Decentralized AI is in a trough but real opportunities are emerging, crypto VCs say

The intersection of crypto and artificial intelligence (AI) has entered a quieter, more selective phase, according to two prominent venture capitalists.

Anand Iyer of Canonical Crypto and Kelvin Koh of Spartan Group described the current climate as a post-hype moment for decentralized AI protocols, with capital and talent shifting toward more focused, utility-driven applications during Consensus Hong Kong 2026.

“I think we’re in the trough right now,” said Iyer, whose San Francisco-based firm backs early-stage infrastructure and applications built on decentralized networks. “We went through a frothy period. Now it’s about figuring out where the real strength lies.”

Both Iyer and Koh criticized what they see as overinvestment in GPU marketplaces and attempts to build decentralized alternatives to large AI models like those from OpenAI or Anthropic. The capital required, Koh noted, is “night and day” compared to what’s available in crypto.

Instead, they see potential in purpose-built, full-stack solutions, tools that start with a specific problem and build down to the model, compute, and data layers.

Iyer pointed to startups skipping expensive SaaS tools and using AI to build custom internal systems in days. “Speculation won’t drive product anymore,” he said. “We have to think about users first.”

Both investors emphasized the importance of proprietary data, regulatory advantages, or go-to-market edges as new forms of competitive moats.

For founders looking to raise capital, Koh offered blunt advice: “Twelve months ago, it was enough to have a wrapper on ChatGPT. That’s no longer true.”