Intel is preparing to launch a new GPU, the Arc Pro B70, with 32GB of VRAM, positioning itself in the market for graphics cards dedicated to artificial intelligence.

Technical Specifications

The GPU, on direct sale for $949 starting March 31, offers a bandwidth of 608 GB/s and a power consumption of 290W. These features make it potentially suitable for the local execution of AI models, such as Qwen 3.5 27B with 4-bit quantization.

Market Implications

The introduction of a GPU with these features at this price could make the development and inference of large language models (LLM) more accessible locally, reducing dependence on cloud solutions.

For those evaluating on-premise deployments, there are trade-offs to consider carefully. AI-RADAR offers analytical frameworks on /llm-onpremise to evaluate these aspects.