- Founders: Burkay Gur & Gorkem Yurtseven
- Sector: AI Infrastructure
- Location: San Francisco, CA
The Opportunity
With the advancement of image, video, audio, and 3D models over the past year, the AI industry is rapidly moving from text-based outputs to more complex, longer-form multi-modal content generation. AI-based multimedia is increasingly ubiquitous because it compresses months of planning and production into high-quality content generation at a fraction of the traditional time and cost. While this industry is still in its early innings, each model release seems to materially push the frontier forward. In a few short years, we’ve moved from static low-resolution images to high fidelity, complex videos and audio powering critical business cases across every industry.
We envision a not-too-distant future where AI will be the default vehicle for multi-media content creation. However, AI generated multi-media is still computationally intensive, requiring massive GPU clusters, specialized infrastructure, and complex optimization techniques. For the industry to reach its full potential, users require greater maneuverability of the models, creating the nuances and details indicative of high-quality production across different industries (i.e., retail, e-commerce, etc.).
In short, creators need a tool for AI-based multimedia generation that doesn’t require them to trade off between performance, cost, and model choice.
The Solution
fal provides the generative media platform for developers. Rather than managing hardware, building inference pipelines, and setting up dedicated software environments, fal abstracts all the infrastructure complexity behind simple APIs. fal’s inference engine automatically and dynamically applies advanced compiler and quantization techniques to optimize models, and provides developers with unbeatable reliability, performance, and cost.
Generative media comes in many different flavors — image-to-image, text-to-image, text-to-video, image-to-video, video-to-video, text-to-speech, speech-to-text, and everything in-between. And each flavor has dozens of models that are available for consumption. For developers, having one platform where all the relevant models are assembled for easy access, and where they can build workflows involving multiple different models easily, can be a huge unlock.
fal works closely with leading model providers like Google, Black Forest Labs, and ElevenLabs to offer the most cutting-edge, capable models on the market. Through these partnerships, users can access hundreds of different models for endless use cases, ranging from lip-syncing and music generation to end-to-end video creation and beyond. fal also actively supports the open-source community through initiatives like the fal Research Grants program, providing free compute resources to developers working on open-source projects. This commitment not only fosters innovation but also positions fal as a collaborative leader in the AI ecosystem.
Why We’re Backing fal
We first met fal co-founders Burkay Gur and Gorkem Yurtseven in early 2024, and over the past 18 months we’ve been consistently impressed by their progress tackling one of AI’s most complex infrastructure challenges. Scaling reliable, high-performance inference isn’t easy — but Burkay’s ML chops honed from his time at Coinbase and Gorkem’s tooling background from his experience at AWS bring the technical depth required to get the job done. Their team has built a delightful experience, allowing them to amass a vibrant community and developer flywheel.
In the coming years, as more enterprises turn to AI for complex, multi-modal use cases — and as models grow more powerful — fal’s platform is primed to play a leading role. Since publicly launching last year, fal has already become the central source for generative media, serving millions of developers and billions of inference requests with 99.99%+ uptime. Leading companies like Canva, Adobe, and Perplexity entrust fal for some of their most mission-critical features. By serving the most popular and best models on one platform, fal is also creating unique network effects: over 1M developers use fal’s APIs, benefitting from fal’s access and continuous improvements, which in turn lead to more model releases on the platform.
Further, the market timing for fal couldn’t be better. We’re witnessing a generational shift in how content is created, with AI-generated media moving from experimental to essential. Marketing teams are using AI for personalized campaigns, entertainment companies are streamlining production workflows, and e-commerce platforms are generating product videos at scale. fal’s infrastructure makes these transformations possible by removing the technical barriers that previously limited AI adoption to only the most technically sophisticated organizations.
What’s Ahead
Salesforce Ventures is thrilled to participate in fal’s $125M Series C alongside Meritech Capital, a16z, Notable Capital, and Bessemer Venture Partners, among others. We believe that fal is uniquely positioned to power the next generation of creative tools, and are excited to back their mission of enabling generative media for every company.
The generative media journey is just beginning, and there is so much work still to do. The fal team would agree — since the deal closed, they have already brought on multiple SOTA models with fast and optimized inference as well as LoRA training support. fal continues to innovate on both the product and infrastructure of its platform at a breakneck pace. We look forward to having a front-row seat for this exciting journey.
Welcome to Salesforce Ventures, fal!
