The AI company behind Canva and Perplexity just hit $8 billion
Fal powers the AI features inside Canva, Perplexity, and Poe. Now it's reportedly raising $350M at an $8B valuation — with $400M in annual revenue.
Every time you generate an image in Canva, ask a question on Perplexity, or create content on Poe, there's a company you've never heard of making it all work. That company is Fal — and it just reportedly hit an $8 billion valuation.
According to The Information, Fal is in talks to raise $300–350 million in new funding. The company already generates roughly $400 million in annual revenue, making it one of the fastest-growing AI infrastructure companies most people have never heard of.
The invisible engine behind your favorite AI tools
Fal is what's called an AI inference platform — think of it as the factory floor where AI models actually run. When an app like Canva needs to generate an image, it doesn't run the AI model on its own servers. Instead, it sends the request to Fal, which runs it on powerful GPU hardware (the specialized chips AI needs) and sends the result back in milliseconds.
This behind-the-scenes approach means 1.5 million developers and hundreds of companies rely on Fal without most end users ever knowing it exists.
- Canva — AI image generation and design features
- Perplexity — generative media for search results
- Poe (by Quora) — powers 40% of its official image and video bots
- PlayAI — text-to-speech infrastructure
1,000 AI models under one roof
What makes Fal different from running your own AI servers is the sheer variety. The platform hosts over 1,000 production-ready models — not just one company's AI, but the best from everyone:
Video creation: Veo 3.1 (Google), Kling 3.0, Sora 2 (OpenAI), LTX-2
Voice & audio: Chatterbox TTS, MiniMax Speech, Lux-TTS voice cloning
Avatars: Aurora studio-quality video avatars, Omnihuman animation
For developers, this means one account, one API (a connection point for apps), and access to every major AI model — instead of signing up separately with OpenAI, Google, and a dozen other providers.
How much does it actually cost?
Fal uses pay-per-use pricing — you only pay for what you generate, with no monthly subscription required:
- Images: Starting at $0.02 per image (~50 images for $1)
- Videos: Starting at $0.05 per second of video
- GPU access: From $0.99/hour (A100) to $2.10/hour (H200)
For comparison, running your own H100 GPU hardware would cost tens of thousands of dollars upfront. Fal lets developers tap into that power for under $2 per hour.
From two ex-engineers to an $8 billion company
Fal was founded in 2021 by Burkay Gur (formerly at Coinbase) and Gorkem Yurtseven (formerly at Amazon). Based in San Francisco, the company built its platform to solve a specific problem: AI models are expensive and difficult to run at scale.
Their solution — globally distributed GPU clusters running NVIDIA's H100, H200, and next-generation B200 chips — lets any developer deploy AI without managing hardware. The result: apps can scale from zero to thousands of AI requests per second instantly.
Why the $8 billion matters beyond Fal
This valuation signals a major shift in where AI money is flowing. While consumer-facing companies like OpenAI and Anthropic grab headlines, the infrastructure layer — the companies that actually run the AI models — is quietly becoming just as valuable.
If Fal closes this round, it would join a small club of AI infrastructure companies valued above $5 billion, alongside CoreWeave and Lambda. For anyone building AI features into their products, Fal's growth confirms that "AI-as-a-service" is becoming the default — just as cloud computing replaced private data centers a decade ago.
Related Content — Get Started with Easy Claude Code | Free Learning Guides | More AI News
Stay updated on AI news
Simple explanations of the latest AI developments