Our Investment in Fireworks AI: the Inference Platform Aiming to Power Every GenAI Application

AI is poised to be even more significant than the software, internet, and mobile evolutions. Yet, beneath the daily AI headlines lies a critical infrastructure challenge: how do companies actually deploy AI at scale?

We believe Fireworks is the platform answering this challenge. We are thrilled to lead Fireworks’ Series C alongside Index and Evantic, joining Sequoia and Benchmark to support Fireworks in its mission to build artificial autonomous intelligence.

The Fireworks Mission

Fireworks is the inference and infrastructure platform that provides developers with the tooling needed to take AI into production. It provides access to state-of-the-art open-source models across text, image, and audio, gives developers the tooling to customize these models for any case, and handles all the complexity in procuring compute. Fireworks can scale AI workloads at sub-second inference speeds, with extreme reliability and optimal cost.

As companies look to scale their AI offerings, the Fireworks team saw that they want control over their entire stack — not dependency on someone else’s API. Fireworks enables companies to ‘own’ their AI stack — meaning customizing models with their own data, controlling their costs, and avoiding vendor lock-in. For Fireworks, ‘artificial autonomous intelligence’ is the North Star, where automated product and model co-design reach maximum quality, speed, and cost-efficiency using generative AI.

Before investing, we spoke with numerous startups and large enterprises about their AI architectures. We consistently heard these companies were turning to Fireworks, which was giving them the ability to own their AI development. We heard Fireworks made it ‘dead simple’ to run AI applications, managing the low-level architecture to optimize cost, quality, and latency while giving developers the tools to customize their product experience to their user data. Moreover, Fireworks was productizing the techniques and capabilities used by frontier labs and putting these tools in the hands of every developer.

From Zero to 10 Trillion Tokens Daily

We believe Fireworks has become the inference platform of choice for companies that are serious about AI in production.

When customers needed to serve millions of users at scale, they told us Fireworks was the vendor they trusted as the backbone for their entire application, achieving performance levels that simply weren’t possible with other solutions. The numbers tell the story:

• Fireworks serves over 10,000 customers
• The company processes more than 10 trillion tokens daily
• Up to 40x faster inference than alternatives
• 8x cost reduction compared to other platforms
• Powering mission-critical AI features for companies like Cursor, Uber, Notion, Samsung, and Verizon

Our Partnership

A big driver of our investment was our deep belief in Lin Qiao and her founding team — Benny Chen, Chenyu Zhao, Dmytro Dzhulgakov, Dmytro Ivchenko, James Reed, and Pawel Garbacki. As the group responsible for PyTorch at Meta, Fireworks has assembled one of the highest-quality technical teams we have met. They bring a rare combination of technical depth and production battle scars to Fireworks.

Over the last three years, we believe the Fireworks team has stayed at the forefront of the industry, making the right bets and maintaining a maniacal focus on delivering extreme simplicity to their users.

Fireworks is democratizing AI — giving every developer the tools to scale AI applications. We are honored to partner with them in their journey to empower this generation of builders, startups, and enterprises driving AI innovation across every industry.

The content here does not constitute an offer to sell or a solicitation of an offer to buy any securities or investment advisory services.

Certain information contained herein is based on information from various sources prepared by third parties. While such sources are believed by Lightspeed to be reliable, neither Lightspeed nor its affiliates assume any responsibility for the accuracy or completeness of such information, and such information has not been independently verified by Lightspeed. For more details please see https://www.lsvp.com/legal.