The Most Efficient AI Computer, Built in Software

We build AI infrastructure software for frontier models, hyperscale datacenters, and diverse accelerators.

The Most Efficient AI Computer, Built in Software

We build AI infrastructure software for frontier models, hyperscale datacenters, and diverse accelerators.

Our Product and Solutions

With Moreh’s comprehensive software suite, any AI model and application can run on cost-efficient hardware, including AMD GPUs and Tenstorrent accelerators.

Inference

Software

MoAI Inference Framework

The fastest distributed inference on AMD GPU clusters, powering cost-efficient generative AI at scale.

Software

Moreh vLLM

Optimal inference performance on AMD GPUs.

Solution

End-to-End Deployment

On-premises inference box, cluster, or cloud-hosted API, delivering maximum tokens per dollar.

Training

Software

MoAI Training Framework

Automatic distributed training on massive scale AMD GPU clusters.

Solution

Large Scale Training

Infrastructure software support for developing powerful models faster and at lower cost.

Infrastructure

Solution

AMD GPU Appliance

Fully integrated AMD GPU-based cluster systems with scalable RoCE networking.

Solution

Tenstorrent Appliance

The lowest TCO with inherently scalable, network-integrated chips for both inference and training.

Software

MoAI Platform

A universal and integrated K8s-based AI platform with flexible GPU virtualization.

Blog Posts

Distributed Inference on Heterogeneous Accelerators Including GPUs, Rubin CPX, and AI Accelerators
September 23, 2025

MoAI Inference Framework supports automatic and efficient distributed inference on heterogeneous accelerators such as AMD MI300X + MI308X and NVIDIA Rubin CPX + GPU.
Moreh vLLM Performance Evaluation: Llama 3.3 70B on AMD Instinct MI300X GPUs
August 30, 2025

Moreh vLLM achieves 1.68x higher output TPS, 2.02x lower TTFT, and 1.59x lower TPOT compared to the original vLLM for Meta's Llama 3.3 70B model.
Moreh vLLM Performance Evaluation: DeepSeek V3/R1 671B on AMD Instinct MI300X GPUs
August 29, 2025

Moreh vLLM achieves 1.68x higher output TPS, 1.75x lower TTFT, and 1.70x lower TPOT compared to the original vLLM for the DeepSeek V3/R1 671B model.

Latest News

Moreh and SGLang team up to showcase distributed inference system on AMD at AI Infra Summit 2025
September 11, 2025

Introducing distributed inference systems on AMD with greater efficiency than NVIDIA, and unveiling collaborations with Tenstorrent and SGLang.
The non-US AI startups competing with Silicon Valley heavyweights
January 28, 2025

Global Corporate Venturing — The startup that perhaps comes closest to DeepSeek’s approach is South Korea’s Moreh, which has created a software tool that allows users to build and optimise their own AI models using a more flexible, modular approach.
Moreh partners with Tenstorrent to challenge NVIDIA in AI data center market
November 18, 2024

Joint R&D of AI data center solutions by integrating Tenstorrent's semiconductors with Moreh's software; Targeting NVIDIA dominant market with competitive solutions.