Perp DEXs still don’t work for institutions, consensus panelists explain why

Institutional investors have increasingly gained exposure to bitcoin BTC$79,512.22 and other major tokens through ETFs and centralized exchanges.

However, they have largely stayed away from decentralized exchanges (DEXes) offering perpetual (perp) futures tied to crypto and tradfi assets, panelists said at Consensus Miami, citing security risks and a mismatch between DeFi’s permissionless design and institutional identity and compliance requirements.

The session titled “Perp DEX Explosion: Bullish Volumes & Bear Market Resilience” featured Wizard of SoHo, a veteran trader and family office manager; Michaël van de Poppe, founder and CIO of MN Fund & MN Capital; and Michael Anderson of Canary Labs. Jason Atkins, chief commercial officer at liquidity provider Auros, moderated the discussion.

The discussion focused on perpetual-focused decentralized exchanges and what it would take for them to attract institutional capital and scale up.

Wizard of SoHo said that institutions are unlikely to move onto perp DEXs easily due to recurring security/exploit risks highlighted by the recent multi-million-dollar hack of Drift, and that the next major competitive battleground for all perp DEXs will be whether any of them can safely onboard institutional capital.

“How do you convince the big institutional players to go on the perp devs? I think that’s going to be the biggest challenge, especially given the exploit on Drift. And, you know, we’ve had a lot of exploits lately,” he said.

Canary Labs’ Anderson struck a cautious tone on decentralized finance, saying he is reluctant to use it despite having explored parts of the ecosystem.

“I’m scared to use DeFi right now,” he said. “It does feel like a bit of a minefield, and you’re just waiting for the next headline each day.”

Anderson added that while activity has picked up in some areas, particularly from Asia amid tighter KYC enforcement on centralized exchanges, the overall environment still feels risky.

“Right now, it feels slightly dangerous on the product side,” he said.

Anderson argued that the risk perception makes it difficult to see large institutional players adopting decentralized exchanges at scale, especially compared with centralized platforms.

“I think it’s gonna be very difficult for some of the larger firms to use it on the institutional level, versus some of the centralized exchanges,” he said.

Anderson also pointed to product innovation gaps as another constraint, noting that centralized exchanges are increasingly integrating trading tools, such as bots, into futures markets. In contrast, decentralized exchanges have yet to match that pace of development.

KYC, or know-your-customer verification, is another key point of divergence. DeFi is built around open, permissionless participation, where users can interact without formal identity checks or traditional onboarding requirements.

Institutions, by contrast, operate under strict regulatory obligations and must meet full KYC and compliance standards, which makes that permissionless model difficult to adopt at scale.

“Crypto wants to be more non-KYC,” he said, “but to bring on institutional [players] you need to have some form of KYC at the larger size.”

The discussion also broadened into adjacent themes shaping market structure, including the rise of AI-driven trading tools and Hyperliquid’s dominance.

Michaël van de Poppe said AI agents are effectively an evolution of algorithmic trading, rather than a fundamentally new concept.

“To be honest, I think that AI agents are just the next level algorithmic trading anyways, so it’s just a little different execution,” he said. Responding to a moderator’s point about reduced human control in automated systems, he acknowledged the shift in oversight but argued the direction is inevitable.

“Yeah, there are some risks, but I think that at the end of the day, we are not going to be trading ourselves anymore. Nothing will be manual,” he said. “AI agents will be doing it for us, and they are probably better.”

van de Poppe added that the technology is still early and highly dependent on how it is deployed.

“If you start using those AI protocols or LLMs and you’re not putting in the right context or framework, it’s going to build a bad trader for you,” he said. “So if you are not a good trader, then it’s not going to build anything for you.”