Enterprise AI Infrastructure Plans Revealed by Global Survey

According to a survey conducted by FuriosaAI, ClearML, and the AI Infrastructure Alliance of 1,000 companies across North America, Europe, and Asia, 96% of respondents said they plan to expand their AI computing infrastructure in on-premise or cloud environments in 2024.

While many companies are rushing to adopt Generative AI, concerns were also raised that moving too fast could lead to mistakes such as prioritizing the wrong use cases. This suggests that while many tech companies are increasing infrastructure capacity, they are still at an early stage of establishing concrete execution plans.

The importance of open source tools and frameworks was also highlighted. 95% of respondents said that using open source tools is important, with PyTorch (61%), TensorFlow (43%), and Jax (16%) showing the highest preference among frameworks, in that order.

Of particular note is the demand for infrastructure for Inference. 52% of respondents said they are looking for cost-efficient alternatives to replace GPUs, a higher proportion than in the Training area (27%).

In response to these market needs, FuriosaAI offers the Gen 1 Vision NPU for computer vision acceleration, and is preparing to launch a 2nd generation accelerator for LLM and multimodal inference.