ZML, a French AI startup backed by Turing Award winner Yann LeCun, has launched a free product to enhance inference performance across various AI chips. The new software, called ZML/LLMD, allows open-source large language models to run on Nvidia, AMD, Google TPU, Apple Metal, and Intel Arc chips. The company aims to break existing silos and enable AI use cases to operate at maximum speed on different hardware platforms, according to founder Steeve Morin.
The release of ZML/LLMD comes as AI integration into work and daily life continues to grow, with inference optimization becoming more critical than model training. Morin noted that while inference has become more important, it often remains behind the scenes with software and architecture barriers leading to vendor lock-in. The ability to achieve peak performance across a variety of chips is seen as a significant technological challenge, but it could also disrupt the market amid rising concerns over AI costs.
Morin emphasized that ZML’s goal is to provide enterprises and cloud providers with the flexibility to use a mix of chips, potentially reducing costs and energy consumption. He highlighted the importance of collaboration with European AI chipmakers like Axelera and Kalray, as well as the startup’s strong relationship with Nvidia, despite competition in the inference space. Morin also credited ZML’s small, well-funded team as key to its rapid development and future plans for more releases.
Source: techcrunch
