As AI industry booms, Lin Qiao, CEO of $4 billion startup processing 15 trillion AI tokens a day, calls 2026 the year of ...

The global AI industry is increasing and very few people embody that momentum more than Lin Qiao, CEO of Fireworks AI, a $4 billion startup now processing 15 trillion AI tokens per day. As reported by Business Insider, a former Meta engineer who helped build PyTorch, Qiao has seen firsthand how quickly AI can move from experimental frameworks to mainstream adoption. “This is the year token consumption is going to grow exponentially,” she said in a recent interview. Fireworks AI’s inference cloud platform has now scaled rapidly from 10 trillion tokens in late 2025 to 13 trillion earlier this year, and now 15 trillion daily. Tokens, the basic units AI models use to process language, are not just powering tech teams. Qiao also points to finance departments automating forecasts, legal teams building internal AI tools, gig workers generating music demand, and sounds using multiple AI systems simultaneously. “Literally every single person is using these tools,” she noted.The surge in demand is rippling across the technology stack. GPU supply is tight, prices are rising, and even power grids are under strain as companies race to expand AI capacity. Qiao described the system as “saturated,” with bottlenecks stretching from semiconductor components to energy infrastructure.With hyperscalers like Amazon, Google, Microsoft, and Oracle already renting out GPUs, some question why startups like Fireworks AI are necessary. Qiao’s answer: complexity and speed. Enterprises struggle to keep pace with rapidly evolving hardware and models. Fireworks AI manages that churn optimizing performance, migrating workloads, and ensuring customers can adapt quickly.Qiao’s credibility stems from her role in building PyTorch, which democratized AI development and spread adoption beyond Silicon Valley into industries like agriculture and manufacturing. She sees a similar wave unfolding now, but at a much faster pace. “Once AI becomes usable, adoption accelerates dramatically,” she said.