Fal raises $125M Series C and crosses $100M ARR on AI-generated media platform

Fal, a generative media platform providing API access to image, video, and audio models, raised $125 million in Series C funding and crossed $100 million ARR this month. Co-founder Gorkem Yurtseven announced both milestones, with the ARR figure arriving ten days before month-end.

Customers include Shopify, Canva, Adobe, and Lvel.io. Growth comes from both directions: large enterprises signing direct deals and indie developers adopting through self-serve, who then bring Fal into larger companies as they advance their careers.

“Gorkem Yurtseven: 'Two weeks ago, we announced our Series C — raised $125 million.' On revenue: 'We crossed $100 million run rate — we still have 10 more days but we crossed it. It's been an incredible summer for us and the whole industry.' On the business model: 'We actually make more money on cost-effective models, people who want to use it for higher volume but the model is good enough. That's a very interesting sweet spot.' Enterprise customers include Shopify, Canva, and Adobe.”

Fal makes more money on cost-effective, high-volume models than on frontier ones, a different economics from LLM inference providers where consolidation around flagship models compresses margin. The fragmentation of image and video use cases, and the model-chaining behavior that fragmentation drives, is what gives Fal its advantage.

Fine-tuning matters more for image models than most realize. Companies can fine-tune on a specific product and generate better images of it in ways that don't translate cleanly to LLMs. The same dynamic is emerging in video as open-source, fine-tunable video models mature, with companies already experimenting on camera angles, styles, and product or person-specific outputs.

Alibaba's video model, released a few weeks ago, is currently the best open-source option available. Yurtseven stops short of calling it a DeepSeek moment for media generation, arguing the broader inflection—where image and video generation becomes mainstream rather than an add-on feature inside Canva or Adobe—hasn't happened yet.

Fal's structural opportunity is the gap between raw open-source model capability and production-ready enterprise deployment. Image and video models haven't generalized the way LLMs have. You cannot write a long prompt and expect the model to handle complex edits end to end. Workflows still require chaining multiple models together, and that orchestration layer is where Fal operates.