Tenstorrent QuietBox 2 Brings RISC-V AI Inference to the Desktop
A PCIe card with a 96-core RISC-V CPU and 16GB RAM for on-device AI, no cloud required.
Tenstorrent, the AI chip company led by industry veteran Jim Keller, has launched the QuietBox 2, a PCIe accelerator card that aims to democratize powerful, local AI inference. Unlike typical GPUs, the card is built around a 96-core RISC-V CPU (the Tenstorrent Ascalon) and is paired with 16GB of fast LPDDR5 memory. Designed to fit into a standard desktop PC, it operates on a modest 75W of power, drawing directly from the PCIe slot without needing extra cables. This makes it a plug-and-play solution for adding substantial on-device AI compute.
The card is specifically engineered for running inference on large language models and other generative AI workloads. By processing data locally, it addresses growing concerns about data privacy, security, and the recurring costs of cloud API services. Developers can use it to build and test AI applications that require low latency or must operate in disconnected environments. Tenstorrent is supporting the platform with its open-source software stack, including kernel drivers and compiler tools, to foster an ecosystem around RISC-V for AI.
The launch represents a significant push to establish the open RISC-V architecture as a viable alternative to proprietary GPU and AI accelerator designs from companies like Nvidia and AMD. For the tech industry, it offers a tangible path toward hardware diversity and vendor independence in the AI space. For professionals, it provides a practical tool for prototyping edge AI solutions, conducting sensitive research, or simply avoiding cloud egress fees and potential downtime.
- Hardware: A PCIe card with a 96-core Tenstorrent Ascalon RISC-V CPU and 16GB of LPDDR5 RAM, powered solely through the PCIe slot (75W).
- Use Case: Enables local inference for LLMs and AI models, providing data privacy, low latency, and independence from cloud services and their associated costs.
- Ecosystem Play: Part of Tenstorrent's strategy to build an open software and hardware ecosystem around RISC-V for AI, challenging incumbent accelerator architectures.
Why It Matters
Offers professionals a practical, private, and cost-effective path to run powerful AI models locally, reducing reliance on cloud APIs.