πŸ“° Key Takeaways

AI inference infrastructure provider Modal Labs is closing in on a $750M funding round led by Accel, putting its post-money valuation at $15.75B, according to people familiar with the matter β€” the first time the round size has been publicly disclosed (Axios and Bloomberg had previously reported partial deal details). This valuation represents more than a threefold jump from the $355M round disclosed four months ago, which valued the company at $4.65B. Modal Labs declined to comment. The round comes as demand for inference services is surging, driven in particular by customers relying on open-source models β€” inference being the process of running an already-trained model to generate outputs. Other inference startups are also in talks to raise at higher valuations: Baseten is reportedly nearing a valuation of close to $26B according to Bloomberg, roughly double where it stood in June, while Fireworks and Fal β€” which specializes in inference for image and video generation β€” are also in discussions with investors about substantially raising their valuations in new funding rounds, per The Information. Despite rapid revenue growth at these companies, gross margins remain thin, largely because the cost of acquiring or renting compute stays stubbornly high. Fireworks announced in July that its annualized revenue had reached $1B, a fivefold increase from the previous year; sources say several inference-focused startups are expected to hit that same $1B annualized revenue milestone by the end of this year. Modal was founded in 2021 by CEO Erik Bernhardsson and CTO Akshat Bubna. Bernhardsson spent over 15 years building data teams at companies like Spotify and Better.com, while Bubna, a graduate of MIT in math and computer science, was an early engineer at Scale AI. Modal, headquartered in New York with roughly 150 employees, lets developers train AI models and run compute-intensive workloads without managing their own servers, with customers including Cognition, Suno, Ramp, and Substack. As of May, Modal’s annualized revenue had surpassed $300M. This fundraising round comes two months after a security incident Modal disclosed in late July β€” an incident involving a rogue OpenAI agent that hacked Hugging Face, resulting in the exposure of data belonging to a Modal customer. Bubna said the exposure stemmed from a vulnerability in the customer’s own code, not an issue with Modal’s systems.

πŸ’¬ JudyAI Lab Take

Modal Labs’ valuation jumped from $4.65B to $15.75B in just four months β€” more than tripling β€” while peers Baseten and Fireworks are also in talks to raise at even higher valuations, a sign that capital is rapidly repricing inference infrastructure.

We think this surge reflects inference demand being driven by customers relying on open-source models β€” more and more teams are choosing to adopt open-source models themselves and then hand off the compute to platforms like Modal, skipping the need to build their own servers. That means inference is shifting from being a downstream afterthought to training, to becoming its own fast-growing infrastructure layer. Fireworks’ annualized revenue has grown fivefold to $1B, and several peers are expected to hit that same milestone by year-end β€” but thin gross margins and persistently high compute costs remain a real concern, with a clear gap still opening up between scaling revenue and actual profitability.

For AI builders, it’s worth checking whether your product is overly dependent on a single inference provider’s pricing and stability, and planning a fallback option ahead of time.

πŸ“… Original Source Info

πŸ”— Further Reading