Field Notes on Radix
A ground-up rethink of what a DeFi-native L1 requires — evaluated across scale, security, developer experience, and end-user experience. Where the architecture is genuinely differentiated, and where execution risk remains real.
Most L1s optimise one axis. Solana chose throughput at the cost of decentralisation and stability. Ethereum chose security and composability at the cost of scale. Polkadot and Cosmos chose interoperability and sacrificed atomic composability across chains. Radix is a rarer kind of attempt: a ground-up redesign that targets all four axes simultaneously — and, in several cases, formally demonstrates results that its competitors have simply conceded as unsolvable.
That does not make it a certain bet. Ecosystem size is small, the non-EVM choice is a deliberate trade-off that severs access to the largest developer pool in the space, and the founder who conceived the core technical approach died in July 2025. What follows is a structured assessment of where the architecture holds up under scrutiny and where the remaining risk sits.
Core technical advantages
Scalability
Cerberus uses an effectively infinite pre-sharded ledger rather than a fixed shard set. Each transaction specifies which shards it touches; unrelated transactions execute in parallel with no coordination overhead. Adding nodes linearly increases throughput — there is no theoretical ceiling. Ethereum depends on L2 rollups for throughput. Solana and Sui are bounded by single-node vertical limits.
No ceiling on TPSComposability
Most sharded systems break composability — a DeFi transaction touching two shards cannot atomically succeed or fail as one unit. Cerberus "braids" multi-shard consensus into a single atomic outcome. This is the key unsolved problem in blockchain scaling. Polkadot, Cosmos, and Ethereum sharding all sacrifice composability; Radix formally proves it is preserved.
UC Davis formal proofAsset Security
Solidity treats tokens as smart contracts — a developer bug can allow assets to be created from nothing or stolen. Scrypto (Rust-based) treats assets as first-class physical objects with enforced conservation laws baked into the VM. The Radix Engine guarantees assets cannot be duplicated, destroyed without intent, or extracted without authorisation — regardless of developer mistakes.
Exploits eliminated at VM levelUser Experience
Every transaction on Radix shows a plain-language manifest before signing — not a hex blob. Smart Accounts replace seed phrases with multi-factor recovery using badges and access controllers; losing a device does not mean losing assets. On all other L1s, blind signing and seed phrase management remain unsolved mass-adoption barriers. This is built into the L1 stack, not bolted on.
No blind signingToken Model
Tokens on Radix are native network resources, not separate smart contracts. This eliminates entire classes of ERC-20 approval exploits, token import friction, and the need for wallets to manually discover balances by querying hundreds of contracts. Malicious token injection — a persistent MetaMask problem — is structurally impossible.
Incentives
When one developer's Scrypto blueprint is reused by another dApp, a royalty can automatically flow to the original author — enforced at the protocol level, not via a side agreement. No other L1 has this. It creates financial incentive for a composable, reusable component ecosystem rather than siloed contracts.
Identity
Users can maintain separate identity personas per dApp — logging in with wallet-controlled identities rather than passwords. This is a full-stack identity layer built into the network, not a third-party add-on. No other L1 ships this natively.
Design Philosophy
Ethereum, Solana, and Avalanche are general-purpose VMs that DeFi happened to be built on. Radix was designed exclusively around the requirements of financial applications from the ground up — consensus, VM, language, wallet, and identity are co-designed for that single use case.
"The most important unsolved problem in blockchain scaling is atomic composability across shards. Every other chain has either abandoned sharding or abandoned composability. Radix is the only one to formally prove both are preserved simultaneously."
Head-to-head comparison
| Dimension | Radix | Ethereum L1 | Ethereum L2s | Solana | Sui | Polkadot |
|---|---|---|---|---|---|---|
| Scalability ceiling | None (linear) | ~30 TPS | High per-rollup | Hardware-bound | Hardware-bound | ~1,500 TPS/chain |
| Atomic composability | ✓ across shards | ✓ single chain | ✗ broken cross-L2 | ✓ single chain | ✓ single chain | ✗ broken cross-chain |
| Asset security model | VM-enforced conservation | Smart contract logic | Smart contract logic | Smart contract logic | Object model (partial) | Smart contract logic |
| L2s required to scale | ✓ No | ✗ Yes | N/A (is L2) | ✓ No | ✓ No | ~ parachains |
| Human-readable txns | ✓ native | ✗ | ✗ | ✗ | ✗ | ✗ |
| Seed-phrase-free recovery | ✓ native | ✗ | ✗ | ✗ | ✗ | ✗ |
| Native assets (no ERC-20) | ✓ | ✗ | ✗ | ~ SPL tokens | ✓ objects | ~ |
| Developer royalties | ✓ protocol-level | ✗ | ✗ | ✗ | ✗ | ✗ |
| Smart contract language | Scrypto (Rust, asset-oriented) | Solidity | Solidity | Rust | Move | Rust / ink! |
| EVM compatible | ✗ (deliberate) | ✓ | ✓ | ✗ | ✗ | ~ |
| Ecosystem / TVL | Small (~$41M TVL) | Dominant | Large | Large | Growing | Medium |
The honest caveat
Radix's technical architecture is genuinely superior for DeFi across nearly every axis — but the non-EVM choice is a deliberate trade-off: the existing Solidity developer pool and EVM tooling cannot be reused. Ecosystem size remains small. Dan Hughes passed in July 2025, triggering a 40% price drop, and the Hyperscale upgrade (Rust-based mainnet targeting ~2027) is still incomplete. The thesis is architecturally sound. Execution risk is real.
The comparison above is an attempt at honest accounting rather than advocacy. On most of the dimensions that matter for financial applications — composability guarantees, asset security, user-facing friction, and scaling architecture — Radix holds up. The gaps are real too: $41M TVL is two orders of magnitude below serious DeFi destinations, the developer ecosystem is thin, and the Hyperscale upgrade that delivers on the TPS promises has not yet shipped to mainnet.
What the architecture does not do is provide cover for poor execution, regulatory setbacks, or the straightforward risk that the market never arrives at the verticals where Radix has structural advantages. Those are separate questions from whether the technical design is correct. On the technical design, the case holds.