals) (~7.4M ozt) over the same period yields a theoretical maximum gain of ~175.8M ozt or ~5.5K metric tons.

We can apply a percentage factor to the LATAM mining output to estimate something more reasonable. If the USA were commandeering 20% of LATAM mining, we would estimate 36.6M ozt LATAM mining inflow less 7.4M ozt export leakage yielding a ~29.2M ozt (~910 metric tons) shadow inventory inflow from January to July.

My data only runs from January to July. For Goldman's claim to make sense, the USA would need to have commandeered LATAM mining output much earlier than Jan 2026 - since at least the start of 2025 or earlier. I don't know exactly when the shadow inventory build up started, but I'm having a difficult time squaring the circle on based upon what I can see.