Phemex CEO Federico Variola calls AI a “net negative” for crypto
He argues AI is pulling liquidity from crypto while raising exploit risk and security costs that could centralize DeFi.
Phemex CEO Federico Variola said AI has been a “net negative” for crypto, citing capital rotation into AI, more capable attackers, and rising cybersecurity costs for protocols. He still endorsed limited use of AI agents for portfolio construction and trade decision support, with humans retaining final execution decisions.
Key Takeaways
- Phemex CEO Federico Variola said AI has been a “net negative” for crypto and added, “It’s difficult to be bullish about AI in crypto.”
- Variola framed AI as a cross-market competitor for capital, saying liquidity has been significantly diverted from crypto into the AI sector.
- He warned AI is improving offensive capabilities like social engineering and vulnerability discovery, raising the baseline security burden for DeFi and self-custody.
- A July incident tied to a Coldcard hardware wallet flaw saw roughly $116 million in Bitcoin drained from more than 5,200 addresses, with the flaw described as widely believed to have been found via malicious AI use.
Phemex CEO: AI Has Been a “Net Negative” for Crypto
Federico Variola put a clean label on a messy trend. AI, in his view, has been a “net negative” for crypto.
He made the comments on the Chain Reaction show on Sep. 16, 2026. The argument had three pillars: capital rotation away from crypto, attacker capability rising faster than user defenses, and a security cost curve that could push the industry toward larger, more centralized operators.
Variola’s sentiment read was blunt. “It’s difficult to be bullish about AI in crypto,” he said. He then tied the macro and the micro together: “Liquidity have been significantly diverted to to that industry on one side. On the other hand, AI has empowered a lot of bad actors that have been exploiting protocols.”
The tension is that Phemex itself announced an AI-focused transformation in February, with plans to embed AI across product development and internal operations. Variola’s latest remarks were not about Phemex’s roadmap. They were about the net effect of AI on the broader crypto market.
Capital Rotation Meets Security Risk: The Trader-Relevant Read-Through
Variola’s “liquidity diversion” claim matters less as a precise flow statistic and more as a positioning narrative. No dataset or timeframe was provided, so it cannot be treated as measured fact. But the framing is familiar: crypto is not competing only against other risk assets, it is competing against adjacent tech trades for attention, capital, and talent.
For traders, that narrative tends to express itself as a higher bar for sustained bids in smaller tokens and higher sensitivity to risk-off impulses. If marginal capital is being pulled toward AI, crypto needs either stronger catalysts or cleaner market structure to keep liquidity from thinning at the edges.
The second leg is more immediate. If AI makes exploits cheaper to attempt and harder to detect, the market’s risk premium rises in the places that rely on user competence and operational hygiene: DeFi (decentralized finance, where trading and lending run via smart contracts) and self-custody (holding keys directly rather than leaving funds with a custodian). Variola explicitly warned that as AI becomes more pervasive, the threat surface expands from code to people.
He also argued the “fixes” can be centralizing. “It’s difficult to envision a world in which AI is going to favor crypto specifically as an industry, since a lot of the fixes that we see actually encourage more centralization rather than less centralization,” he said. The implication is structural: if security becomes a scale game, the counterparty that benefits is the incumbent with budget, tooling, and process.
AI-Enabled Exploits and the Security Budget Squeeze
Variola’s offensive-case framing leaned on two vectors: social engineering and vulnerability discovery. Social engineering is the low-tech attack that scales with better persuasion, better targeting, and better automation. Vulnerability discovery is the high-tech side, where faster code review and pattern matching can compress the time between a new deployment and a viable exploit.
He described both as already moving. “AI has empowered a lot of bad actors that have been exploiting protocols, whether with social engineering or… finding vulnerabilities,” Variola said. He added that small teams working on protocols will no longer operate without a massive cybersecurity budget.
That is the centralization pressure point. If the minimum viable security spend rises, smaller teams either ship slower, ship riskier, or outsource security to a narrow set of providers. None of those outcomes are clean for decentralization.
The source also pointed to a concrete loss event used to illustrate the concern. In July 2026, attackers drained roughly $116 million in Bitcoin from more than 5,200 addresses tied to a Coldcard hardware wallet flaw that was described as widely believed to have been discovered through malicious use of AI. “Widely believed” is not attribution, and the packet does not include a technical writeup confirming AI involvement.
Still, the incident captured the direction of travel. Coinkite CEO Rodolfo Novak warned developers at the time that the “sober reality” is AI-assisted code review can uncover bugs that outpace the industry’s most seasoned experts. That is a warning about speed. The defender’s loop has to tighten.
Variola also linked the threat model back to adoption. “As AI becomes more pervasive, and whether it is your devices being hacked or social engineering, or all these kinds of strategies that are empowering threat actors, that makes DeFi a lot less appealing for a retail user because you have to worry about so many things that you didn’t have as much before,” he said.
What Would Change the Thesis: AI as Defense, and Agents as Assistants
The counterweight in the same packet is that AI is not only an offensive accelerant. CertiK senior blockchain investigator Natalie Newson said in April 2026 that “AI can also be one of the biggest defenses,” even as she warned it was making attacks more sophisticated.
If that defensive tooling diffuses broadly, the “net negative” framing weakens. The near-term signals are practical rather than philosophical: whether major security firms and protocol teams start publicly reporting higher AI-driven attack volume, especially social engineering, and whether that forces visible budget expansion that smaller teams cannot match.
The Coldcard example also needs clarification to carry analytical weight. Follow-on disclosures or technical writeups would need to establish whether the flaw was actually discovered with AI assistance versus conventional research. Right now, the claim is suggestive, not proven.
Variola also carved out a user-side use case that does not require handing the keys to a bot. He said AI agents can help investors build portfolios or make better trading decisions, but he does not expect them to replace the user’s final decision to execute. “At the end of the day, still it will be up to the user to make the final decision. So I don’t think that agents will ever replace that action of taking the trade,” he said.
Adoption metrics for those assistive agents are the other tell. If agents remain decision-support tools rather than autonomous execution engines, the upside case looks more like better UX than a new source of DeFi growth.
My Take: AI’s Near-Term Edge Looks Like an Arms Race, Not a DeFi Growth Catalyst
The threshold that matters is whether AI becomes a defensive equalizer faster than it becomes an offensive multiplier. Variola is effectively arguing the opposite: that the first-order effect is higher baseline security spend, and that advantage accrues to teams with budget and process.
If security firms can credibly demonstrate that AI-assisted audits, monitoring, and code review are lowering loss rates for smaller teams, the “centralization by security” thesis breaks. If not, AI reads less like a growth catalyst and more like an arms race that prices out the long tail of DeFi builders.
