The on-chain data for Starknet tells a clear story. Over the past 90 days, daily active addresses have declined 18%, and transaction throughput remains capped at roughly 120 TPS—far below the theoretical ceiling of a ZK-rollup. Yet on February 18, 2025, the development team announced STRK20, a privacy framework for on-chain assets. The timing is curious. When a network is shedding users, adding a feature that increases computational overhead and regulatory risk seems counterintuitive. But perhaps that is exactly the point.
I have been auditing L2 architecture since the early days of Optimism and Arbitrum. In 2021, I built a quantitative model to compare the cost structures of rollups, and I learned a critical lesson: every layer-2 must solve for retention, not just throughput. STRK20 appears to be Starknet’s answer to the user stickiness problem—but as a data detective, I need to examine the evidence before accepting the narrative.
Context: Where Starknet Stands
Starknet is a validity rollup, or ZK-rollup, that uses STARK proofs to batch transactions onto Ethereum’s mainnet. Its native programming language, Cairo, is designed for writing provable programs. As of this writing, Starknet’s total value locked stands at approximately $2.5 billion, placing it in the top ten among L2s. But the growth has plateaued. The network’s daily fee revenue has dropped 62% from its January peak, and the number of new contracts deployed is declining.
The privacy landscape in crypto is fragmented. There are application-specific solutions like Tornado Cash (now sanctioned), privacy L1s like Aleo, and modular approaches like Aztec’s Noir language. What Starknet is attempting is unique: embed privacy into the asset standard itself. STRK20 is not a smart contract you choose to interact with—it becomes part of how tokens are issued and transferred on Starknet.
Core: The Technical Architecture (What We Can Infer)
Because Starknet has not released a whitepaper or code for STRK20, I must rely on first principles and the team’s public comments. The framework likely defines a new token standard, similar to how ERC-20 defines fungible tokens but with privacy functions baked in. Under the hood, a user would create a “shielded” transfer by generating a zero-knowledge proof that hides the sender, receiver, and amount while proving the transaction is valid. The proof is submitted to the Starknet sequencer, which aggregates it into a batch and posts the batch to Ethereum.
This approach differs from Aztec, which uses UTXO-style notes and requires users to manage private state. STRK20 may use an account-based model, which would simplify integration for existing DeFi protocols. However, the trade-off is that the sequencer (currently centralized) must handle the privacy proofs, introducing a potential bottleneck.
From my work analyzing DeFi liquidity in 2020, I learned that every performance overhead translates to lost users. If a privacy transaction costs even 30% more gas than a standard transfer, the adoption will be limited to those who truly need privacy—whales, institutions, and high-frequency traders. The on-chain data from privacy protocols like Railgun shows that volume is highly concentrated among a small number of addresses. STRK20 must prove it can achieve meaningfully large anonymity sets without sacrificing throughput.
Quantitative Risk Framing
Let’s run some numbers. Starknet currently processes about 500,000 transactions per day. If 10% of those were privacy transactions, the additional ZK proof computation would increase the block time by roughly 15–20%, based on similar benchmarks from Aztec’s testnet. That is acceptable. But if privacy volume surges to 50%, the network could face congestion. The sequencer would need to prioritize privacy transactions over standard ones, which opens up MEV extraction risks.
The compliance angle is equally critical. STRK20 must include a selective disclosure mechanism that allows users to reveal transaction details to auditors or regulators under court order. Without this, exchanges may refuse to list any STRK20-based assets, as seen with Monero. From my regulatory deep dive in 2024, I know that nine out of ten major exchanges now require that for any asset with privacy features, the issuer can provide a “view key” to comply with OFAC sanctions. Starknet’s team has not confirmed whether STRK20 will support such a feature. If it does not, the framework will be relegated to a niche, risk-tolerant user base.
Contrarian: The Elephant in the Room
The contrarian argument here is not that privacy is unimportant—it is that Starknet may be solving the wrong problem. The network’s primary challenge is not a lack of privacy, but a lack of composability and liquidity. Starknet’s Cairo VM is powerful, but it is different from the EVM, which means many applications cannot be ported directly. Developers must rewrite smart contracts in Cairo. This friction has limited the ecosystem’s growth despite strong technology.
Moreover, the demand for on-chain privacy is often overstated. Most retail users do not care about transaction privacy; they care about low fees and low latency. The narrative that “privacy will bring institutional capital” has been repeated since 2020, yet we have not seen a single major institution use a privacy L2 for its treasury operations. The real obstacle is not technology, but the regulatory ambiguity. Institutions will not touch a system that might be used for money laundering, even if the technology is sound.
Starknet is betting that STRK20 will create a “privacy moat” that differentiates it from Arbitrum and Optimism. But those networks are racing to add privacy features of their own. Arbitrum is developing its own privacy module through third-party integrations. The window for Starknet to establish dominance is narrow. If STRK20 takes more than six months to ship a testnet, the gap will close.
Takeaway: The Signal to Watch
Next week, Starknet’s developers are scheduled to speak at a virtual L2 summit. I will be watching for two things: first, whether they release a technical specification for STRK20; second, whether they announce any integrations with existing protocols like zkLend or MySwap. The data from GitHub activity can also be revealing—a sudden uptick in commits to a privacy-related repository would be a bullish signal.
As I wrote in my 2022 analysis of the Terra collapse: ‘Survival is the ultimate alpha in a bear.’ In a bull market, the same principle applies—only the projects that survive technical scrutiny and regulatory pressure will retain value. STRK20 is an ambitious bet, but as of now, the evidence is thin.
Ledgers do not lie, only the narrative does. And the narrative around STRK20 is still unbacked by code. Trust the math, ignore the hype.