Aztec just relaunched zk.money with a hard cap of $2,500 per transaction and $50,000 in total daily deposits across the entire network. Read that again. In a bull market where freshly funded L2s are pushing nine-figure TVL campaigns before their testnets even stabilize, Aztec is voluntarily capping its flagship privacy wallet at the daily throughput of a mid-tier car dealership.
That number is not a marketing decision. It is a risk-fuse.
The Context You Need
zk.money is the application-layer privacy wallet built on Aztec Network, the zk-rollup L2. The original v1 launched in 2021 and pulled in over 75,000 wallets handling roughly $100 million in volume before the team shut it down in 2023. CEO Joe Andrews framed that closure bluntly: privacy payments were not a failed proposition, they simply required infrastructure that did not yet exist.
Now it's back. Deposits cost $0.35, withdrawals $0.20, and users get 100 sponsored transactions per day. Supported assets are USDC, USDT, and DAI, with everything converted to DAI on arrival. Human-readable identities like bob.zk.money route payments through ENS. Deposits and withdrawals are both publicly visible and screened against sanctions lists, while the middle layer stays private.
That last detail is the entire thesis. This is not Tornado Cash. It is a deliberately regulator-compatible privacy tool.
The Part Nobody Is Pricing Correctly
Here is where my audit background forces me to slow down, because the interesting information is not in the feature list. zk.money went live while a known critical vulnerability in the Alpha V5 proof system remains unpatched, and the fix is not scheduled until V6, which lands in late 2026.
The vulnerability matters. In the Aztec V5 proving system, a specific flaw allows malformed transactions to be accepted — bad proofs passing validation. I have audited reentrancy in fund distribution logic before, at three ICO projects back in 2017, and I can tell you the failure mode is comparable in spirit: the system accepts an input it should reject, and the accounting drifts from reality.
So how does Aztec ship anyway? Two layers of defense.
First, the economic constraint — the $2,500 and $50,000 caps. These caps do not prevent an attack. They bound the blast radius. That is the entire design intent, and it is an unusually honest admission from a team that it does not fully trust its own cryptography yet.
Second, and far more consequential, is Oxide. The Oxide protocol operates a sealed server — a trusted execution environment enclave — that co-signs every send and withdrawal. The model is dual-key. Even if the Aztec proof system is compromised and a bad transaction gets accepted, an attacker still cannot extract clean funds unless the enclave is also breached.
The Contrarian Angle: TEE Reliance Is a Regime Shift, Not a Feature
Most commentary will treat Oxide as a clever safety net. I think that reads it backwards.
In a pure zero-knowledge privacy design, the trust assumption is cryptographic. You trust math. Tornado Cash and RAILGUN, whatever their regulatory problems, never asked users to trust a sealed server run by anyone. zk.money now does. That is not a security upgrade. It is a trust-model downgrade — from minimal-trust cryptography to a multi-party failure assumption where you must believe both the L2 proof system AND an opaque enclave operator will not fail simultaneously.
And here is the operational risk almost nobody is modeling: withdrawals require enclave co-signature. If Oxide halts, degrades, or is compromised, fund movement stops entirely. A centralized TEE with undisclosed audit status, undisclosed operator, and undisclosed code is a single point of failure sitting directly beneath a privacy product.
I want to be fair here. The team's willingness to cap deposits at a level it can absorb, and to publicly acknowledge a known bug rather than quietly patch and pretend, is more intellectually honest than most of what I reviewed during the 2020 DeFi yield cycle. Honesty, however, is not safety. Those are orthogonal variables.
The strategic logic is legible once you stop expecting a normal launch. Aztec is running a controlled experiment. Validate demand now, ship the fix in V6, then scale. The tradeoff is that this high-risk, capped state could persist for years, not months.
Two other quiet risks deserve flagging. DAI-only settlement concentrates exposure into the MakerDAO stability mechanism. And Ethereum's own Hegotá upgrade proposal for native L1 private transfers is the structural threat nobody wants to name — if base-layer privacy arrives, the value proposition of a compliance-friendly L2 privacy wallet compresses fast.
Takeaway
The cap is not a weakness. It is the most honest number in the announcement. Aztec is telling you exactly how much it is willing to lose if its own cryptography fails, and the answer is $50,000 a day. Position your expectations around a validation-stage product, not a payments network. The question worth holding onto is whether a compliance-first privacy tool can survive long enough for its own cryptography to catch up — before Ethereum builds the same primitive into the base layer.
