The Open-Source AI Reality | How Token Costs Will Fall 10X & Usage Will Explode 100X | Lin Qiao

By 20VC

Categories: VC, Startup

Summary

Fireworks founder predicts a 10x reduction in token costs over three years will trigger a 100x explosion in AI usage, driven by specialized inference optimization for private enterprise data—not generalized models. The real AI frontier isn't training; it's activating proprietary data locked inside companies.

Key Takeaways

  1. Token costs will drop 10x in 3 years, driving 100x usage growth. This economic shift unlocks the inference layer as a valuable business segment rather than a commodity, creating defensibility through specialized intelligence optimization.
  2. Majority of enterprise data (80%+ of world data) is private, proprietary, and never used for training general models. Specialized AI companies win by activating this locked data through custom models, not competing on generalized APIs.
  3. Positioning between chip providers (Nvidia) and model providers (OpenAI) is defensible. The inference optimization layer captures value by solving for latency, cost, and specialized use-cases that generalized APIs can't optimize for.
  4. Late-stage founder advantage: Lin waited until 48 to start, spending 15 years learning systems at LinkedIn and Facebook. Deep technical expertise in distributed systems and database architecture proved more valuable than early-stage execution.
  5. Specialize vs. generalize depends on business diversity. AGI premise assumes one model solves everything; reality shows different regions, policies, and workflows require specialized solutions—validation that the market demands customization.

Related topics

Transcript Excerpt

What I don't want to see is there's only one company owns intelligence. That doesn't make sense to me. I think last year is the year of coding and this year is the year of co-work. And in the hot seat today, a founder who I wrote a $10 million check for after just a 15-minute meeting. Lynn Quo, founder at Fireworks. This was one of the easiest investment decisions that I've made in a 10-year investing career. >> I do think the cost of token will go down drastically. 10x cost reduction in the next three years and this 10x cost reduction will drive a 100x usage. We absolutely are not going to move into application layer. very clear to us whether we will move down into data centers and so on that could be always be on the table but the question is >> ready to go Lynn I am so excited for this …

More from 20VC