TL;DR
- While the stablecoin discourse argues about Tether vs. Circle vs. x402, a third layer has quietly developed: tokenized bank deposits. JPM Kinexys currently does ~$5B in daily volume on public chains (April 2026, JPMD on Base).
- McKinsey frames the reality as a three-layer stack: Stablecoins (money in motion) + Tokenized Deposits (money at rest) + Tokenized Central Bank Money (settlement money).
- Most directly relevant for Europe: the ECB launches Pontes as a production service in September 2026, connecting DLT platforms to TARGET. The "pilot" label was quietly dropped.
- Consequence: "Stablecoin disrupts banking" misses the point. Banks are building their own layer — and today it's already the largest.
The Gap in the Discourse
In the first three posts of this series, we measured the magnitudes (x402 vs. Tron-USDT), compared the geopolitical settlement visions (PAPSS, CIPS, mBridge), and drew the market boundary between agent-payments and remittances.
One layer is missing in the picture so far: what are Western, OECD-regulated megabanks actually doing right now? In the first three posts, they only appeared as "discussing disruption, but building nothing." That was imprecise. They're building. Just not what the stablecoin discourse expects.
The Three-Layer Model
In May 2026, McKinsey published in "Beyond Stablecoins: The Emerging Architecture of On-Chain Money" what Fireblocks and the ABA Banking Journal have been framing similarly since early 2026: on-chain money isn't one layer, it's three.
Layer 1: Stablecoins as "money in motion" Fast, low-value cross-border transactions where banking access is limited. USDC, USDT, FDUSD, PYUSD. This is the use case Tron-USDT effectively serves in the Global South — and that x402 addresses for agent-payments.
Layer 2: Tokenized Deposits as "money at rest" Treasury balances, interbank settlement, institutional payments at scale. Banks tokenize their own deposits, keep the funding relationship, and gain programmability on top. Examples: JPM Coin / JPMD (on Base), Citi Token Services, BNY pilots, the Canton Network with Goldman, UBS, JPM.
Layer 3: Tokenized Central Bank Money as "settlement money" The ultimate settlement asset without counterparty risk. That's what mBridge is doing with CBDCs in pilot. In the Western context: what the ECB is planning with Pontes as a production service from September 2026.
The point of the model: these three layers don't replace each other — they cooperate. Stablecoins for friction reduction in small transactions, tokenized deposits for institutional scale, CBDCs as final settlement guarantee between layers.
The Underrated Number: Kinexys Does $5B Per Day
In Post 1, we established that x402, at ~$28,000 daily volume, is six orders of magnitude smaller than Tron-USDT at $20-30B. The honest number in the Western banking stack is another order of magnitude above that:
Kinexys (JPMorgan's blockchain division) publicly stated in April 2026: $3T cumulative transaction volume since inception (2015), averaging $5B per day currently.
For context: that's a single bank with a single blockchain division. JPMorgan's entire payments division processes $10T daily — so Kinexys is a tiny fraction of that. But in the on-chain money comparison, it's a giant.
Ratios:
- x402: ~$28,000 / day
- Tron-USDT: $20-30B / day
- Kinexys: $5B / day (just this one bank)
Add Citi Token Services, BNY pilots, the Canton Network, and HSBC initiatives, and the tokenized-deposits layer is probably the largest part of real on-chain money volume — and it didn't appear in any of our three previous posts because tech Twitter doesn't care about this layer.
What Makes Tokenized Deposits Structurally Different
Three properties distinguish tokenized deposits from stablecoins:
1. Direct bank liability, not private reserve. When you hold USDC, you hold a claim against Circle, backed by Treasuries. When you hold JPMD, you hold a direct liability of JPMorgan — on your balance sheet it counts as a regular bank deposit, with deposit insurance and everything that comes with it.
2. Interest-bearing possible. Stablecoins typically pay no interest (in the US for regulatory reasons, in the EU under MiCA from 2025 structurally difficult). Tokenized deposits can bear interest because they're legally deposits. Naveen Mallela, Co-Head of Kinexys, explicitly framed this as a selling point against stablecoins in November 2025.
3. Compliance is built in, not bolted on. AML, KYC, OFAC screening, balance-sheet compliance — all part of the bank infrastructure sitting behind the token. Stablecoin issuers have to build these layers separately and have repeatedly come under regulatory pressure for them (USDT-NYAG settlement, BUSD shutdown, EU MiCA reserve rules).
This solves the fundamental bank incentive problem: banks can offer programmable on-chain money without losing their funding relationship to Tether or Circle. They modernize their balance sheet instead of hollowing it out.
Pontes: What the ECB is Concretely Building from September 2026
For Europe, the direct concretization is Pontes — part of the ECB's Appia initiative published in March 2026.
What Pontes is: a DLT settlement layer connecting blockchain-based market platforms to the Eurosystem's TARGET. This enables wholesale transactions happening on DLT platforms to settle in central bank money — no longer in USDC or USDT, but with the highest possible settlement anchor.
Important: the Q3 2026 launch label "pilot" was quietly removed from ECB communications in March 2026. Pontes launches as a production service in September 2026, not as an experiment. Appia is the larger framework with a 2028 target date.
Background: the ECB sees in the growing stablecoin dominance ($316B stablecoin market early March 2026, 58% USDT) a direct threat to EU monetary sovereignty. Cipollone explicitly warned in March 2026 that tokenized markets "will not scale on private digital money alone" — a statement politically loaded against the Trump-administration's pro-stablecoin policy.
For European practice this means: from autumn 2026 there will be a production DLT settlement layer in central bank money in Europe. That's neither a crypto pilot nor a think-tank whitepaper — that's infrastructure banks can use.
Five Implications for the Discussion
1. "Stablecoin disrupts banking" is doubly wrong
The statement misses two things: first, that banking in emerging markets wasn't disrupted by Tron-USDT but replaced (Post 1). Second, that OECD banking is currently building its own stablecoin-equivalent layer — and thereby pre-emptively neutralizing stablecoins in parts. Disruption presupposes that the disruptee stays passive. JPM, Citi, BNY, and the ECB are not passive.
2. The volume skew is more extreme than Post 1 described
If you add the tokenized-deposits layer, x402 isn't 6 orders of magnitude smaller than Tron-USDT — it's more like 7-8 orders of magnitude smaller than the summed on-chain money reality. That doesn't change the option-value logic for x402 plays — but it makes the magnitude comparison even more honest.
3. The compliance frame structures the layers cleanly
Stablecoins must build compliance separately (and regularly fail at it). Tokenized deposits bring it built-in. CBDCs are by definition compliance-conformant. That's not "crypto vs. banking" — that's "regulated vs. unregulated layers of the same on-chain stack".
4. Pontes is the most DACH-relevant story across all four posts
Unlike CIPS (China), PAPSS (Africa), mBridge (BIS pilot), or Kinexys (US bank on Base), Pontes is directly in the EU, directly in euros, directly in your regulatory framework. If you're a European treasury or fintech integrating on-chain money into your strategy, Pontes from September 2026 is the obvious entry point.
5. Three layers means: no layer wins
The most common trap in tech pitches is "X will win." Three-layer models say: no layer wins alone. Stablecoins stay relevant for low-value cross-border transactions (especially in emerging markets). Tokenized deposits stay relevant for institutional scale. CBDCs become the final settlement guarantee. Anyone pitching a single winner hasn't understood the model — or is selling a stake in a single layer.
Where Backtesting Arena Stands
We do exclusively Layer 1. Our x402 endpoints address agent-payments — exactly that part of Layer 1 designed for AI agents without account setup.
Tokenized deposits aren't our layer. We have no bank license, no balance sheet, no deposit infrastructure, and don't plan to build them. When Pontes goes live as a production service in September 2026, we won't integrate it — that would be Layer 3, not ours.
But: we understand why the other two layers are larger, and we understand why they'll stay. The option-value logic for x402 plays (see Post 1) remains correct under exactly this assumption — we're building a small, focused piece of Layer 1 for a growing agent audience, without pretending it's the only relevant layer.
That's the honest cluster synthesis: four posts, four perspectives, one picture — on-chain money is multi-rail, multi-layer, multi-jurisdictional. Anyone seeing only one layer doesn't see the market.
Conclusion
Three layers, three roles:
- Stablecoins = money in motion (small, fast, cross-border, often unbanked)
- Tokenized Deposits = money at rest (institutional, balance-sheet-integrated, programmable, already $B/day at Kinexys)
- Tokenized Central Bank Money = settlement money (CBDC, Pontes from Sep 2026, final settlement guarantee)
Anyone discussing "the future of payments" who only knows one of these layers is discussing a partial model.
That closes the cluster. Four posts, one picture: on-chain money is not a stablecoin religion. It's a layered, multi-rail reality — and honest discourse needs the full picture.
Data sources: McKinsey "Beyond Stablecoins" (May 2026), JPMorgan Kinexys 2026 Milestones report (April 2026), ECB Appia roadmap (March 2026), ECB Pontes page, Cipollone Brussels speech (March 2026), Fireblocks stablecoins-tokenized-deposits-CBDCs overview (November 2025), Capital Markets Law Journal (Buckley 2026), Ledger Insights. As of May 2026.
Backtesting Arena is a backtesting platform for systematic trading. Our x402 endpoints address Layer 1 — explicitly not tokenized deposits or CBDC settlement. Free tier available without signup.