Originally reported by PANews on November 10, 2026, at 16:25 (UTC+8), in a Wu Blockchain podcast recap.
The pitch that crypto infrastructure could serve AI agents is gaining attention from some investors. But in this account of two venture funds, the immediate constraint is less futuristic: many existing investments are difficult to monitor, sustain or exit. The recap also reports that a16z renamed its crypto investment arm a16z Frontier and broadened its scope to include AI and robotics. The two fund representatives describe their own decisions, not results for the VC industry as a whole.
Waterdrip Capital shifts its investment mix
Waterdrip Capital co-founder Dashan said the firm has invested in nearly 300 projects since its founding in 2017. In a review conducted in the first half of 2026, he said fewer than one-third—roughly 60 to 70—were still operating well by the firm’s criteria. He estimated that 20%–30% had involved some combination of founders disappearing, projects going to zero, defaults or damage to investor rights. Those categories should not be collapsed into a claim that every such project was a scam.
Of the more than ten projects Waterdrip invested in during 2026, Dashan said only three or four were crypto-related; the rest were mainly in AI and chips. His proposed role for crypto is as infrastructure for applications such as AI-agent payments, data privacy and auditable operations, rather than as a standalone investment story. He argued that wider agent adoption would have to come first; the interview offers no evidence that agent payments are already a large market.
A smaller fund pauses new deals
Xiao Yang, who works at an unnamed fund of approximately $20 million, described a different response. The fund invested in more than 70 projects in 2024 and about 100 in total, but stopped making new investments after the first quarter of 2025 across crypto and other fields. He said roughly two-thirds of its portfolio was still operating, while more than half of its projects were difficult to communicate with effectively.
Xiao Yang described the fund’s paper return as slightly positive and close to break-even, but said the amount it could actually realize might leave it with a loss. Unlocked tokens and project valuations are not the same as cash from an exit. He said some of its AI investments had performed well on paper, while one AI graphics-analysis investment of nearly $300,000 had been written down to zero.
Together, the accounts show why a shift in investor attention toward AI is not, by itself, evidence of adoption or returns. Both speakers emphasized founder reliability, real user needs and sustainable businesses over short-lived market themes. Their portfolio figures and assessments are their own reported estimates.