Tokenized Treasuries drawdown: who redeemed, which issuers grew, and the T-bill vs DeFi yield spread

Partial answer, because most of BUIDL's decline sits on chains I could not trace. Window: transfers since 2026-09-16 (Surf indexed Ethereum and BNB Chain transfers), against rwa.xyz per-asset 30-day changes read Oct 16. (1) Size of the declines: BUIDL $2.17B, -17.62% over 30d (about -$464M). Supply is spread across Solana 871.6M, Avalanche 481.5M, two Ethereum listings (256.1M and 195.8M), Aptos 157.0M, BNB 146.8M and smaller chains. USYC $2.38B, -8.43% over 30d (about -$219M); 98% of supply is on BNB Chain (2,056.9M) and 34.4M on Ethereum. (2) BUIDL on Ethereum (contract 0x7712c34205737192402172409a8f7ccef8aa2aec): net burns to the zero address 16.31M, only about 3.5% of the dollar decline. Largest net decreases, all unlabelled EOAs in Surf labels: 0xf2f489b17effb79b3f389df66728c0efe73d510f -11.07M (11 transfers; Etherscan shows it was funded from an OKX deposit address, not proof of identity); 0xf19a4fcc98c96a063ab06f3f5781e34a845a49e8 -6.39M; 0x9ef9f05250d8d484358e335f14be2a06767a8192 -5.03M; 0x12c0de58d3b720024324d5b216ddfe8b29adb0b4 -4.23M; 0x1e695a689cf29c8fe0af6848a957e3f84b61fe69 -0.83M. The largest net increase was unlabelled 0xa318d9af160110330d7e3b56cb7da0d9f7d6d67b +10.72M. Crypto-native share of the BUIDL decline: UNAVAILABLE. The traced Ethereum holders are unidentified, and the Solana, Avalanche and Aptos legs, which hold most of the supply, are outside my data access. (3) USYC on BNB Chain (contract 0x8d0fa28f221eb5735bc71d3a0da67ee5bc821311, 6 decimals): 42,000,154 USYC burned (about $47.8M at about $1.138 NAV, i.e. rwa.xyz value divided by total supply) in 5 transfers from 0x87433fec6f8d9df13d1e17c4b11364ecd2e93a51, which Surf labels as entity Binance (cex). USYC on Ethereum (0x136471a34f6ef19fe571effc1ca711fdb8e49f2b): net burns 2.74M; the largest decrease is 0xdd82875f0840aad58a455a70b88eed9f59cec7c7 -2.93M, labelled Usual Treasury (stablecoin issuer). So about $51M (about 23%) of USYC's $219M decline is traced, and essentially all of the traced part is crypto-native: one exchange and one stablecoin issuer. The other about 77% is untraced. The BNB transfer index shows only these burns over 30d, so it may be incomplete, or the rwa.xyz 30-day baseline may include flows outside the window. Overall: USYC evidence leans toward crypto-native redemptions; BUIDL attribution is unresolved. Not a trade signal.

Rate-side evidence for check 4bf0f312 (T-bill vs DeFi stablecoin yield spread). Observation: US September CPI printed 3.7% YoY (from 3.4%) and +0.6% MoM; core 2.6% YoY, +0.3% MoM (BLS, Oct 14). Kalshi's October Fed decision market now shows hold 71% and hike 25bps 25%, versus hold 81% and hike 17% in the Oct 13 snapshot already cited, so the hike tail widened. Caveat: that market had zero trades in the prior 7 days, so both readings may be stale quotes. Crypto deleveraged the same day: roughly $483M of BTC and ETH futures were liquidated in 24 hours and BTC fell below $81K (secondary reporting citing CoinGlass). Inference, not yet tested: hotter inflation and a larger hike tail support short-dated T-bill yields staying high or rising, which would favour the tokenized Treasury side of the spread if DeFi stablecoin borrow demand keeps falling with deleveraging. This does not measure the rwa.xyz 7-day APY or any DeFi rate directly; check 4bf0f312 still needs those readings at two or more points in the window. Not a trade signal.

The spread does not favour tokenized Treasuries, and it moved further against them over the window. (1) Tokenized T-bills (rwa.xyz category 7-day APY): 3.68% (data as of Oct 12) and 3.36% (as of Oct 16, shown as -8.33% vs 7d ago). (2) DeFi stablecoin yields, from on-chain reads at Ethereum blocks 26118198 (2026-10-04 09:49 UTC) and 26205644 (2026-10-16 14:18 UTC). Aave v3 USDC supply: currentLiquidityRate 3.695% to 3.772% APR (about 3.76% to 3.84% APY). Aave v3 USDT supply: 3.686% to 3.267% APR (about 3.76% to 3.32% APY). sUSDS savings rate (ssr): 3.60% to 3.80% APY. It was raised inside the window; the change date isn't pinned, but it is consistent with the Oct 12 executive. Realized sUSDS share-price growth over the 12.19 days was 3.75% APY. sUSDe realized share-price growth Oct 4-16 was about 4.91% APY. (3) Current snapshot only (Surf yield ranking, Oct 16): Compound v3 USDC 5.51%, Maple USDC 5.20%, Maple USDT 5.54%, Spark savings USDT 3.75%, sUSDe 4.83%. Result: on Oct 16 the 3.36% tokenized T-bill APY sits below Aave USDC (3.84%), sUSDS (3.80%) and sUSDe (about 4.8-4.9%), and roughly at Aave USDT (3.32%). From Oct 12 to Oct 16 the T-bill measure fell 0.32 pts while sUSDS rose 0.20 pts and Aave USDC rose 0.08 pts, so the gap widened against tokenized Treasuries. Only Aave USDT moved toward them. This does not support renewed inflows on a yield basis. Uncertainty: the rwa.xyz figure is a category-wide 7-day measure, not a single fund's yield, and its decline despite hike-odds swings may reflect product mix. DeFi yields carry smart-contract, liquidity and (for sUSDe) funding-rate risk, so this is a raw comparison, not risk-adjusted. Aave rates are instantaneous point readings that move with utilization. The two rwa.xyz points (Oct 12 and 16) do not exactly match the on-chain points (Oct 4 and 16). Not a trade signal.

The premise that all five products are growing does not hold on rwa.xyz's own 30-day figures. Only two grew, and the growth that can be traced is mostly net Ethereum supply, not new-chain deployments. (1) Per-asset rwa.xyz pages, read Oct 16 (30-day change): USDY $2.36B (+6.29%); USTB $606.0M (+11.94%); JLTXX $738.27M (-8.28%); JTRSY $345.00M (-47.80%). The UMIU per-asset page did not return asset data (UNAVAILABLE). The up arrows for JLTXX and JTRSY on the dashboard table, cited at promotion, do not match their per-asset 30-day figures. Those arrows may reflect a different window. (2) USTB: Ethereum holds 53.33M of 53.89M tokens (99%; Plume 0.37M, Solana 0.19M). Ethereum totalSupply() on the USTB contract (0x43415eB6ff9DB7E26A15b704e7A3eDCe97d31C4e, 6 decimals) went from 47,843,162 at block 25986552 (2026-09-16) to 53,953,416 at block 26205662 (2026-10-16), +12.77%. That is about +$68.7M at about $11.24 per token, versus a reported value gain of about +$64.6M. Essentially all of USTB's growth is net Ethereum minting, consistent with net subscriptions rather than new chains. (3) USDY: Ethereum totalSupply() (0x96F6eF951840721AdBF46Ac996b59E0235CB985C) went from 1,046.63M to 1,092.72M over the same blocks, +46.09M (+4.40%). That is about +$53.3M at about $1.157 per token (rwa.xyz value divided by summed chain supply), roughly 38% of USDY's about +$139M 30-day value gain. NAV accrual (about $7M a month at about 3.8% APY) explains a little more. The remainder sits on non-Ethereum chains: Stellar 467.5M, SEI 225.8M, Solana 156.5M, BNB 88.5M and others. rwa.xyz shows only direction arrows per chain, so I could not split it into new subscriptions versus bridging versus new deployments. No per-chain launch dates appear on the pages. (4) JLTXX is almost entirely on Ethereum (738.27M of supply; Solana 206 tokens) and shrank. JTRSY is mostly Ethereum (291.8M) with small balances on Base, Stellar, Monad, BNB and Plume, and shrank. Uncertainty: Ethereum totalSupply changes can include bridge mints and burns, not only primary subscriptions. The USDY per-chain split relies on rwa.xyz supply figures without dates for the 30-day baseline. Issuer announcements were not reviewed. Not a trade signal.