Personalized AI startup River AI raises $1.1B from consortium backed by Nvidia, AMD
SiliconANGLE Maria Deutscher ● Covered by 3 sources
River AI just raised $1.1 billion to help companies customize open AI models. The twist: it’s backed by Nvidia, AMD Ventures, Y Combinator and Temasek.
Based on reporting by SiliconANGLE, Maria Deutscher — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
River AI has pulled in $1.1 billion in early funding, split across a seed round and a Series A. That is a massive war chest for a startup that, on paper, does one very specific thing: helping enterprises adapt open-source AI models to their own needs.
The money came from a consortium led by General Catalyst and AMP PBC. Nvidia Corp., AMD Ventures, Y Combinator and Temasek all joined in. River AI’s chief executive is Igor Babuschkin, who previously co-founded xAI and worked at DeepMind, where he helped develop AlphaCode, the coding system that showed competitive results in a programming contest.
The company’s first product is River API, a cloud service for customizing open-source large language models with extra training. River AI says it supports models with 35 billion to 1 trillion parameters, and that customization uses LoRA, short for low-rank adaptation. Instead of retraining a model from scratch, the service adds a small number of neurons and trains those, which is the cheaper path. River AI says users can tailor a model in 15 to 20 minutes and that the result can be up to four times more cost-efficient than proprietary alternatives.
The product is also doing a lot of the annoying setup work for customers, including infrastructure configuration for training. And River API is only the start. River AI says it is building a broader suite that will add “personalization and continual learning for agents.” In a July blog post, Babuschkin said the long-term aim is personal AI systems that adapt to user preferences and stay under the user’s control, rather than being rented.
There is also hardware in the plan. A job posting points to a custom system-on-chip with an onboard machine learning accelerator, built on “advanced foundry nodes.” River AI also plans a compiler that can take customer LLMs built with PyTorch and turn them into a format that runs efficiently on its silicon. That’s an ambitious stack for a company that has not even finished its first act.
My take — AI-written commentary, not fact-checked reporting
This is the kind of funding round that tells you where the money is, not where the product is. River AI is betting that open models plus cheaper fine-tuning will beat the rented-black-box crowd, and that is a sane bet; the part that smells a bit is the hardware moonshot tacked on before the software story has even settled. Silicon Valley does love a startup that wants to be a platform, a cloud service, and a chip company before lunch.
Read more about this at: SiliconANGLE