Consensus HK VCs: decentralized AI in a trough but market is recalibrating — play a 15‑year game and focus on proven verticals; high‑profile exits and safety disclosures deepen turmoil

At Consensus Hong Kong 2026, VCs Anand Iyer (Canonical Crypto) and Kelvin Koh (Spartan Group) described decentralized AI protocols as being in a "trough" as capital and talent move away from overhyped GPU marketplaces and broad attempts to replicate large models. They criticized expensive, broad infrastructure plays and warned that simple wrappers around models like ChatGPT no longer attract funding, advising founders to prioritize user-focused products, proprietary data, regulatory or go‑to‑market advantages, and defensible moats.
AI Analysis
VCs at Consensus HK said capital and talent are shifting away from GPU marketplaces and model-replication projects, calling the sector in a "trough"; they criticized expensive, broad infrastructure plays and warned that simple model wrappers no longer attract funding while urging focus on utility, proprietary data and defensible moats—facts that imply downside for speculative decentralized-AI plays but opportunity for product-focused projects.