Which COMEX warehouse movements in 2025 were attributed to arbitrage versus end-user physical demand, according to public delivery notices?
Executive summary
COMEX public delivery notices and warehouse reports show two distinct 2025 phases: a massive inbound wave in Q1 that multiple analysts attribute to tariff-driven arbitrage (metal moved from London/Swiss vaults to New York to capture the spread) and a subsequent period from April onward where elevated delivery notices coincided with shrinking warehouse stocks that market commentators interpret as genuine end‑user/physical demand and withdrawals (noted in CME and independent reporting) [1] [2] [3]. The public notices document volumes and timing but do not, on their face, reveal each claimant’s commercial intent, so attribution relies on contemporaneous market structure signals and industry reporting [4] [5].
1. Q1 inbound surge: delivery notices match an obvious arbitrage play
Between December 2024 and March 2025 COMEX warehouse inventories rose from roughly 17.1 million ounces to a peak near 43.3 million ounces—a 153% increase—while contemporaneous delivery‑notice volumes and registration activity spiked, and market participants described daily airlifts from Swiss refineries into New York as traders executed a London‑to‑COMEX convergence trade created by tariff risk and a spot‑futures spread [1] [6]. Multiple analysts and newsletters interpreted the timing and pattern—large eligible registrations and rapid inward flows coupled with widening spreads—as classic arbitrage rather than genuine industrial or retail end‑user demand [1] [3].
2. Public delivery notices show elevated volumes but not buyer motives
The CME publishes daily and monthly COMEX delivery notices and warehouse/depository stock reports that record which contracts stand, which bars are registered, and net stock movements, and 2025’s notices documented unusually large delivery activity and warranting events [4]. Those notices prove the scale and timing of movements—how much changed hands and when—but they do not encode counterparty economic purpose; therefore, attributing each movement definitively to arbitrage or end‑use requires inference from market context [4] [5].
3. April onward: delivery notices plus falling stocks point to physical withdrawals
After the March 2025 inventory spike, public reporting shows warehouse stocks declining from spring into later months even as delivery notices remained elevated, a pattern analysts interpret as metal leaving COMEX for end‑users or external custody rather than continuing accumulation for arbitrage [2] truegoldrepublic.com/blog/comex-deliveries-a-potential-game-changer-for-the-gold-market" target="blank" rel="noopener noreferrer">[7]. Commentators argue the change from rising eligible registrations to falling total stocks—while delivery notices stayed high—signals that more metal was being claimed and physically removed, consistent with true physical demand rather than mere repositioning inside the exchange system [2] [3].
4. Divergent narratives: arbitrage, physical demand, and vested interests
Industry outlets broadly agree about the Q1 arbitrage story but diverge on what followed: some warn of a “vault drain” and systemic stress as end‑users and investors claimed registered metal (discoveryalert, SchiffGold, AInvest coverage), while others frame later flows as normalization after an unusual arbitrage event (seekingalpha, truegoldrepublic) [3] [8] [9] [7]. Those differences often reflect editorial slants—some publications emphasize systemic risk and physical scarcity, which can amplify alarm, while others stress market mechanics and that fewer than 2% of contracts typically end as physical deliveries historically [10] [11].
5. What public delivery notices can and cannot confirm
Public COMEX delivery notices and CME warehouse reports indisputably document the timing and magnitude of 2025 movements—Q1’s inbound deluge and the post‑April drawdown are visible in the data [1] [4] [2]. However, these notices do not, by themselves, label transactions “arbitrage” or “end‑user demand”; that attribution rests on corroborating price behavior (spot‑futures spreads), reportage of cross‑border shipments (Swiss to U.S. accounts), and analyst interpretation, all of which point to Q1 arbitrage and later genuine physical withdrawals but stop short of proving intent for every single movement [1] [3] [5].