Moreh & SGLang Demo Distributed Inference on AMD at AI Infra Summit

Introducing distributed inference techniques on AMD with excessive effectivity, and unveiling collaborations with Tenstorrent and SGLang

Moreh, an AI infrastructure software program firm, unveiled its distributed inference system on AMD and showcased the progress of its collaborations with Tenstorrent and SGLang at the AI Infra Summit 2025 in Santa Clara, held September 9–11.

The AI Infra Summit is the world’s largest and most established AI convention devoted to the infrastructure layer of AI & Machine Learning. Originating because the AI Hardware Summit again in 2018, the summit has advanced from a semiconductor-focused convention right into a full-stack AI infrastructure occasion.

The 2025 summit attracted 3,500 attendees and over 100 companions, with content material designed for {hardware} suppliers, hyperscalers, and all enterprise IT and AI infrastructure specialists constructing quick, environment friendly, and reasonably priced AI.

At the Enterprise AI session on September 10, Moreh CEO Gangwon Jo launched the corporate’s distributed inference system and offered benchmark outcomes demonstrating that it optimized the most recent deep studying fashions, resembling DeepSeek, extra effectively than NVIDIA. He additionally unveiled a next-generation AI semiconductor system combining Moreh’s software program with Tenstorrent’s {hardware}, providing a spread of cost-competitive options to NVIDIA.

During the summit, Moreh co-hosted a presentation with SGLang, a frontrunner within the deep studying inference software program ecosystem, and arranged a sales space and networking classes collectively. This serves as a chance to additional strengthen collaboration with the worldwide AI ecosystem, significantly within the North American market. Furthermore, Moreh plans to collectively develop an AMD-based distributed inference system with SGLang to speed up its enlargement of the quickly rising deep studying inference market.

Moreh CEO Gangwon Jo said, “Moreh possesses the strongest technical capabilities amongst AMD’s world software program companions and is at present conducting proof-of-concept (PoC) initiatives with a number of main LLM firms,” and added, “Through shut collaboration with AMD, Tenstorrent, and SGLang, we purpose to determine ourselves as a world firm offering clients with various AI computing options.”

Moreh is growing its personal core AI infrastructure engine and, by way of its basis LLM subsidiary Motif Technologies, is securing complete technological capabilities that span the mannequin area. Simultaneously, the corporate is making its mark within the world market by way of collaborations with key companions resembling AMD, Tenstorrent, and SGLang.

The put up Moreh & SGLang Demo Distributed Inference on AMD at AI Infra Summit first appeared on AI-Tech Park.