Fireworks Raises $1.5B As AI Platform Reaches $17.5B Valuation
Fireworks AI raised a $1.5 billion Series D at a $17.5 billion valuation, SiliconANGLE reported. The company says it processes more than 40 trillion tokens a day, while contract values, total GPU capacity and customer commitments tied to the new capital remain outside the public record.
Fireworks AI has raised $1.5 billion in a Series D round that values the AI infrastructure startup at $17.5 billion, according to SiliconANGLE.
The funding gives the company new capital for managed graphics-card clusters, fine-tuning tools and inference services used by developers running open-source AI models.
Series D Values The AI Infrastructure Startup At $17.5 Billion
Atreides Management, Index Ventures and TCV jointly led the Series D round.
The Wednesday funding announcement listed more than a half-dozen other backers in the financing, including Nvidia.
SiliconANGLE attributed the valuation partly to rapid sales growth.
The outlet reported that annualised revenue recently passed the $1 billion mark, and Fireworks put platform processing at more than 40 trillion tokens per day for users.
Samsung Electronics and GitLab were listed among the company's customers.
Fireworks did not tie the new valuation to named contract values, customer spending commitments or a disclosed order backlog.
Developers Use Fireworks For Fine-Tuning And Inference
The company operates a cloud platform for fine-tuning open-source AI models.
The service gives developers access to managed graphics-card clusters under usage-based pricing and includes an AI agent that can automate parts of the training workflow.
The agent can use a developer's task description and uploaded training dataset to choose technical settings for a model.
The system searches for hyperparameter combinations that maximise output quality and can extend a dataset with DPO files when needed.
The platform also supports four parallelisation techniques for splitting training calculations across multiple chips.
The four methods can run side by side or individually, depending on the model type.
Dedicated GPU Clusters Sit Behind Fireworks Deployments
After fine-tuning, customers can host models through two inference services.
The serverless option removes the need to configure underlying infrastructure, while the Deployments service gives customers dedicated graphics-card clusters.
The company says Deployments provides better performance than the serverless service.
The same interface lets developers adjust autoscaling and compress models through quantisation to reduce infrastructure requirements.
Co-founder and Chief Executive Officer Lin Qiao said companies hold knowledge in their data, workflows, customers and quality standards.
The proceeds will be used to expand infrastructure and hire more engineers.
The public record does not show customer commitments tied to the new capital.
