Those who have been around for long enough remember DeFi – a financial market so decentralised that you weren’t subject to any single actor shutting down their services, frontends, or backends. Sufficiently open source so that you could interact, validate, and verify both logic and transactions directly by utilising third-party tools or through interacting with the blockchain using your own node.
Then came CeDeFi – a promise of creating a centralised layer of services on top of those decentralised protocols to allow more people to access the primitives through more “user-friendly” interfaces. Think a mash up between traditional finance experience powered by decentralised finance building blocks. Therein lay the risk – a single, often custodial, service provider orchestrating all the money movement, rebalancing, transactions, and so forth. Just one such a provider, Celsius, lost almost $5b of user funds. It blew up, not because of DeFi, but because of fraud.
Now, in 2025, we’ve seen the rise of a new era of platforms. Platforms I call DeCeFi. These platforms utilise decentralised primitives but incorporate a major centralised component in providing a financial service.
Here are two examples of what the architecture might look:
| Perpetuals Exchange A | Savings App B | |
|---|---|---|
| De | Smart contracts hold deposited assets and canonical state root powering verifiable TVL attribution | On-chain, verified, source-available lending protocol powering loan origination and yield distribution |
| Ce | Closed-source single sequencer L2 enabling high throughput OR Insufficiently decentralised validator-set controlling the L2 bridge withdrawals enabling security measures in case of breaches | Embedded wallet enabling easy social login without private key export features |
| Fi | Perpetual swaps trading | Interest on assets (yield) |
Table 1: Examples of DeCeFi architectures
These services have begun to proliferate for a reason. Decentralisation is expensive, makes progress slower, and reduces focus (source? first-hand experience). The regulatory vibe has also been shifting towards the more permissive. There are now many good reasons to build in a centralised manner. A product falling under the DeCeFi category is not, in itself, a reflection on neither the security of the product nor the motivations of the team.
Furthermore, DeCeFi is a spectrum. It’s a spectrum of choices in the aspects of a product’s centralisation.
Some products might have a very valid reason for using a third-party provider to manage your keys, be it easier logins or end-to-end control over the user-experience. A single sequencer L2 architecture allows lower latency and higher throughput – and it’s easier to set up and maintain.
Some of these design choices also leave the option of making a DeCeFi product into a DeFi product onto the user themselves. Embedded wallets with the option for private key export allow the user to take control at any time – if they realise to do so before it is too late.
The point?
As DeCeFi products labeled as DeFi continue to proliferate, understand the centralisation vector. Accept it, mitigate it, or reject the product. It’s your money and it’s your choice.
If you are building a DeCeFi product, think about the priorities and what makes sense. Also consider DeFi ethics.