DeFi and CeDeFi: Andrei Grachev's Take on Risks in DeFi and CeDeFi’s Solutions
Updated On 31 August 2026
Published On 1 May 2026

The following was originally published on X by Andrei Grachev, Managing Partner at DWF Labs.
DeFi or CeDeFi? A Story of Lost Responsibilities
The fundamental problem with the DeFi market today is that its infrastructure is not designed to protect itself from AI-driven hacker attacks.
Pursuing full decentralization and purely smart-contract-based execution sounds good, but it has created risks that cannot be fixed in a short period of time, and it has reduced the flexibility of the overall design.
In my opinion, the future is not fully automated and decentralized. It is a merge of human control and decentralized execution. Humans can be exploited through social engineering; algorithms can be exploited through hacks. We need two layers of control, where each layer has its own unique advantages and solves the risks of the other.
Money is not the fix
Not everything can be solved by capital injections and recovery funds. Free money, community support, and donations are great, but they breed irresponsible behavior and the feeling that any problem is not really a problem, because someone else will fix it.
Take 2022. After the FTX collapse, the whole industry was looking for funds to pay salaries, maintain liquidity, keep building, and launch new projects. The financial injections that followed that demand — investments, OTC deals, M&A, and more — helped many founders and projects stay alive.
The FTX case was a black swan. It was not caused by any single market participant. So the response made sense: fix it with money, learn from the mistakes, rebuild confidence, and move forward.
But there is a fundamental difference between that situation and the current attacks on DeFi markets. Yes, the community and some great people and companies donated a lot of support — the bad debt on Aave was patched somehow, Drift received funds for recovery — but the fundamental risks are still there. It is a question of "when," not "if."
When we stop paying for our mistakes, we stop learning, and we stop improving ourselves.
The threat is accelerating
The rapid growth of AI, and the quantum technologies that will follow, will exploit everything that can be exploited. It is a hard truth, but it is the truth we have to listen to as we adjust our strategies.
I have been here since 2016, and from the earliest days the whole industry wanted to be recognized, wanted to be big. The good news: we are recognized, and we are big. But this is exactly the moment to abandon childish, idealistic plans, face the challenges bravely, and act responsibly.
Not for others — for ourselves. Everything starts with us. We should forget about any outside help or support: if you can manage your risks, you are sustainable; if you can't, you are not. Simple.
The lessons keep repeating
DeFi Summer 2020 showed that yields can be enormous — and then showed that losses can be even bigger. 2022 showed us that even giants can fall, and that there is no safe haven. 2025 showed that "risk-free" farming can be hugely profitable, and then showed that it was never really risk-free at all, eliminating the entire narrative.
Today we face new challenges again, and many bright minds are building the things that will solve them. Even your favorite, Hyperliquid, behaves in a centralized way when it comes to risk and insurance funds — because as traders, they know DeFi logic will be exploited.
DeFi is a great technology. But like any technology, it needs human control and supervision. Time is flying fast, and my bet is on CeDeFi.