Fireworks AI and Fal Weigh New Funding Rounds as Inference Demand Surges

Fireworks AI and Fal are both considering raising new funding rounds, The Information reports, as demand for AI inference compute continues to outstrip supply.

The two startups operate in adjacent corners of the AI infrastructure market. Fireworks AI provides an inference platform that lets companies run large language models and other AI models at scale. Fal specializes in inference for generative media, powering image and video generation for developers and creative tools.

The reported deliberations come as inference — the stage of AI computing where trained models actually generate outputs — has become the fastest-growing cost center in enterprise AI adoption. Startups that own inference infrastructure have attracted outsized investor attention over the past year, as application-layer companies shift spending from training compute toward serving models to end users.

Neither Fireworks nor Fal has disclosed target valuations, round sizes, or lead investors, and The Information's report does not state whether either company has mandates out with investors or is still in internal discussion. Both companies have raised capital previously, and a new round for either would extend a run of financings across the inference layer.

For Fireworks, a fresh raise would deepen its position in the crowded market for LLM serving, where it competes with hyperscalers and well-funded rivals such as Together AI and Groq. For Fal, additional capital would fund capacity expansion in generative media inference, a segment that has seen usage climb as image and video generation moves into mainstream consumer products.

The timing signals where investors see the next phase of AI value accruing: not in building models, but in running them cheaply and at scale. If either round closes in the coming months, expect inference-focused infrastructure to remain among the most competitive fundraising categories in the sector.