AI inference platform Fal completes a $1.3 billion Series F funding round at a $16 billion valuation
U.S. cloud inference platform Fal has completed a $1.3 billion Series F funding round at a post-money valuation of $16 billion, double the $8 billion valuation from its Series E in April this year. The round was led by Bessemer Venture Partners, with Nvidia continuing to participate; the funds will be used to procure more compute and expand its current team of 185 people. Fal provides hosting and inference services for image, video, and audio models to creators and developers, and its CEO said that this summer it had to limit customer request volumes due to a surge in demand.
Why read this
AI score 58
Fal's valuation doubles to $16 billion in half a year; the funding is earmarked for compute expansion, with capacity bottlenecks and customer structure at inference service providers worth watching.
Topics
Body · AI translation
Zhihui Dongxi, November 4: In the early hours of today, US cloud inference platform Fal announced that it has completed a $1.3 billion (approximately RMB 8.733 billion) Series F funding round, at a post-money valuation of $16 billion (approximately RMB 107.488 billion). The round was completed in October this year. In April this year, its Series E valuation was $8 billion (approximately RMB 53.744 billion), and the valuation has doubled after this round of funding.
According to Bloomberg, Fal plans to use the new funds to procure more computing power and expand its current team of 185 people. This summer, the company was once unable to meet customer demand and had to limit customer request volumes.
This round of funding was led by well-known US investment firm Bessemer Venture Partners, with Nvidia continuing to participate.
Fal was founded in 2021 and is headquartered in San Francisco. Its core business is providing image, video, and audio model hosting and inference services for creators and developers, helping customers run models and generate content.
Its two co-founders are Burkay Gur and Gorkem Yurtseven. CEO Gur previously led Coinbase's machine learning business, and co-founder and CTO Yurtseven was formerly an Amazon developer.
According to Bloomberg, citing people familiar with the matter, Fal's annualized revenue had already reached $800 million (approximately RMB 5.374 billion) earlier this year.
According to an investment announcement published by Bessemer, Fal provides developers with a unified model invocation API, provides distribution services for model developers, and provides inference infrastructure for enterprises deploying their own models.
According to disclosures by Bessemer, the lead investor in this round, companies such as Adobe, Canva, and Shopify have already used Fal's services in actual business operations. Fal's official website states that its customers cover entertainment studios, advertising agencies, e-commerce brands, game developers, and marketing teams, and that the relevant models have been integrated into customers' content production workflows.
Fal disclosed that its post-training partners include Chinese large-model unicorn MiniMax and German visual model developer Black Forest Labs. In August this year, Fal launched H3 Max, post-trained based on MiniMax H3, optimizing prompt adherence, visual quality, and generation speed; in October, the company partnered with Black Forest Labs to launch FLUX 3 Max, continuing to optimize video generation quality and inference efficiency.
Fal stated that both the training and inference of H3 Max use Nvidia GB200 NVL72 systems, and generating a 5-second video takes about 3 seconds.
According to Bloomberg, Gur said that surging customer demand is an important reason for Fal's large funding round. This summer, the company was once unable to meet customer demand and had to limit customer request volumes. Fal disclosed on its official website that these two funding rounds will support its business development through 2027 and provide funding for the team to carry out new research and explore content creation.
Conclusion: Demand exceeds capacity, and the inference platform supports expansion through financing
In July this year, AI inference service provider Fireworks completed a $1.505 billion (approximately RMB 10.111 billion) Series D funding round, with Nvidia also participating. The funds will be used to expand the engineering team and global computing power scale. Customer demand is growing, but taking on this business requires spending money first to procure computing power and expand teams. Both companies have chosen to support expansion through financing.
For Fal, financing must solve the immediate service capacity problem, while cooperation with MiniMax and Black Forest Labs focuses on improving model generation speed and quality. Advertising, e-commerce, and gaming customers have already integrated AI models into daily content production, and platform rate limiting, generation time, and material quality will all affect customers' choices.
Next, Fal needs to turn the newly added computing power and model optimization into more stable and efficient services, and strive for these customers to continue investing their model invocation budgets.
Source: Bloomberg, Fal, TechCrunch