fal and AWS: Building for the Next Phase of Generative Media
By Emir Lise, Compute Partnerships @fal
Today fal announced a strategic partnership with Amazon Web Services (AWS).
I interned at AWS in college, working on compute pricing. What surprised me most was how many people were in the room for any given decision. Go-to-market, product, engineering, a lot of thoughtful input went into things before they shipped. When I was there, I didn’t anticipate how quickly AI and the compute demands around it would evolve into what they are today.
Getting to now sit on the other side of that table, across from some of the same people I once worked alongside at AWS, is a pretty good measure of how much the industry has grown.
Growing in the same direction
AWS didn't set out to become the infrastructure backbone for media, entertainment, retail, healthcare, and financial services all at once. It grew into that because the underlying need for reliable, elastic compute turned out to be universal.
fal has been on a similar arc. We started as a generative media platform built for developers, and the assumption was that the people reaching for image, video, audio, and 3D generation were builders, technical people standing up AI-powered tools. But generative media didn't stay neatly inside that box. Studios started using it, then retail brands, then enterprise software teams with very specific and operational needs at scale, and industries we hadn't originally targeted found us because fast and reliable inference across a growing catalog of models turned out to matter everywhere creative work happens. Today over 2.5 million developers build on fal, and companies like Amazon MGM Studios, Canva, and Adobe run production workloads on our platform every day across image, video, audio, and 3D.
That's what makes this partnership feel like a natural next step. We've grown into an ecosystem that touches many of the same industries AWS does, and as generative media continues to evolve, having AWS alongside us for that next phase feels complementary.
Reliability and flexibility at scale
fal's serverless infrastructure doesn't just power our customers' applications, it powers fal itself, every model, every inference call, every workload running on our platform. So the reliability bar here isn't theoretical.
Generative media workloads can be unpredictable by nature. A new model drops, a campaign goes live, a product ships, and suddenly you're absorbing a traffic spike that wasn't in anyone's forecast. You need a partner whose infrastructure handles that kind of moment without requiring you to negotiate for it, where capacity exists across regions and compute types precisely when you need it most.
That's what we were looking for, not just solid uptime on a normal day but genuine elasticity at the moments that actually test the system. AWS brings that, along with a global footprint that lets us serve enterprise customers wherever they happen to be building.
As Gorkem Yurtseven, our CTO and Co-founder, put it:
The inference problem for generative media is unlike anything else in AI infrastructure, the parallelism, the model variety, the latency requirements. We've spent years optimizing fal's engine for exactly this. AWS gives us the global scale and reliability layer that lets us take that work to its full potential, for every team running production workloads on fal.
Even in a space that has moved as fast as this one, it still feels like we're at the beginning. The workflows and solutions coming out of generative media keep getting more ambitious, and the infrastructure decisions that underpin all of it matter more than ever. My work is grounded in thinking about that long term, about the partnerships and foundations that will still make sense years from now.
For press inquiries, contact press@fal.ai
To start building on fal, visit fal.ai