Fed +25 bps: BUIDL moved 20 in one day and Hyperliquid 0

The Fed raised the intervention rate by 25 basis points on September 16, 2026, and by the following day BUIDL, BlackRock’s tokenized money market fund, had already passed through 20.0 of those 25 basis points, reaching 23.1 by October 4; meanwhile, Hyperliquid’s native lending was still charging exactly 5.000% on October 5, for the seventh consecutive reading, and Sky’s ssr() remained at exactly 3.60000% for the sixth. What separates them is the source of the rate they pay: only a portion of the on-chain dollar has a direct line to the Fed. Where the rate is the yield of a Treasury bill portfolio, the hike flows in automatically, and its speed depends on the portfolio's weighted average maturity. Where the rate is a constant written into a contract (a formula, a credit quota, or a number that only changes via vote), it never enters, and what moves instead is pool utilization. Then there is a fourth case: on September 30, 2026, the USDC rate for Aave V4 on Arc—Circle’s permissioned chain—hit 10.57% within a 2-hour and 24-minute window during which a single account withdrew 73 million USDC and then returned it. This article classifies on-chain dollar rates by their fixing mechanism, read directly from each protocol's contract or API, and places them alongside the spread against a four-week Treasury bill, which paid 3.89% on October 1, 2026.

What did the September 16, 2026 Fed hike move in SOFR, EFFR, and the four-week bill?

The Federal Open Market Committee statement from September 16, 2026, literally states that the committee "decided to raise the target range for the federal funds rate by 1/4 percentage point to 3-3/4 to 4 percent," approved by a 12-0 vote. This is a 25-basis-point hike in the target range, up to 3.75-4.00%.

The rates the Fed controls almost mechanically performed as expected. The EFFR (Effective Federal Funds Rate, the weighted average of overnight interbank lending) rose from 3.63% on September 16, 2026, to 3.88% the following day and has remained there, unchanged, across the twelve daily observations from September 17 to October 2, 2026: exactly 25 basis points, according to the New York Federal Reserve series. The rate the Fed pays banks for their reserves did the same: 3.65% on September 16, 2026, 3.90% since the 17th, and still 3.90% on October 5. The SOFR (Secured Overnight Financing Rate, the rate for repos collateralized by Treasury debt, which is the benchmark for the wholesale dollar) rose from 3.62% on September 16, 2026, to 3.85% on the 17th, touched 3.90% on September 25, 28, and 30, closed October 1, 2026, at 3.87%, and October 2 at 3.88%, the last published close: a cumulative rise of 25 basis points by October 1 and 26 by October 2.

The four-week Treasury bill behaved differently, and the details matter for everything that follows. According to the FRED DTB4WK series, it went from 3.82% on September 16, 2026, to 3.89% on October 1, 2026: only 7 basis points. However, from September 8 to 16, 2026, before the committee voted, it had already risen 16 basis points, from 3.66% to 3.82%. The bill anticipates the decision; the repo reflects it the day after. Anyone using the bill as a benchmark must understand two things: that a large part of the adjustment occurred before the meeting, and that the bill itself moved 5 basis points in a single week—from 3.85% to 3.90% between September 25 and October 1, 2026, with a maximum daily jump of 4 basis points—which is significant when the spread being measured is 29 basis points.

The figures in this article use the four-week bill as a reference, at its October 1, 2026 close, which is the latest FRED publishes for that series: the October 2 close, the next business day, is not yet in DTB4WK. When the next one appears, all spreads in the table will move in lockstep with the bill's movement that day, without changing the row order. The four-week segment is used because it is what the portfolios of the tokenized money market funds in the table replicate. The three-month bill reached 4.10% on September 28, 2026, and reversed the move: it closed October 1 at 4.00%, only 11 basis points above the four-week bill, whereas on September 28, 2026, three days earlier, they were 21 basis points apart.

Who sets the on-chain dollar rate: BUIDL, Hyperliquid, Aave V4 on Arc, Sky, or Ondo?

The classification organizing this article is based on the nature of the rate paid, rather than by protocol or asset type. There are four ways an on-chain interest rate can move, and only one is directly related to what the Fed does.

• The rate is the yield of a portfolio. The contract sets nothing: it distributes what a basket of Treasury bills and repos produces. The hike flows in automatically, without authorization, and the speed of entry is determined by the portfolio's weighted average maturity. This is the case for BUIDL (BlackRock’s tokenized money market fund, issued by Securitize), USYC (Circle), and USDY (Ondo).
• The rate is a constant written into the contract. It could be a formula with a flat segment (Hyperliquid), a curve with a utilization kink (HyperLend), that same curve plus a credit quota set by governance (Aave V4 on Arc), or a single number that only changes if there is a vote (Sky’s ssr() and SparkLend’s reserve parameters). None of these have a direct channel to the Fed. What moves here is utilization, and utilization does not obey the Federal Open Market Committee.
• The rate is a periodic decision by the issuer. Ondo states on its own page that "APY is set monthly by Ondo." Excluded from the table due to lack of a primary source is the Coinbase USDC reward, which Digital Today, citing CoinPost, placed at 3.75% on September 17, 2026, for subscribers of Coinbase One, its paid service. The channel exists, but it is subject to a calendar and commercial criteria.
• The rate is moved by deposit concentration. In a market where one depositor holds half the balance, their entry or exit shifts utilization enough to cross the curve's kink and reprice everyone for as long as it lasts. Tracked on September 30, 2026, in Aave V4 on Arc. HyperLend saw a similar peak that day, though the mover was not tracked.

It works like a thermostat. If the sensor is outside, the boiler reacts to the outdoor cold with a delay based on the wall's thickness; if the sensor is in the living room, the boiler reacts to people being in the room, and the outdoor temperature is irrelevant. A rate linked to a bill portfolio has its sensor outside; a rate linked to a utilization formula has it inside the house. The limit of the analogy is that utilization does respond, with a delay and indirectly, to money being more expensive off-chain: if bills pay more, some depositors leave, utilization rises, and the depositor rate rises with it. But that path goes through utilization and is triggered by a depositor's exit, with the Fed at the far end of the chain.

The following table reads, protocol by protocol, the written rate and whether it moved. The pass-through column measures how many of the 25 basis points the Fed raised on September 16, 2026, appear in that rate.

Two clarifications regarding the table, as it mixes two measurement methods. Values for BUIDL, Maple, and the USDY series come from the DefiLlama daily pool series and correspond to the last closed day, October 4, 2026. The data point from the download day itself or October 3 is not used, because the October 3 point returns zero for the three BUIDL pools: with the October 4 close, all three return to 3.8093%, 3.7778%, and 3.4673%, exactly the same values as October 2. Values for Hyperliquid, Aave V4 on Arc, HyperLend, Sky, Maple (its own API), and Ondo are proprietary readings from October 5, 2026, taken from the contract or API between 16:55 and 16:58 UTC, two days after the previous reading. There is up to a one-day lag between the two bases, which is why a delta from the aggregator series is not subtracted against a point-in-time contract reading.

Why did BlackRock’s BUIDL pass through 20 bp in one day while Ondo’s USDY waited until Oct-1-2026?

The three BUIDL pools tracked by DefiLlama climbed the same step on the same day: on September 17, 2026, the day following the committee's decision, the yield rose from 3.578 to 3.778% in the Ethereum institutional class, from 3.547 to 3.747% on Solana, and from 3.237 to 3.436% in the Ethereum open class. Twenty basis points at once across all three, and the rest—up to the 23.1 cumulative by October 4, 2026—arrived in small adjustments over the following two weeks. That represents 92.5% of the Fed hike: 23.1 of the 25 basis points, a proprietary calculation using both ends of the series.

The coincidence across three pools on two different chains is what allows the step to be attributed to the Fed rather than a movement of money. The Solana pool held $967.6 million on October 4, 2026, and did not participate in the unwinding of the Grove position—the credit vehicle for the Sky ecosystem—which emptied the Ethereum institutional class in the same fortnight, from $524.5 to $239.9 million; yet it still rose the same 23.1 basis points. Neither BlackRock nor Securitize publish a searchable daily rate, so the figure comes from the aggregator; the evidence that it is the Fed is the triple coincidence, and the primary source for the same event is the Treasury bill.

Ondo’s USDY belongs to the same family and took fifteen days to move, but move it did. Until September 30, 2026, its page declared a 3.60% APY; on October 1, 2026, in its monthly reset, it moved to 3.75%, where it remained on October 5. That is 15 of the Fed's 25 basis points, with a two-week delay and by issuer decision. The aggregator series still lags behind: on October 4, 2026, it marked 3.62%, which is the slow path of the same movement.

The two brakes on USDY are found in its own portfolio sheet. The first is maturity: in the dashboard dated October 2, 2026, the $2,305,922,902 in assets backing the $2.06 billion of USDY in circulation have a weighted average maturity of 164.51 days, with 90.08% in Treasury bills—$2,077,275,054—with an average life of 182.51 days at 4.06%, and a collateralization of 132.54%. A portfolio with nearly six months of average maturity cannot reprice in a day: only the portion that matures and is reinvested reprices. The portfolio's average maturity is the variable that orders this family. BUIDL holds paper with maturities of days and repriced in one day; USDY holds paper of nearly six months and needed two weeks. And it is lengthening: that same average maturity was 147.42 days in the September 28 dashboard and 144.9 days in the October 1, 2026 dashboard.

The second brake is that the rate the holder receives is a number Ondo sets each month, not the yield the portfolio produces. In the October 2, 2026 dashboard, that portfolio yielded 4.01% to maturity while the published APY remained at 3.75%: 26 basis points that the issuer keeps between both sides. Ondo’s page does not keep a history of this dashboard, so the margin series consists of three of our readings of the same table, each with the date the dashboard itself declares: 24 basis points in the September 28, 2026 reading (3.84% yield vs 3.60% APY), 21 in the October 1 reading (3.96 vs 3.75), and 26 in the October 2 reading (4.01 vs 3.75). In two days of dashboard data, the portfolio yields 5 basis points more and lasts 19.6 days longer, yet the holder receives the same 3.75%.

Circle’s USYC is from the same family but is excluded from the table for measurement reasons: its series moves 50 basis points in nine days without an identifiable step, from 2.985% on September 20, 2026, to 3.482% on the 29th, and returns to 3.070% at the October 4 close. Measured between September 16 and October 4, 2026, that series shows −12.2 basis points for an asset backed by Treasury bills, which is the opposite sign of the mechanism's intent.

What does the Hyperliquid native lending formula fix and what does it leave loose?

The Hyperliquid native lending book, which finances the portfolio margin of its perpetuals (futures without expiry), calculates the borrower rate using the formula in its documentation: 5% plus 4.75 percentage points of rate for every percentage point of utilization above 80%. Below that kink, the rate is flat at 5.000% and there is no macro variable within the expression. On October 5, 2026, the API returned, for the USDC reserve, a lending rate of 5.000% with 536,691,680 USDC supplied, 413,809,717 borrowed, and a utilization of 77.1038%. The same 5.000% it charged on September 16, before the meeting, and the same as the seven proprietary readings taken between September 19 and October 5, 2026.

What did move in Hyperliquid is the depositor rate, and it is worth looking at closely because it is the trap in this map. The depositor receives the lending rate multiplied by utilization and by 0.9—the protocol retains 10% of the interest. With the lending rate locked at 5.000%, this leaves the depositor receiving 0.045 times utilization and nothing else: an arithmetic identity, which on October 5, 2026, gives 0.045 × 77.1038% = 3.4697%, the exact value returned by the API. The last two readings prove this on their own, as they are separated by two days: on October 3, 2026, utilization was 74.750% and the deposit paid 3.3637%; on October 5, utilization was 77.1038% and the deposit paid 3.4697%. The deposit rose 10.6 basis points in two days with the lending rate stationary at 5.000% and without the Fed meeting.

The seven readings draw the full band: 3.1977% with 71.06% utilization on September 19, 2026; 3.2649% with 72.55% on September 21; 2.9733% with 66.07% on September 25; 3.1703% and 3.1393% in two readings on September 30 with 70.45% and 69.76%; 3.3637% with 74.750% on October 3, and 3.4697% with 77.1038% on October 5. A 50-basis-point range drawn entirely by the credit demand of the book itself, with the floor and ceiling ten days apart. Anyone seeing the Hyperliquid depositor rate rise after a Fed hike is looking at utilization, not the Fed.

The formula also has a ceiling worth knowing before comparing it to a bill. If utilization reached exactly 80%, the edge of the flat segment, the Hyperliquid USDC depositor would receive 0.05 × 0.8 × 0.9 = 3.60%, still 29 basis points below the 3.89% of the four-week bill on October 1, 2026. To match the bill, one must exit the flat segment: solving the full formula, the required utilization is 80.0836%, which with the 536.7 million USDC supplied on October 5, 2026, requires approximately 15.99 million USDC more in live loans. Both numbers are proprietary calculations based on the published formula. As long as utilization stays in the 66% to 77% band where it has lived since September 19, 2026, the Hyperliquid deposit moves between 2.97% and 3.47%, below the 3.89% bill rate throughout the entire range.

What moved the Aave V4 on Arc USDC rate: the credit quota or a single account?

Both, in very different proportions. On October 5, 2026, at Arc block 24,421,640 (16:57:23 UTC), the Aave V4 hub—the contract that concentrates liquidity and distributes it to the lending modules, or spokes—returned a drawnRate—the rate borrowers pay—of 2.14% annually on 67,323,195 USDC of live debt and 143,058,437 deposited, with a utilization of 47.06%. On September 24, 2026, in our reading of the same hub that day, that same rate was 1.62% with a utilization of 35.56%. That is a 52-basis-point rise, and none of it comes from the Fed.

The cause of those 52 basis points is dated and has a primary source: the credit quota of the main spoke was increased from 51 to 100 million USDC, executed on-chain on September 25, 2026, as part of round 18 of parameters from LlamaRisk, the firm that proposes risk parameters for Aave markets. With more quota, more debt entered—from 51,003,620 to 67,323,195 USDC, 16.32 million more—and utilization rose with it. The curve was not touched: the four parameters of the Arc USDC rate strategy on October 5, 2026, are the same as on September 25 (optimal utilization 90%, base rate 0, first slope 4.10%, second slope 10%), and with utilization below the optimal, the first slope reproduces the read rate: 4.10 × 0.470599 / 0.9 = 2.144%.

The deposit side, however, remains where it was. On October 5, 2026, there were 143,058,437 USDC deposited: 0.25% below the 143,416,920 on September 25, 2026, and 0.013% above the 143,040,169 from the October 3 reading, 18,268 USDC more in two days. The quota expansion brought new debt without the deposit side moving.

This data matters because there is a reading from September 30, 2026, which, taken alone, describes a different market. It fell within a round trip: the market's largest depositor withdrew 73,008,994.71 USDC from the hub and returned 73,009,000.00 2 hours, 24 minutes, and 42 seconds later, on that same September 30, 2026. During that window, the market deposit appeared as 70.09 million USDC—half—and utilization was above the 90% kink, where the second slope dominates. The first reading that day showed 96.45% utilization with the borrower rate at 10.57% and the depositor rate, by proprietary calculation, at 9.17%; twenty-nine minutes later, utilization was 95.71% and the rate had dropped to 9.83%. The 2 hours and 24 minutes measure the account's absence and the period the rate was above the kink; within that window, the rate already moved 74 basis points between two readings separated by half an hour. As soon as the account returned the money, the market returned to its place: 2.14% on October 3, 2026, and the same 2.14% on October 5. Measured against that return level, the account moved the rate by 843 basis points; the 895 separating 10.57% from the 1.62% on September 24 include the 52 from the quota and are not additive to them.

This episode represents the fourth mechanism on the list, and the only one not written into any parameter. A market in which one depositor holds half the balance has a rate that the depositor can move at will for as long as they choose to be out, without asking governance for permission and without a single line of the curve changing. On the same September 30, 2026, HyperLend saw a similar small peak, with no tracking of who moved it: its borrower rate hit 9.24% with 92.22% utilization, above its 90% kink, and by October 5 it was back to 4.77% with 85.47%, below it. The aggregator's closed point for HyperLend on that same September 30 gives a 3.74% deposit rate, because it is taken at the end of the day, after the peak had passed.

Both sources measure the same reserve: the aggregator takes its deposit rate from the same liquidityRate in the contract that the API returns, and what it calls total value locked is the available liquidity, not the total deposit (7.0 vs 57.0 million USDC on October 3, 2026). That day, the API returned 3.780% at 20:00 UTC and the aggregator's closed point 3.794% three hours later. On October 5, 2026, the API gives a deposit rate of 3.6728% and a lending rate of 4.7748% with 85.47% utilization, and the curve reproduces them without any quota term: 4.30 + 0.50 × 85.47 / 90 = 4.775%. The same curve explains the September 30 peak, with utilization above the kink: 4.30 + 0.50 + 20 × (92.22 − 90) / 10 = 9.24%.

The Arc reading, by contrast, is verified by repetition. The two readings on October 3, 2026, are separated by 6.5 hours and match to the fourth decimal, 2.13652% and 2.13656%; the October 5 reading, two days later, gives 2.14461%, 0.8 basis points more. With the rate steady over five days and the deposited balance rising 0.013%, 2.14% describes the market. The 10.57% on September 30, 2026, was also a correct contract reading, and what it measured was the 2-hour and 24-minute window in which its largest depositor was away.

What is ssr() in Sky’s sUSDS and who has to vote to move it?

Sky’s sUSDS is the pure example of a rate by vote, and the largest on the map by size. On October 5, 2026, at Ethereum block 26,127,507 (16:57:35 UTC), its contract returned an ssr() equivalent to 3.60000% APY on 4,300,419,009.13 sUSDS, which at that block's chi() is 4,781,164,271 USDS, read directly on-chain. It is the sixth proprietary reading with the same value to the fifth decimal: two from October 3 separated by 6.5 hours, one from October 5 two days later, and the intermediate ones from September 25 and 30, 2026. That number is a constant that a contract applies per second until a Sky executive vote replaces it with another. Only two things enter the calculation of accrued interest: that constant and time. The Fed has no way to touch it.

What moved in those two days was the size, and in the direction that makes the observation stronger: the 4,659,118,358 USDS from the second October 3, 2026 reading rose to 4,781,164,271 on October 5, a +2.62% increase—122 million USDS more—with the rate identical. Money enters the contract and the rate it pays remains the same, because the rate does not depend on how much money is inside.

The aggregator series for sUSDS in 2026 shows five changes: from 4.00% to 3.75% on March 9, to 3.65% on April 22, to 3.60% on May 26, to 3.52% on July 23, and back to 3.60% on September 3. The last of those changes, an 8-basis-point upward move, arrived thirteen days before the Fed meeting, with the hike not yet voted on. Since then, and with SOFR 26 basis points higher, the ssr() has not moved once. The channel exists—Sky has used it five times in 2026—but it requires a ballot, and between September 16 and October 5, 2026, that ballot was not written.

A sister case within the same ecosystem clarifies the difference between the fixed rate and the earned rate. The yield on USDS deposited in SparkLend—the Sky ecosystem's own lending market—rose 24.2 basis points, from 2.318% on September 16, 2026, to 2.561% on October 4, in a single step on September 21. The explanation lies in the same point of the series: the pool's total value locked went from $479.0 million to $447.4 between the September 20 and 21, 2026 closed points of the DefiLlama daily series, the same two points where the APY jumps from 2.31841% to 2.56067%. With debt stationary, a drop in deposits raises utilization and, with it, what the remaining depositors earn. The same pattern as in Hyperliquid: the governance-set rate stayed where it was, and the rate the depositor earns rose with utilization.

How much less do Hyperliquid, Arc, and Sky depositors earn compared to a Treasury bill?

This is the practical consequence of the map: almost everything lending dollars in these markets earns less than a four-week Treasury bill. The two exceptions in the table are Maple’s syrupUSDC and Ethena’s sUSDe, and neither lends dollars against Treasury bills: Maple pays the yield of a private credit portfolio to institutions and Ethena that of a hedged perpetual derivatives position, so their premium over the bill pays for a different risk and falls outside what this map measures as pass-through. The following table shows, for each asset, what the dollar lender earned and its distance from the October 1, 2026 bill and SOFR. The measurement base column states, row by row, where each figure originates.

Put into monetary terms, the spread has a concrete size. Hyperliquid native lending USDC depositors had 536,691,680 USDC in the book on October 5, 2026, earning 3.4697%, 42 basis points below the bill: a $2.26 million annual difference, proprietary calculation based on supply and rates on that date. Sky sUSDS holders had 4,781,164,271 USDS in the contract earning 3.60%, 29 basis points below: $13.87 million per year. And Aave V4 on Arc USDC depositors had 143,058,437 USDC earning 0.9083%, 298 basis points below: $4.27 million per year on a market 33.4 times smaller than Sky’s. All three figures measure the same thing: the annual cost of holding the dollar in a contract whose rate is set by a formula, a quota, or a vote, with the Treasury bill as the comparison term.

Two caveats regarding these spreads. The first is that the Hyperliquid depositor is not just earning a rate: their deposit serves as collateral for portfolio margin, so they are comparing a bill against an instrument that also enables them to trade. The book's USDC supply grew from 427.9 million on September 19, 2026, to 536.7 on October 5, earning between 2.97% and 3.47%: depositors accept the spread, and they accept it with the book 25.4% larger than on September 19, 2026. The second is that Maple’s syrupUSDC appears with two figures because it has two: its own API returned 4.9037% on October 5, 2026, and the aggregator series 5.1474% on the 4th, a 24-basis-point difference between two measurements of the same thing, and both are provided here because there is no third party to arbitrate.

Why didn't the Fed hike make leverage more expensive on Hyperliquid?

The consequence is symmetrical. A rate hike does not squeeze leveraged positions in the Hyperliquid native book as it would a credit indexed to a reference rate: the cost of financing margin there was 5.000% before the September 16, 2026 meeting and remained 5.000% on October 5, with the EFFR 25 basis points higher. Where the rate depends on the pool's local capacity, however, the borrower pays a bill the Fed did not send: 52 basis points more in Aave V4 on Arc USDC between September 24 and October 5, 2026, due to a quota expansion, compared to the 0.25 percentage points the Fed raised.

The September 30, 2026 episode adds the risk that appears in no rate table. For 2 hours and 24 minutes, an Arc borrower who rolled over their position saw a rate between 9.8% and 10.6% instead of the 2.14% they pay on October 5, 2026, and the cause was neither the Fed nor a governance vote, but the fact that the largest depositor was away. In a market with such concentration, the rate a borrower pays depends on when that depositor enters and exits, and that variable is not listed in any of the four published curve parameters.

Looking ahead, the probability priced by the betting market for an October hike has dropped to less than a third of what it was ten days ago. On October 5, 2026, Polymarket gave a 20.5% chance of a 25-basis-point hike at the October 28 meeting and 78.5% for no change; the series for that market went from 66.5% on September 25 to 33.5% on September 30 and 17.5% on October 3, with a 3-percentage-point rebound in the last two days. For the December 9, 2026 meeting, a 25-basis-point hike was priced at 74.5%, and "another Fed hike in 2026" at 77.5%.

What would have to happen at BUIDL, Hyperliquid, and Sky after the Dec-9-2026 meeting for the map to be false?

With the October hike priced at 20.5% on October 5, 2026, the useful test for this map has shifted to the December 8 and 9, 2026 meeting, with re-measurement a week later, on December 16. What we should see is this.

• That the side that passes through, passes through. If there is a hike on December 9, 2026, and the three BUIDL pools have not risen by at least 15 basis points by December 16, the bill portfolio family ceases to behave like a bill portfolio and the classification falls apart on that side.
• That the side that does not pass through, continues not to. If there is a hike on December 9, 2026, and by December 16 Sky’s ssr() has risen without a vote, or Hyperliquid’s lending rate has moved from 5.000% with utilization below 80%, then a channel to the Fed exists that this map does not see.
• That the measurement is correct. The threshold in the previous paragraph is measured against the Hyperliquid lending rate, or against the depositor rate at constant utilization. Measured against the depositor rate alone, it can detect nothing: with the lending rate locked at 5.000% in the seven readings from September 19 to October 5, 2026, that rate is 0.045 times utilization and only measures utilization. The 10.6 basis points the depositor gained between October 3 and 5, 2026, are the small-scale proof: utilization rose from 74.750% to 77.1038% and the Fed did not meet.
• That Ondo has done what it already did. The October 1, 2026 reset raised the USDY APY by 15 basis points, below the 20 that would have moved USDY out of the slow family, but the discretionary channel is proven, and the margin retained by the issuer widened to 26 basis points in the October 2 dashboard. If the January 1, 2027 reset does not reflect a December hike, then the issuer's decision overrides the portfolio's maturity, and the row must be rewritten.
• That concentration still matters. The new category rests on a single observation, that of September 30, 2026, and five days later it lacks a second: on October 5, 2026, the Arc rate is 0.8 basis points above the October 3 rate and the deposited balance has risen by 18,268 USDC. It is sought via a sweep of hub events, because an episode that reverts in 2 hours and 24 minutes leaves no trace in two readings taken hours apart. If in the two months following October 5, 2026, that sweep finds no account that has moved a market rate on the map by more than 100 basis points in less than six hours, the Arc episode remains an anecdote and the category is removed.
• That there is a useful meeting. If there is no hike on December 9, 2026, the map remains with a single observation, that of September 16, 2026: the classification by mechanism remains legible in the contract, but the measured pass-through is anecdotal.

Until that date arrives, what is verifiable is already in the contract. Before comparing an on-chain dollar yield with a Treasury rate, it is worth knowing if the number being looked at is a portfolio yield, a contract constant, or an issuer's monthly decision, because all three are called "APY" on the same screen and only the first responded to the 25 basis points of September 16, 2026. And it is worth looking more than once and with hours of separation, because the fourth possibility is that what you are looking at is the schedule of a single depositor.

Sources and links: FOMC Statement Sep-16-2026 · SOFR (New York Fed) · EFFR (New York Fed) · 4-week Bill, DTB4WK series (FRED) · 3-month Bill, DTB3 series (FRED) · Hyperliquid Portfolio Margin Documentation · USDY Factsheet and Portfolio (Ondo) · Securitize, BUIDL issuer · Sky Executive Votes · LlamaRisk, Aave Parameter Rounds · BUIDL-I Pool Series on DefiLlama · HyperLend Markets API · Maple Finance · Digital Today on Coinbase USDC Reward (Sep-17-2026)