Open-weight AI companies are the Valley’s hottest acquisition targets
There's a lot of capital pouring into the business of giving models away.
Rumors suggest that Nvidia is on the verge of making a $13 billion deal to acquire Hugging Face, a platform for sharing open weight AI models and benchmarks. Hugging Face has emerged as a key player in the ecosystem of developers building and deploying large language models (LLMs) that aren't owned by major AI labs. This acquisition would enable Nvidia to reduce its dependence on its agreements with hyperscalers and frontier labs, especially with other AI model builders like OpenAI and Google developing their own inference chips.
While Nvidia already has its own Nemotron family of open-weight models, they haven't gained significant traction. By acquiring Hugging Face, Nvidia will have access to a large user base that can be directed towards its chips and standards. Open-weight models have seen limited adoption, with only 6% of companies using them, according to a Ramp survey, or just 2% of software engineers surveyed by Jellyfish.
Open-weight models are primarily used by businesses that rely on repeated inference workloads, such as customer service chatbots. They can be more cost-effective for these high-volume, repetitive tasks. Stripe acquired OpenRouter, the top provider of open-weight models to businesses, for over $7 billion, signaling confidence in the potential of open models for cost savings in compute resources.
However, frontier models still have an edge in coding and agentic tasks due to easier access and token subsidies provided by the proprietary labs.
The trend towards open-weight models is driven by a desire for control, configurability, and the need to reduce reliance on expensive models from leading AI companies. As AI workflows become more mature, companies may find it increasingly attractive to invest in self-hosting models to meet their specific needs. The AI sector is still in its early stages, and the dominance of OpenAI and Anthropic is not guaranteed.
As the tech giants explore the potential of open technology, it seems that the allure of open systems is proving hard to resist.
Written by urgent.news from TechCrunch's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
