Key Takeaways
- Animoca Brands launched a $10M fund targeting early-stage agentic AI, calling the sector as early as OpenAI circa 2020.
- Yat Siu says the $900B advertising industry will shift to AI agents, routing discovery value to blockchain networks.
- Siu projects 1,000x productivity gains and advises retail investors to build wide baskets across 50 to 70 AI companies.
Yat Siu Says AI Productivity Gains Are 1,000x, Tells Retail Investors to Build Wide Baskets Now
Speaking in Miami on The David Lin Report with David Lin and Bonnie Blockchain, Siu described the current state of agentic AI as comparable to seed-stage investing in OpenAI or Anthropic roughly five years ago. He told investors to act accordingly.
“The productivity growth in AI is literally a 1,000x,” Siu said. “So we think the return profile is going to be a thousand-x amongst sort of the top companies.”
A $10M Bet on the Next Web
Animoca Brands, which holds more than 620 portfolio companies across gaming, decentralized finance ( DeFi), real-world assets, and non-fungible tokens ( NFTs), announced the fund to target early-stage agentic AI infrastructure. Siu said the firm treats the current moment the way it treated gaming and NFTs in 2021.
“When you’re early to the game then you have an advantage,” Siu said, “and that’s the same thing that we saw with gaming and with NFTs.”
He pointed to December 2024 as the approximate start of social agentic AI as a real movement, noting that most of the public is still thinking in ChatGPT mode: asking questions rather than delegating full tasks.
Agents as the New Internet Layer
Siu frames agentic AI as the infrastructure of what he calls Web 4. Blockchain, he argued, gives agents the ability to hold wallets, run microtransactions, and establish digital identity without relying on centralized platforms like Facebook or Apple.
“Advertising is an almost $900 billion industry a year, which is all about discovery, and it will completely be flipped upside down,” Siu said. “All of that value is going to go to the agents who are going to find stuff for you.”
He said companies pitching to consumers in the next two to three years will be building MCP servers and CLI tools designed to reach agents, not humans.
Ghost GDP and New Economies
On the question of so-called ghost GDP, where AI agents transact with each other without human enrichment, Siu pushed back on the pessimistic read.
“Entire new economies come where humans either create agents or themselves want to go and say, ‘Hey, look at this, I want to offer you something here,’” Siu said. He compared the dynamic to Adam Smith’s invisible hand: the activity looks abstract until the downstream value materializes.
He said AI agents owning assets creates a new class of buyers, and humans who can produce things those agents value will benefit directly.
Every Worker Becomes a Manager
On labor, Siu said the most valuable skill going forward is agent orchestration, not task execution.
“You’re hired because you can manage 10 agents well or 20 agents,” he said. “The person who can manage 100 agents or 200 agents will be more valuable than the person who can manage only one.”
He projected GDP expansion, not contraction, drawing a parallel to the personal computing era. The same argument, he said, applies to music, programming, and media: the barrier to creation drops, output proliferates, and the people with the clearest vision rise.
For retail investors, Siu said broad basket exposure to early-stage agentic AI companies is the most practical path, even if it means holding 50 to 70 positions. One Anthropic-scale outcome in that group, he said, justifies the approach.
