fal Raises $1.65B Across Series E and F for Media AI

An advertising deadline gives an AI video model very little credit for being impressive yesterday. For fal, the San Francisco company supplying infrastructure for generative media, the commercial assignment keeps arriving with the next image, clip or audio request. The company has disclosed $1.65B across Series E and Series F to finance more of that work.

Sequoia Capital led the Series E, which closed earlier in 2026. Bessemer Venture Partners leads the Series F, with continued investment from NVIDIA. Announced together on November 3, the financings support inference capacity, the team and research into 2027, putting substantial capital behind the infrastructure that carries generative media into customers' production workflows.

What the two financing rounds actually cover

fal's official announcement presents the rounds together, and its accompanying artwork states the combined $1.65B amount. The Series E belongs to an earlier financing moment; the disclosure brings it alongside the Series F. That distinction matters when assessing how the company has financed its expansion over the year.

The company identifies entertainment studios, advertising agencies, e-commerce brands, game developers and marketing teams among the users putting generative models into production. It directs the proceeds toward the capacity to serve those workloads and the people developing its platform. The announced ambition extends into 2027, without attaching a specific delivery date to additional capacity or a new product.

Bessemer's Series F investment statement confirms its leadership of the latest round. The firm's explanation follows fal's expansion from model serving into a broader collection of tools and services. It is the perspective of an investor increasing its commitment to the business, and it helps explain which parts of that business the new capital is backing.

The financing therefore reaches into an ongoing supplier relationship. A customer integrating generation into a service needs the underlying platform to remain useful as the models and creative requirements change. fal is raising money while accepting more responsibility for that continuing relationship.

Why creative production changes the infrastructure job

Co-founders Burkay Gur, CEO, and Gorkem Yurtseven, CTO, started fal in 2021, according to the company's account of its origins. Their earlier work at Coinbase and Amazon placed them close to the infrastructure problems behind software products. fal now applies that experience to workloads where a customer's judgment about an output can send the system back to work immediately.

Inference is the execution of an already trained model to produce an answer or asset. With generative media, the result might be a product image or a sequence of video frames, and the user may want further variations before accepting it. That makes the relationship between speed, quality and cost especially tangible for someone assembling a campaign or building a creative application.

fal's founders have chosen to sell access to that machinery through developer services. Their customer can concentrate on the experience being created while fal handles the model-serving work underneath. As more of a customer's workflow passes through those services, a delay or awkward integration becomes part of the customer's own production problem.

That is the operating consequence behind the funding. Supporting a creative experiment is one assignment; remaining useful through its revisions, deployment and continuing use gives the infrastructure supplier a longer job. The capital helps fund the people and computing resources attached to that job.

The buyer's economics follow the usable result

fal's pricing page separates model API charges from custom GPU compute. Video endpoints can charge by output duration or by video, depending on the model. The listed unit gives a buyer a starting point, while the actual workflow determines how many units the customer needs.

For a team evaluating generated media, an acceptable asset is a more informative commercial unit than an isolated request. Several attempts may be needed before a result fits the brief, and a model that performs well on one task can produce a different experience on another. Those are evaluation questions for the buyer's own workload, rather than outcomes established by the financing announcement.

This gives model infrastructure a demanding sales conversation. Developers can inspect an endpoint and its price quickly; understanding its place in a production process takes actual use. A creative organization also has to decide how generation fits alongside its existing approval, editing and delivery work. An attractive demonstration opens that conversation without finishing it.

fal participates in those decisions through both its hosted services and enterprise offering. The opportunity is to stay valuable as customers move between experimentation and recurring production. The customer's spending then follows a process with human judgments inside it, even when the generation itself happens automatically.

Investors are following fal further into the workflow

Bessemer's involvement predates this round: its February 2025 announcement documented participation in fal's $49M Series B. Sequoia announced its Series D leadership in December 2025. The new E and F disclosure extends a financing history tied to a developing infrastructure business.

The company has also described improving existing models through post-training, adding another area of work alongside serving them. That creates a closer connection between the organization supplying a model and the platform delivering it to a developer. It gives fal more places to contribute to the final experience and more engineering decisions to carry through to production.

The practical demands extend to hiring. fal's careers page describes a San Francisco headquarters and a team spread internationally, with recruitment spanning the work of building the company. The financing statement explicitly includes the team alongside inference capacity, treating both as resources needed for the next period of development.

Burkay Gur and Gorkem Yurtseven now have the backing to keep extending that assignment. Their customers will continue to arrive with changing briefs, new models to evaluate and work already due. The next stage of fal's business will be built in those repeated encounters, where creative ambition keeps asking the infrastructure for a little more room.