Welcome to Shaping Tomorrow

The Summer Money Left the Bank

Inaugural edition of the Financial Services & Future of Money series. Subsequent cycles will track what materially changes against this baseline. Cycle: 23 July 2026 · Audience: Senior strategy, foresight, banking, payments, risk and policy professionals · Horizon: 6 to 18 months primary, 2 to 5 years secondary
Read estimate: 3 min Executive Synthesis · 22 min full read
⤢ Expand

What you'll learn

  1. Why stablecoins crossed inside the regulated perimeter in 2026, and why the still-unwritten yield rule now decides whether the deposit base moves in an orderly way or a disorderly one.
  2. Why the card networks conceded rail-neutrality and re-underwrote themselves as the identity and mandate-verification layer for AI-agent payments.
  3. Why tokenization reached production infrastructure at DTCC, and why the systemic risk shifted from whether it scales to whether it fails together, faster.
  4. Why central-bank digital money divided over a single question, whether it may pay interest, and why the dollar is ceding store-of-value share to gold rather than to rivals.

Key takeaways

  1. GENIUS moved from statute to a live rulebook (OCC 25 Feb, Treasury/FinCEN 8 Apr 2026), USDC overtook USDT on volume, and JPMorgan, Bank of America and Citi announced a shared tokenized-deposit network, all while the CLARITY market-structure bill, and its yield question, stayed stuck.
  2. Card networks went rail-neutral: Mastercard launched Agent Pay for Machines and agreed to buy stablecoin platform BVNK for up to $1.8bn; Visa's stablecoin settlement hit a $7bn run-rate; the x402 protocol processed 169m payments in year one.
  3. DTCC, custodian of $114tn, moved its tokenization service toward limited production in July and launch in October 2026, as the IMF, BIS and FSB warned that atomic settlement, 24/7 markets and AI model monoculture compress reaction time.
  4. The US legislated a four-year retail-CBDC ban with a wholesale carve-out, the digital euro went to trilogues, China's e-CNY went interest-bearing, and the dollar's reserve share fell below 57%, migrating into gold, which now rivals Treasuries in official reserves.

Executive Synthesis

Why this matters now

This is the inaugural Financial Services & Future of Money cycle, so it sets a baseline rather than grading a prior scan. The organising observation: in 2026 the digital-money question stopped being whether it arrives and became who issues it, who clears it, and who absorbs the reflexivity. Money is being unbundled from the bank deposit onto programmable rails, faster than the rulebook or the banking perimeter can adapt.

Four shifts this cycle establishes as the baseline, and the pattern that names them:

  • Private money crossed inside the regulated line. GENIUS went from statute to a live rulebook, USDC overtook USDT on volume, and the banks answered with a tokenized-deposit network. The unwritten CLARITY yield rule is the switch between an orderly migration and a deposit-flight shock.
  • The networks conceded the rail and claimed the agent. Visa and Mastercard went rail-neutral across cards, accounts and stablecoins, and re-underwrote themselves as the identity and mandate layer for AI-agent payments.
  • The plumbing reached production and the risk turned reflexive. DTCC moved tokenization toward an October launch; the systemic question shifted from whether tokenization scales to whether atomic, always-on, model-correlated markets fail together and faster.
  • Central-bank money split over interest. The West barred retail CBDCs while China made the e-CNY interest-bearing; the dollar kept trade dominance but ceded store-of-value share to gold, not to rivals.
  • Name the pattern: money left the bank. Deposits, settlement and monetary transmission are migrating onto programmable rails, and the perimeter that used to contain them is being redrawn from both sides at once.

This cycle names the variable that has become the defining question in financial-services strategy: not which technology wins, but who is allowed to issue money, who clears it, and who bears the reflexivity when it moves at machine speed. Between January and July 2026 that shifted from projection to operating fact. Four developments carry the pattern.

First, private money crossed inside the regulated perimeter. The GENIUS Act moved from statute to a working rulebook: the OCC issued implementing rules (25 February 2026) and Treasury, FinCEN and OFAC a joint AML rule (8 April 2026) treating stablecoin issuers as regulated financial institutions. USDC overtook USDT on adjusted volume, and JPMorgan, Bank of America and Citi answered with a shared tokenized-deposit network. The decisive nuance: the market-structure bill that would let stablecoins pay yield is still stuck in the Senate, and that unwritten yield rule is the switch between a slow deposit migration and a fast one.

Second, the payment networks conceded the settlement rail and moved up a layer. Mastercard launched Agent Pay for Machines, settling agent transactions across cards, accounts and stablecoins, and agreed to buy the stablecoin platform BVNK for up to $1.8bn while Visa's stablecoin settlement reached a $7bn annualised run-rate. The contest has moved beyond which rail settles the payment, to who owns the agent's identity and verifies its mandate.

Third, tokenization crossed from pilot to production infrastructure. DTCC, whose depository custodies $114tn, moved its tokenization service toward limited production in July and launch in October 2026. As it did, the IMF warned that automated margin calls in tokenized markets "can amplify volatility, rather than dampen it," and the BIS flagged that shared AI models and concentrated providers raise correlated behaviour. The risk migrated from whether tokenization scales to whether it fails together.

Fourth, central-bank money divided over a single design choice. The US legislated a four-year retail-CBDC ban with a wholesale carve-out, the digital euro went to trilogues, and China's e-CNY went interest-bearing, crossing the deposit-protecting line the West holds sacred. Beneath the "retail CBDC is dead in the West" consensus, the real divergence is whether central-bank digital money may pay interest and disintermediate banks.

The system's first response was not to pick a winning technology but to redraw the line around what counts as money, and who may issue it.

The historical rhyme worth holding is the money-market-fund era of the 1970s and 1980s: an out-of-perimeter instrument that paid a better rate, pulled deposits out of banks, forced a regulatory response, and permanently changed the shape of the deposit base without destroying it. The Bank Policy Institute models the analogous risk today: a $4tn yield-bearing stablecoin market by 2030 could drain $3.7tn of deposits and cut lending 19%. The Federal Reserve reads the same history as compositional, not catastrophic. The question this cycle is which precedent governs.

Where this analysis could be wrong

The most consequential assumption is that the deposit unbundling is structural rather than cyclical, that stablecoins and tokenized money-market funds represent a durable migration rather than a yield-chasing episode that reverses when rate differentials compress. If the CLARITY yield ban holds, stablecoin growth stalls near the current ~$300-400bn, and no large bank reports sustained deposit outflow attributable to stablecoins, then the "money left the bank" framing over-weights what is really a compositional shift the banking system absorbs, as the Fed argues it will. The falsifiable tests are the CLARITY yield provision as enacted, whether any large deposit franchise reports stablecoin-attributable outflow through 2026-27, and whether stablecoin market cap breaks clearly above $500bn or stalls. Signals to watch are the Senate market-structure text, the bank deposit-beta disclosures at the Q3/Q4 prints, and whether tokenized-deposit networks launch on schedule in H1 2027.

What this cycle establishes as the baseline

  1. Stablecoin perimeter, baseline set. Private money is now inside the regulated line via GENIUS; the open variable that decides the pace of deposit migration is the CLARITY yield rule.
  2. Identity layer, baseline set. The settlement-rail contest is effectively over (the networks went rail-neutral); the live contest is over agent identity and mandate verification.
  3. Tokenized markets, baseline set. Production infrastructure arrives in 2026 (DTCC launch October); reflexivity, correlated and faster failure, is the emerging systemic risk regulators are pricing first.
  4. Sovereign split, baseline set. The CBDC divergence is about remuneration, not existence; and the dollar's store-of-value erosion runs to gold while its trade role endures.
  5. Inaugural scan delta. The scan ran 38 sources across four themes at roughly 74% Tier 1-2, drawing on central banks, the BIS, IMF, FSB and primary filings. Subsequent cycles will grade what materially changes against this baseline rather than restating it.

Each is developed below, with a decision posture, in the four Strategic Implications.

Baseline ledger: the dated commitments this cycle sets for the next

As the inaugural edition, this cycle has no prior obligations to resolve. Instead it fixes the dated watch-items the next cycle will grade, each with a source and a date by which it should be checked.

Watch-item Type What to check, and by when
CLARITY market-structure bill DECISIVE Whether the enacted text permits stablecoin yield. A merged Senate draft was expected on the floor from late July 2026; recess and midterms threaten passage. Check by Q4 2026.
DTCC tokenization launch MILESTONE DTCC targeted limited production in July and full launch in October 2026. Check the go-live and initial asset scope by November 2026.
FSB responsible-AI standard MILESTONE The FSB consultation (comments closed 22 July 2026) is due to finalise in October 2026; watch whether it names agentic AI, which US model-risk guidance left out of scope.
Digital euro legislation DECISIVE Trilogues opened June 2026 on the regulation and the ~EUR3,000 holding cap. The negotiated cap and any remuneration stance are the variables to watch; check by mid-2027 against a potential 2029 issuance.
US bank-capital re-proposal IN TRAIN The Basel III re-proposal that cuts CET1 (comments closed 18 June 2026) is a capital reduction justified as re-internalising shadow credit. Check the final rule and any divergence from the international standard by early 2027.

Audience Snapshots

Four functional lenses on the same intelligence base. Each card names the shift and the single question this cycle puts to that function's leadership.

Banking & treasury strategy

Is your deposit base modelled for a stablecoin yield rule you do not yet control?

The shift: GENIUS put stablecoins inside the regulated line; the Bank Policy Institute models up to $3.7tn of deposit loss if yield is permitted; money-market funds already hit a record $8.3tn; and the largest banks answered with a shared tokenized-deposit network targeted for H1 2027.

The question to brief: should we run deposit-flight scenarios keyed to the CLARITY yield outcome, and set our tokenized-deposit posture now rather than after the rule is written?

Payments & fintech

Are you positioned to own agent identity, or only to settle the payment?

The shift: the networks went rail-neutral, with Mastercard's Agent Pay for Machines, its up-to-$1.8bn BVNK acquisition and Visa's $7bn stablecoin run-rate; standards such as AP2 and x402 are consolidating around who verifies an agent's mandate; and settlement economics commoditise as the rail stops being the moat.

The question to brief: should we stake a position in the identity and mandate-verification layer now, before the agentic-payment standards harden around a few incumbents?

Asset management & market infrastructure

Is your operating model ready for tokenized collateral that settles atomically, around the clock?

The shift: DTCC, custodian of $114tn, moves tokenization to production this October; tokenized Treasuries have passed ~$7.5bn and private credit is the breakout use case; yet the IMF warns automated margin calls can accelerate selloffs and atomic, 24/7 settlement compresses reaction time.

The question to brief: should we build for atomic, always-on settlement and tokenized-collateral mobility now, and stress-test our funds for machine-speed, correlated liquidity events?

Central banks, regulators & policy

Which precedent set this cycle will bind your mandate first?

The shift: the CLARITY yield rule will decide deposit-system stability; the US left agentic AI out of model-risk scope while the FSB moves to fill the gap; private-credit interconnection is now an official warning; and the CBDC divergence is really about whether central-bank money may pay interest.

The question to brief: should we act now on the stablecoin yield question, the agentic-AI supervisory gap and private-credit interconnection, rather than react once each one binds?

Themes

The cycle's signals are organised into four themes, ranked by impact on near-term decisions. Immediate: changes H2 2026 / H1 2027 capital, siting or offtake decisions. Near-Term: changes competitive position over the next twelve months. Longer-Range: a multi-year shift to track each cycle.

1. The Stablecoin Perimeter: private money crossed inside the regulated line, and the banks moved to meet it

Immediate

The GENIUS Act crossed private money inside the regulated line and, in doing so, turned a payments story into a monetary one. The OCC (25 February 2026) and Treasury, FinCEN and OFAC (8 April 2026) built a working rulebook for issuers while the CLARITY market-structure bill, and its yield question, stayed stuck in the Senate. USDC overtook USDT on adjusted volume; the largest banks answered with a shared tokenized-deposit network. Beneath the payments framing, stablecoins became a monetary-transmission variable: the BIS finds inflows compress Treasury-bill yields and the IMF finds they depreciate emerging-market currencies. The regulated perimeter is now contested from both sides at once, as private rails pull money out while a US Basel reversal pulls credit back in.

The stablecoin perimeter in four numbers

The stablecoin perimeter in four numbers STABLECOIN MARKET ~$300bn USDC ~70% of volume, June record $1.79T CoinDesk, Jul 2026 GENIUS RULEBOOK live OCC + Treasury issuer rules, Feb-Apr 2026 OCC / Treasury, 2026 DEPOSITS AT RISK $3.7tn if yield is permitted by 2030 (BPI model) BPI, May 2026 T-BILL TRANSMISSION ~5bp per $3.5bn inflow (8-10bp under stress) BIS, Jun 2026 Private money crossed inside the line – and the pipe now carries pressure both ways.

The July 2026 stablecoin perimeter in four numbers. Source: OCC, U.S. Treasury, BPI, BIS, CoinDesk.

  • The OCC proposed GENIUS Act implementing rules setting reserve, redemption, custody and capital standards for federal bank and non-bank stablecoin issuers. OCC (25 February 2026).
  • Treasury, FinCEN and OFAC proposed a joint rule making payment stablecoin issuers BSA financial institutions with mandatory AML and sanctions programmes. U.S. Treasury (8 April 2026).
  • USDC overtook USDT on adjusted volume (~70% versus ~25%) as June set a record $1.79T monthly total, with stablecoin volumes now exceeding Bitcoin. CoinDesk (6 July 2026).
  • JPMorgan, Bank of America and Citi, via The Clearing House, are building a shared tokenized-deposit network for H1 2027, explicitly to blunt stablecoin deposit flight. CoinDesk (5 June 2026).
  • A $3.5bn stablecoin inflow compresses 3-month Treasury-bill yields ~5bp (8-10bp during Treasury stress), a channel strengthened post-GENIUS. Bank for International Settlements (revised June 2026).
  • A 1% net stablecoin inflow depreciates the local currency ~5-10bp and widens the dollar premium, with spillovers most pronounced in emerging markets, a monetary-autonomy risk. International Monetary Fund (March 2026).
  • Standard Chartered projects stablecoins reach $2T by 2028, generating $0.8-1.0T of fresh Treasury-bill demand and potentially letting Treasury shift issuance toward bills. Standard Chartered via CoinDesk (23 February 2026).
  • Money-market fund assets hit a record ~$8.3tn, evidence the deposit base is already being contested by higher-yielding cash alternatives. Bloomberg (29 May 2026).
  • The US Basel III re-proposal cuts CET1 requirements (GSIB ~4.8%, regional ~5.2%), explicitly to stop lending migrating outside the regulated perimeter, a deregulatory turn away from the international standard. Holland & Knight (16 June 2026).

Counter-argument

The Federal Reserve reads the deposit effect as compositional, not catastrophic: banks adapted to money funds and sweep accounts before, and about half of large banks already prioritise tokenized deposits (Fed FEDS Note, 1 May 2026). The New York Fed's narrow-banking model (Staff Report 1179, February 2026) shows stablecoin-versus-tokenized-deposit competition can be welfare-optimal rather than destabilising. On this reading the "money left the bank" thesis over-weights a reallocation the perimeter absorbs. The named actors to act on are the OCC and Treasury as rule-writers, Circle and Tether as issuers, and the bank consortium building tokenized deposits through The Clearing House.

Weak signals to watch

  • Weak signal The Federal Reserve now lists stablecoins alongside Treasuries, FX and equities as a shock-transmission node. Would gain weight if a Fed financial-stability report formally models stablecoin runs as a transmission channel.
  • Weak signal The FSB and ECB warn that bank-nonbank interconnection is the contagion channel a capital cut could amplify. Would gain weight if a private-credit fund's stress hits a bank credit line during a deposit-migration episode.
Decision link: Strategic Implications 1.

2. The Identity Layer: the networks conceded the rail and re-underwrote the agent

Near-Term

The payment networks made a strategic concession this cycle: they stopped defending the card rail and re-underwrote themselves one layer up. Mastercard's Agent Pay for Machines settles across cards, accounts and stablecoins; its up-to-$1.8bn acquisition of the stablecoin platform BVNK and Visa's alignment with the x402 protocol mean the incumbents now settle on whatever rail is cheapest and compete instead on agent identity and mandate verification. Standards are consolidating around a small stack, Google's AP2, OpenAI and Stripe's ACP, and the x402 on-chain extension. The systemic subtext is that the same concentration logic is being rebuilt into the agentic-payments stack before any supervisor covers agentic AI, which US model-risk guidance explicitly left out of scope.

The identity layer in four numbers

The identity layer in four numbers VISA STABLECOIN $7bn annualised run-rate up ~50% QoQ Forbes, Jun 2026 x402 IN YEAR ONE 169m payments, 590k buyers, 100k sellers Forbes, Jun 2026 MASTERCARD / BVNK $1.8bn acquisition of a stablecoin platform Forbes, Jun 2026 AGENTIC AI RULES out of model-risk scope (US guidance) OCC, Apr 2026 The networks conceded the rail – and moved to own the agent.

The July 2026 agentic-payments contest in four numbers. Source: Forbes, Mastercard, OCC.

  • Mastercard launched Agent Pay for Machines, settling agent transactions across cards, accounts and stablecoins with 30+ partners including Coinbase, Stripe and Ant. Mastercard (10 June 2026).
  • Visa's stablecoin settlement reached a $7bn annualised run-rate (up ~50% QoQ) and aligned with Coinbase's x402; the x402 protocol processed 169m payments across 590k buyers and 100k sellers in year one; Mastercard agreed to buy BVNK for up to $1.8bn. Forbes (7 June 2026).
  • Google's Agent Payments Protocol (AP2), with 60+ organisations, uses cryptographically signed Intent and Cart mandates, with an A2A-x402 extension for stablecoin agent payments. Google Cloud (16 September 2025).
  • Stripe and OpenAI's Agentic Commerce Protocol powers ChatGPT Instant Checkout, a competing agentic-payments standard consolidating the stack around a few originators. Stripe / OpenAI (29 September 2025).
  • US agencies' revised interagency model-risk-management guidance is non-enforceable and states that generative and agentic AI are not within its scope, promising only a future request for information. OCC (17 April 2026).
  • The FSB opened its first consultation on 12 sound practices for responsible AI adoption across 24 jurisdictions, with comments closing 22 July and a final report due October 2026. Financial Stability Board (10 June 2026).
  • The BIS warned that widespread use of similar AI models, data or decision rules increases correlated behaviour, with hardware, cloud and model provision concentrated among a few providers. Bank for International Settlements (26 January 2026).

Counter-argument

The rail-neutral framing may overstate the incumbents' security. If agent identity standardises on an open protocol such as AP2 or x402 that any wallet or bank can implement, the networks' "identity layer" is a commodity, not a moat, and the value accrues to whoever owns the consumer relationship, the AI assistant, rather than the card network. The named actors are Visa and Mastercard as identity-layer claimants, the AI platforms (OpenAI, Google) as mandate originators, and the stablecoin settlement providers (Coinbase, BVNK, Stripe).

Weak signals to watch

  • Weak signal The x402 protocol settles in ~200ms for a fraction of a cent on Base, economics that could make card interchange uncompetitive for agent-to-agent commerce. Would gain weight if a major merchant routes agent payments off-card by default.
  • Weak signal The US carve-out of agentic AI from model-risk rules, with only a future request for information promised, leaves the fastest-moving payments layer unsupervised. Would gain weight if an agentic-payment fraud or failure event forces an emergency supervisory response.
Decision link: Strategic Implications 2.

3. Tokenized Markets: the plumbing reached production and the risk turned reflexive

Near-Term

Tokenization crossed from pilot to production infrastructure this cycle. DTCC, whose depository custodies $114tn, moved its tokenization service toward limited production in July and launch in October 2026; tokenized Treasuries passed ~$7.5bn and private credit emerged as the breakout use case. The consequential shift is not the asset count but the market microstructure it enables: atomic settlement, 24/7 trading and tokenized-collateral mobility. The IMF, BIS and FSB converge on one warning: the very features that make tokenized markets efficient, speed, determinism and automation, compress the time available for discretionary intervention, so stress is likely to unfold faster and more correlated.

Tokenized markets in four numbers

Tokenized markets in four numbers DTCC TOKENIZATION $114tn custody; launch October 2026 DTCC, May 2026 RWAs ON-CHAIN >$20bn tokenized Treasuries ~$7.5bn (+600%/18mo) Yellow, May 2026 PRIVATE CREDIT ~$1.7tn AUM, the breakout tokenization use case Apollo, Nov 2025 IMF WARNING faster auto margin calls amplify volatility IMF, Apr 2026 The plumbing reached production – and the risk turned reflexive.

The 2026 tokenized-markets picture in four numbers. Source: DTCC, Yellow Research, Apollo, IMF.

  • DTCC advanced a tokenization service with 50+ firms toward limited production in July and launch in October 2026; its depository custodies $114tn and it will tokenize Russell 1000 names, ETFs and Treasuries. DTCC (4 May 2026).
  • The IMF's "Tokenized Finance" note counts $23.2bn of tokenized real-world assets and warns automated margin calls "can amplify volatility, rather than dampen it," with smart-contract features that may accelerate selloffs. International Monetary Fund (April 2026).
  • Tokenized real-world assets passed $20bn, tokenized Treasuries reached ~$7.5bn (up ~600% in 18 months), and BCG and ADDX project $16tn by 2030. Yellow Research (8 May 2026).
  • Private credit, ~$1.7tn in AUM, is emerging as the breakout tokenization use case, extending wealth access to an illiquid asset class. Apollo (November 2025).
  • The BIS warned that AI model monoculture and herding raise correlated behaviour, and that compute, cloud and model provision is concentrated, a concern that compounds with atomic settlement. Bank for International Settlements (26 January 2026).
  • The IMF further cautioned that tokenization could import crypto risks into mainstream markets, with automated, faster markets amplifying volatility. IMF via CoinDesk (6 April 2026).

Counter-argument

The reflexivity risk may be over-weighted at current scale. At ~$20-23bn, tokenized real-world assets are a rounding error against the $114tn DTCC custodies or the multi-trillion Treasury market, and most tokenized Treasuries are held by crypto-native treasuries as collateral, not traded at machine speed by leveraged funds. On this reading the systemic-stress case is a 2028-plus concern and the near-term story is operational modernisation. The named actors are DTCC and the traditional custodians as infrastructure owners, BlackRock, Franklin and Ondo as tokenized-fund issuers, and Apollo and Hamilton Lane in tokenized private credit.

Weak signals to watch

  • Weak signal If tokenized collateral becomes mobile across venues 24/7, a margin call in one market can trigger automated sales in another with no human in the loop. Would gain weight if a tokenized-collateral liquidation cascades across two venues in a stress event.
  • Weak signal Tokenization is concentrating on a few hubs (DTCC, Canton), rebuilding single-point-of-failure risk into the new plumbing. Would gain weight if a major tokenization-platform outage halts settlement.
Decision link: Strategic Implications 3.

4. The Sovereign Split: central-bank money divided over the right to pay interest

Longer-Range

Central-bank money divided this cycle over a single design choice: whether it may pay interest. The US legislated a four-year retail-CBDC ban with a wholesale carve-out; the digital euro went to trilogues, framed geopolitically as breaking "the US payment grip" and designed non-remunerated with a ~EUR3,000 cap precisely to protect bank deposits; and China's e-CNY quietly went interest-bearing and deposit-like from 1 January 2026, crossing the line the West holds sacred. Beneath the "retail CBDC is dead in the West" consensus, the real divergence is whether central-bank digital money may disintermediate banks. Meanwhile the dollar's reserve share fell below 57%, migrating into gold rather than into rivals, and wholesale rails bifurcated into non-interoperable blocs.

The sovereign split in four numbers

The sovereign split in four numbers US RETAIL CBDC banned four years, Senate 85-5; wholesale carve-out CoinDesk, Jun 2026 DIGITAL EURO ~2029 ~EUR3,000 cap, non-remunerated ECB, Oct 2025 e-CNY ~$2.3tn cumulative; now interest-bearing Atlantic Council, Jan 2026 DOLLAR RESERVES 56.9% lowest since 1995; gold now rivals UST IMF COFER, 2026 The split is not CBDC-versus-none – it is whether the money may pay interest.

The 2026 sovereign-money divergence in four numbers. Source: ECB, Atlantic Council, IMF COFER, CoinDesk.

  • The US Senate passed a housing bill 85-5 carrying a four-year (through 2030) ban on a Fed retail CBDC; the ban targets retail, not wholesale or tokenized-reserve work, which the Fed continues inside BIS Project Agora. CoinDesk (22 June 2026).
  • The European Parliament committee approved the digital-euro regulation and mandated trilogues, framed to "break the US payment grip," targeting a potential 2029 issuance. CoinDesk (23 June 2026).
  • The ECB closed the digital-euro preparation phase, testing a ~EUR3,000 non-remunerated holding cap as financial-stability-safe, with ~EUR1.3bn build cost and potential first issuance in 2029. European Central Bank (30 October 2025).
  • China's e-CNY reached 16.7tn RMB (~$2.3tn) cumulatively by November 2025, and a new regime effective 1 January 2026 adds interest-bearing, deposit-like features to out-compete stablecoins and build dollar-alternative settlement rails. Atlantic Council (15 January 2026).
  • The digital pound remains in its design phase with no decision made and a go/no-go due later in 2026, leaving the UK a deliberate laggard. Bank of England (4 March 2026).
  • Wholesale rails bifurcated: BIS and seven central banks delivered the Agora unified-ledger prototype for atomic multi-currency settlement of tokenized deposits and reserves, while mBridge continues without the BIS as a rival bloc, with zero shared members. PostTrade360 (28 May 2026); Forbes (12 May 2026).
  • The dollar's reserve share fell to 56.9% (COFER Q3 2025), the lowest since 1995, as central banks bought ~863t of gold in 2025 and gold rose toward rivalling Treasuries in official reserves. informedclearly (IMF COFER / WGC) (2 May 2026).
  • The G20 cross-border payments roadmap is likely to miss its 2027 targets, with retail payments settled within one day at 67.3% and average remittance cost 6.5%, as ISO 20022 migration continues. Financial Stability Board (9 October 2025).
  • The IMF's 2026 Article IV consultation flagged elevated US debt and current-account deficits with growth of 2.4%, the fiscal backdrop against which the dollar's reserve role is being tested. International Monetary Fund (2 April 2026).

Counter-argument

The "sovereign split" may overstate fragmentation. The dollar retains overwhelming trade-invoicing and funding dominance, and stablecoins, almost entirely dollar-denominated, arguably extend the dollar's reach onto new rails rather than eroding it. Gold buying reflects reserve diversification, not the rise of a rival reserve currency, and no credible alternative to the dollar exists at scale. The named actors are the Fed and Treasury as dollar-system stewards, the ECB and PBOC as CBDC issuers, and the BIS as wholesale-rail convenor.

Weak signals to watch

  • Weak signal If the e-CNY's interest-bearing design shows it can disintermediate banks without collapsing them, it becomes a template that pressures the West's non-remunerated CBDC orthodoxy. Would gain weight if another major economy adopts a remunerated retail CBDC.
  • Weak signal Central-bank gold now exceeds foreign-held US Treasuries on some measures; a continued shift would slowly erode the Treasury market's captive-buyer base. Would gain weight if a major reserve manager publicly rebalances further from Treasuries into gold.
Decision link: Strategic Implications 4.

Strategic Implications

Four decisions the cycle brings forward, each with an owner, a dated action and a decision posture (Decide, Prepare, Monitor). One opportunity line names who profits from each shift.

SI 1: Set the deposit-defence posture now: stress the base for the CLARITY yield outcome and decide the tokenized-deposit play

GENIUS put stablecoins inside the regulated line, and the Bank Policy Institute models up to $3.7tn of deposit loss if yield is permitted, while the largest banks answer with a tokenized-deposit network for H1 2027. Decide the posture before the rule is written: run deposit-flight scenarios keyed to the yield outcome and choose whether to issue, join a consortium, or partner with an issuer. If the Fed’s compositional reading holds and yield is banned, the move over-provisions, so size it to the yield-rule probability and stage it.

Action: by Q4 2026, table a board deposit-defence paper with deposit-beta scenarios keyed to the CLARITY yield outcome and a tokenized-deposit decision (issue, consortium or partner).

Who profits: tokenized-deposit consortia and the banks that move first on compliant on-chain settlement, and the issuers (Circle) that win bank distribution.

Decide Draws on Theme 1. Owner: CFO / Group Treasurer.

SI 2: Stake a position in the agent-identity and mandate layer before the standards harden

The networks went rail-neutral (Mastercard’s Agent Pay, the BVNK buy, Visa’s x402 alignment) and the value moved to identity and mandate verification, with standards (AP2, ACP, x402) consolidating fast. Prepare a position now, as an identity or credentialing provider, a mandate-verification integrator, or an early adopter, before a few incumbents lock the stack. If identity standardises on an open protocol anyone can implement, the layer commoditises, so hedge by owning the consumer or merchant relationship rather than betting on a single protocol.

Action: by 31 December 2026, choose and pilot an agentic-payment stack (AP2, ACP or x402) and define the firm’s identity and mandate role ahead of standard consolidation.

Who profits: identity and credentialing providers, mandate-verification integrators, and the stablecoin settlement rails (Coinbase, Stripe, BVNK).

Prepare Draws on Theme 2. Owner: Chief Payments / Product Officer.

SI 3: Build for atomic, around-the-clock settlement and stress-test funds for machine-speed liquidity events

DTCC brings tokenization to production this October and tokenized Treasuries and private credit are scaling; the readiness gap is atomic settlement, 24/7 markets and tokenized-collateral mobility. Prepare the operating model now and add the IMF’s automated-margin-call channel as a machine-speed, correlated-liquidity stress scenario in fund risk models. If reflexivity is really a 2028-plus concern, early build over-invests, so prioritise interoperability and reversibility (governance override) over speed at any cost.

Action: by the Q1 2027 planning round, define a tokenized-settlement operating-model roadmap and add a machine-speed liquidity stress scenario to fund and treasury risk frameworks.

Who profits: tokenization infrastructure (DTCC, Canton), tokenized-fund issuers (BlackRock, Franklin, Ondo), and collateral-mobility and risk-analytics vendors.

Prepare Draws on Theme 3. Owner: COO / Head of Markets.

SI 4: Monitor the sovereign split and the dollar-to-gold shift, and position reserves and rails accordingly

The CBDC divergence is about remuneration (China’s interest-bearing e-CNY versus the West’s non-remunerated design), the dollar is ceding store-of-value share to gold while keeping trade dominance, and wholesale rails are bifurcating (Agora versus mBridge). Monitor with named checkpoints, the digital-euro trilogue outcome, the e-CNY’s cross-border use, and reserve-manager gold rebalancing, and position payment rails and reserve exposure for a multi-bloc settlement world. If the dollar-extension reading is right, that stablecoins carry the dollar onto new rails, the fragmentation thesis over-weights, so this is Monitor, not Decide.

Action: by the next cycle, stand up a watch on the digital-euro cap and remuneration stance, e-CNY cross-border pilots and official gold flows; flag if a second major economy adopts a remunerated retail CBDC.

Who profits: wholesale-rail and correspondent-alternative providers, gold and reserve-diversification managers, and the dollar-stablecoin issuers extending the dollar’s reach.

Monitor Draws on Theme 4. Owner: Group Treasurer / Head of Policy.

Scenario Matrix

Four 18-month operating environments generated by crossing two axes the evidence base does not yet decide: whether the deposit base migrates in an orderly way or a disorderly one (driven mainly by the CLARITY yield rule), and whether the tokenized and agentic system stays contained or turns reflexive (driven by concentration, atomic settlement and the agentic-AI supervisory gap). The matrix is a planning tool, not a forecast; the value sits in the indicators that would tip a reader from one cell to another.

Deposit migration
System resilience
Orderly
Disorderly
Contained
Reflexive

When the Money Ran

The worst pairing: CLARITY permits stablecoin yield and deposits flee fast, while the tokenized and agentic stack is concentrated and unsupervised. A stablecoin de-peg or a tokenized-collateral margin cascade propagates across atomic, 24/7 venues faster than any supervisor can convene, and the BPI’s $3.7tn deposit-drain scenario collides with a machine-speed liquidity event.

Indicators: enacted CLARITY yield permission plus a large bank reporting stablecoin-attributable outflow; a tokenized-collateral liquidation cascading across venues; an agentic-payment fraud event with no supervisory response.

Fragile Calm

Deposits migrate slowly because yield stays contained, so the surface looks calm, but the tokenized and agentic plumbing concentrates on a few hubs and goes unsupervised. Resilience is untested until a shock, when correlated, atomic markets amplify it. The danger is complacency: the deposit base holds while the failure mode moves into market microstructure.

Indicators: stablecoin growth stalls but DTCC and Canton tokenization concentration rises; the FSB standard omits agentic AI; a single tokenization-platform outage briefly halts settlement.

The Run, Absorbed

Yield is permitted and deposits move fast, but resilient plumbing and supervision absorb it. Banks lose deposit share to stablecoins and tokenized money funds, adapt through tokenized-deposit networks, and lending reprices rather than collapses, the 1980s money-fund rhyme. A painful reallocation in a contained system.

Indicators: sustained stablecoin and money-fund outflow but stable bank funding via tokenized deposits; the Fed’s compositional reading validated in a financial-stability report; no cross-venue reflexive cascade.

Managed Transition

The best pairing: CLARITY contains yield, deposits migrate slowly, banks adapt with tokenized deposits, tokenization modernises operations without machine-speed fragility, and supervision (FSB, ECB) catches up. Money moves onto programmable rails as an orderly upgrade rather than a run.

Indicators: enacted yield limits; tokenized-deposit networks launch on schedule in H1 2027; the FSB names agentic AI in its final standard; no stablecoin-attributable bank outflow.

What We Are Not Planning For

Four scenarios held out of the plan because the evidence base does not yet justify resourcing against them. Each carries the reinstatement trigger that would change that judgement.

Exclusion 1: A near-term collapse of the dollar’s reserve or trade role

The dollar’s erosion this cycle is gradual and runs to gold, not to a rival currency, and no credible alternative exists at scale. We are not planning for a sudden dollar displacement within the 6-18-month horizon; the near-term story is diversification at the margin, not regime change.

Reinstatement trigger: a major trade bloc settling a material share of trade off-dollar, or the dollar’s reserve share falling below ~50% on the COFER measure.

Exclusion 2: A Western remunerated or retail CBDC within the horizon

The US legislated a four-year retail-CBDC ban and the digital euro is designed non-remunerated with issuance no earlier than ~2029. We are not planning for a Western remunerated retail CBDC to reshape the 6-18-month picture; the live Western work is wholesale and tokenized-reserve.

Reinstatement trigger: repeal of the US ban, or the digital-euro trilogue adopting remuneration or a materially higher holding cap.

Exclusion 3: A systemic tokenized-market crisis at current scale

At ~$20-23bn, tokenized real-world assets are too small to be systemic in 2026, and most tokenized Treasuries are held as collateral rather than traded at machine speed by leveraged funds. We treat the reflexivity risk as a 2028-plus concern, not a 6-18-month planning input, while tracking it as a weak signal.

Reinstatement trigger: tokenized real-world assets crossing a materiality threshold (for example above $500bn) with leveraged, cross-venue trading, or a cross-venue tokenized-collateral cascade.

Exclusion 4: A reversal or rollback of the stablecoin regime

GENIUS is law and being implemented through OCC and Treasury rules, and the political direction is accommodative. We are not planning for a statutory reversal of the stablecoin regime within the horizon; the open question is the yield rule, not the regime’s existence.

Reinstatement trigger: a statutory rollback of GENIUS, or an enforcement action freezing or unwinding a top-two issuer.

Discussion Points

  1. If GENIUS has put stablecoins inside the regulated line and the yield rule is still unwritten, do we set our deposit-defence and tokenized-deposit posture now, or wait for CLARITY and risk moving late?
  2. For payments leaders: if the networks conceded the rail and the value moved to agent identity, do we stake a position in the identity and mandate layer now, and which standard (AP2, ACP or x402) do we build against?
  3. If DTCC brings tokenization to production this October and the IMF warns automated margin calls accelerate selloffs, are our funds and treasury operations ready for atomic, 24/7, machine-speed liquidity events?
  4. If the US left agentic AI out of model-risk scope while the FSB moves to fill it, are we running agentic-payment and AI-model risk ahead of the supervisor, or waiting to be told?
  5. If the dollar is ceding store-of-value share to gold while keeping trade dominance, and central-bank money is splitting over remuneration, how do we position reserves and payment rails for a multi-bloc settlement world?

Source Confidence Register

38 verified sources across four themes. Tier 1 (governments, central banks, regulators, multilateral bodies, primary filings): 18. Tier 2 (institutional research, think-tanks, consultancies, industry bodies): 10. Tier 3 (quality journalism and specialist trade press): 10. No Tier 4 vendor/advocacy sources are load-bearing. Non-English originals: 0. Recency window 0-6 months, most sources post-24-January 2026; 6 sources 7-10 months old are retained as structural anchors (foundational regulatory milestones, originating standards and baseline research) and used for structural context, not as fresh board-level readings. 28 of the 38 sources (74%) sit at Tier 1 or 2; the source set was frozen at 2026-07-14. Every source date was verified against genuinely fetched page text in the source-verification pass. Detect → Assess → Decide → Act.

Theme 1: The Stablecoin Perimeter

Source Tier Date Key claim
OCC T1 25 Feb 2026 GENIUS Act NPRM sets reserve, redemption, custody, capital standards for federal bank and non-bank stablecoin issuers
US Treasury (FinCEN/OFAC) T1 8 Apr 2026 Joint proposed rule makes payment stablecoin issuers BSA financial institutions with AML/sanctions programs
CoinDesk (New CLARITY Act) T3 9 Jul 2026 Merged Senate CLARITY market-structure draft may reach the floor; stalled, needs 60 votes; yield question unresolved
CoinDesk (JPMorgan, Bank of) T3 5 Jun 2026 JPMorgan, BofA, Citi via The Clearing House building shared tokenized-deposit network, live H1 2027, to blunt stablecoin deposit flight
CoinDesk (Circle's USDC overtakes) T3 6 Jul 2026 June 2026 record $1.79T stablecoin volume; USDC ~70% vs USDT ~25% of adjusted volume
Federal Reserve Bank of New York T1 Feb 2026 Narrow-banking model: stablecoins can crowd out bank credit; stablecoin-vs-tokenized-deposit competition often welfare-optimal
Standard Chartered (via CoinDesk) T2 23 Feb 2026 Stablecoins reach $2T by 2028, generating $0.8-1.0T fresh T-bill demand
Bank for International Settlements T1 Jun 2026 A $3.5B stablecoin inflow compresses 3-month T-bill yields ~5bp (8-10bp during Treasury stress)
International Monetary Fund T1 Mar 2026 1% net stablecoin inflow depreciates local currency ~5-10bp; EM monetary-autonomy risk
Federal Reserve (Fifth Conference on) T1 16 Jul 2026 Fed frames stablecoins as tokenized money transmitting shocks to Treasury yields, FX and equities
Bank Policy Institute T2 8 May 2026 If stablecoins reach $4tn by 2030, deposits fall $3.7tn and lending drops 19% (~$2.7tn)
Federal Reserve (Banks in the) T1 1 May 2026 Stablecoin cap ~$300bn; ~half of large banks prioritising tokenized deposits; deposit effect compositional, banks adapt
Bloomberg T3 29 May 2026 Money-market fund assets hit record ~$8.3tn in a dash for cash
Holland & Knight T2 16 Jun 2026 Basel III re-proposal cuts CET1 (GSIB ~4.8%, regional ~5.2%); aim to stop lending migrating outside regulated banks
European Central Bank T2 26 May 2026 Euro-area bank private-credit exposure EUR62.5bn; insurers hold ~EUR211bn; risk from spillovers and opacity
Financial Stability Board T1 6 May 2026 Private credit ~$1.5-2.0tn; ~$220bn direct bank credit-line exposure; untested in a prolonged downturn

Theme 2: The Identity Layer

Source Tier Date Key claim
Mastercard T1 10 Jun 2026 Agent Pay for Machines settles agent transactions across cards, accounts and stablecoins with 30+ partners
Forbes T3 7 Jun 2026 Visa stablecoin settlement at $7bn annualized run-rate; x402 did 169m payments in year one; Mastercard buying BVNK up to $1.8bn
Google Cloud T2 16 Sep 2025 AP2 open protocol with 60+ orgs using cryptographically-signed Intent and Cart Mandates; A2A-x402 extension for stablecoin agent payments
OCC T1 17 Apr 2026 Revised interagency Model Risk Management guidance non-enforceable; generative and agentic AI explicitly out of scope
Financial Stability Board T1 10 Jun 2026 First consultation: 12 sound practices for responsible AI adoption across 24 jurisdictions; final report Oct 2026
Bank for International Settlements T1 26 Jan 2026 Similar AI models/data/decision rules increase correlated behaviour; hardware/cloud/model providers concentrated
Stripe / OpenAI T2 29 Sep 2025 Agentic Commerce Protocol powers ChatGPT Instant Checkout, a competing agentic-payments standard

Theme 3: Tokenized Markets

Source Tier Date Key claim
DTCC T1 4 May 2026 Tokenization service with 50+ firms; limited production July 2026, launch Oct 2026; DTC custodies $114 trillion
International Monetary Fund T1 Apr 2026 Tokenized Finance note: $23.2bn RWAs tokenized; faster automated markets could amplify volatility; smart-contract margin calls may accelerate selloffs
Yellow Research T3 8 May 2026 RWAs >$20bn; tokenized Treasuries ~$7.5bn (+600% in 18 months); BCG/ADDX project $16tn by 2030
Apollo T2 Nov 2025 Private credit (~$1.7tn AUM) emerging as the breakout tokenization use case
CoinDesk T3 6 Apr 2026 IMF warns tokenization could import crypto risks; smart-contract margin calls may accelerate selloffs

Theme 4: The Sovereign Split

Source Tier Date Key claim
European Central Bank T1 30 Oct 2025 Digital-euro preparation phase closed; ~EUR3,000 holding cap; potential first issuance 2029; ~EUR1.3bn build cost
CoinDesk (European Parliament clears) T3 23 Jun 2026 European Parliament committee approved digital-euro framework and mandated trilogues; framed to break US payment grip
CoinDesk (US Senate passes) T3 22 Jun 2026 US Senate passed housing bill 85-5 with a four-year ban on a Fed retail CBDC; wholesale work continues
Forbes T3 12 May 2026 mBridge and Agora are rival bloc systems with zero shared members; multilateral interoperability dead
Bank of England T1 4 Mar 2026 Digital pound still in design phase; no decision made; go/no-go due later in 2026
Atlantic Council T2 15 Jan 2026 e-CNY 16.7tn RMB (~$2.3tn) by Nov 2025; new 1 Jan 2026 regime adds interest-bearing/deposit features to compete with stablecoins
Financial Stability Board T1 9 Oct 2025 G20 2027 cross-border targets likely missed; retail settled within 1 day 67.3%; avg remittance cost 6.5%
PostTrade360 T2 28 May 2026 BIS + 7 central banks delivered Agora unified-ledger prototype: atomic multi-currency settlement of tokenized deposits and reserves
informedclearly (IMF COFER/WGC) T2 2 May 2026 Dollar reserve share 56.9% (COFER Q3 2025), lowest since 1995; central banks bought 863t gold in 2025
International Monetary Fund T1 2 Apr 2026 US debt and current-account deficit remained elevated; growth 2.4% in 2026

Conflict notes: Three genuine source disagreements were weighted rather than resolved. (1) Deposits: the Bank Policy Institute's "$3.7tn destroyed" scenario versus the Federal Reserve's "compositional, not catastrophic" reading; the report treats the migration as real but pace-dependent and names the CLARITY yield rule as the switch. (2) Tokenization risk: the IMF and BIS reflexivity warnings versus the scale-skeptic view that ~$20bn of tokenized assets is too small to be systemic; the report frames reflexivity as a 2028-plus risk to build for now, not a 2026 crisis. (3) The dollar: the de-dollarization and gold narrative versus the dollar-extension reading that stablecoins carry the dollar onto new rails; the report frames the dollar as ceding store-of-value share while keeping trade dominance. Author-type coverage: the set is strong on central banks, regulators and policy voices and lighter on business-leader primary filings, so the payments and tokenization actors (Mastercard, Visa, DTCC) are covered via press releases and trade press and treated as directional; crypto-industry trackers (CoinDesk, Yellow Research) are treated as directional on market-size figures.

Claim-Fidelity Appendix: Analyst inferences and editorial framing

This briefing carries a set of analyst-generated interpretations that go beyond what any single source asserts. They are named here so a reader can trace the confidence line and disagree productively.

Analytical inferences carried by the body prose

  • "Money left the bank" is an analyst frame linking deposit migration, settlement disintermediation and monetary transmission into one pattern; no single source asserts the composite.
  • Treating the CLARITY yield rule as "the switch" between an orderly and a disorderly deposit migration is our interpretation: the sources establish that GENIUS is live and the yield question unresolved, but the causal weight placed on that one provision is a judgement.
  • Casting the payment networks’ moves as "conceding the rail and claiming the identity layer" is an analyst reading of the Mastercard and Visa actions; the announcements report the products and the BVNK deal, not the strategic-concession framing.
  • Framing tokenization’s risk as "reflexive," correlated and faster failure, synthesises the IMF, BIS and FSB warnings into one thesis; the "fail together" characterisation is ours and is contested by the scale-skeptic reading.
  • Reading the CBDC divergence as "about remuneration" and the dollar’s erosion as "running to gold, not rivals" are analyst reconciliations of the ECB, Atlantic Council and IMF COFER evidence, not claims any single source frames that way.
  • The 1970s-80s money-market-fund analogy is an interpretive overlay chosen to discipline the deposit-migration thesis, not a claim the sources make.

Editorial framing

The title "The Summer Money Left the Bank" is an editorial compression: deposits are being contested and are beginning to migrate (money funds at a record, stablecoins inside the perimeter, tokenized deposits in build), but no mass deposit exodus has occurred, so the phrase names the direction of travel, not a completed event. Consolidating four parallel shifts under one "money left the bank" synthesis is likewise an editorial choice that compresses independent analyses into a single decision-actionable pattern; readers who prefer to treat the four themes as separate registers can do so without losing any underlying claim. As the inaugural edition, this report sets a baseline rather than grading a prior scan.

The Futures Wheel and the in-body figure SVGs are analyst constructions built from the sourced claims; they carry no separate source-register entry, so a claim or quantification flag on those elements is expected and disclosed here rather than being a provenance failure.

Login