Fal Announces $1.65 Billion in Combined Series E and F Funding

Generative-media infrastructure company Fal announced on November 3, 2026 that it had raised a combined $1.65 billion across its Series E and Series F rounds. Sequoia Capital led the Series E, while Bessemer Venture Partners led the Series F with continued investment from NVIDIA.

The total combines two rounds rather than representing a single Series F raise. An accompanying company post says the Sequoia-led round closed earlier in 2026 and was announced alongside the Series F. The post also identifies NVIDIA as a strategic investor in the later round. Neither the announcement nor the post breaks out the size of either round separately.

Fal plans to use the funding for work across its ecosystem in 2027, pairing inference capacity with investment in its team and room for new research and forms of creation. That split reflects the workload the company describes: entertainment studios, advertising agencies, e-commerce brands, game developers and marketing teams are running image, video and audio models inside pipelines that deliver final work, rather than limiting their use to experiments.

Capacity and Model Turnover Shape the Funding Plan

Fal’s current product overview divides the platform into hosted Model APIs, Serverless Inference for custom models, and Compute for training and other GPU-intensive workloads. The company advertises access to more than 1,500 generative-media models through one API, autoscaling across thousands of GPUs, and interconnected clusters of H100, H200 and B200 GPUs. Those offerings place Fal between model developers and the businesses turning generative-media outputs into production applications.

A strategic AWS partnership announced on May 19, 2026 shows why elastic capacity matters. At the time, Fal reported more than 2.5 million developers building on its platform and daily production workloads from Amazon MGM Studios, Canva and Adobe. Model releases, live campaigns and product launches can create traffic spikes that were not forecast, requiring capacity across regions and compute types. Fal’s serverless infrastructure also runs every model, inference call and workload on its own platform, so the reliability requirement applies internally as well as to customers.

The engineering problem extends beyond adding GPUs. In a Sequoia Capital interview, Fal’s founders and head of engineering described generative video as compute-bound, requiring specialized kernels and careful scheduling across a distributed, heterogeneous GPU fleet. Traffic must be routed to GPUs with warm model caches while hardware choices and customer demand keep changing. The discussion also provided a measure of that complexity: Fal’s top 100 customers were using 14 different models at the same time, sometimes chaining text-to-image, upscaling and image-to-video models in one workflow. Model breadth therefore creates an orchestration burden alongside the raw compute requirement.

Prior Funding Sets Up the 2027 Buildout

The financing disclosure follows Fal’s $140 million Series D, announced on December 9, 2025, which added Sequoia, Kleiner Perkins and NVIDIA as new investors. That round was intended to scale the platform globally. Fal had 70 employees at the time and was hiring across engineering, product, design, go-to-market and operations, providing a historical baseline for the team investment now planned.

The funding plan ties Fal’s 2027 execution to two connected demands: sufficient elastic capacity for unpredictable production workloads and continued research as customers change models and workflows. The central test is whether investment in infrastructure and the team can keep those demands aligned as the model catalog and customer traffic evolve.