AI infrastructure solutions corporations MOREH took part in AMD's annual flagship artificial intelligence (AI) event, Advancing AI 2026, held in San Francisco, and showcased its distributed inference technology. In particular, AMD Chief Executive Officer Lisa Su and Vice Prime Minister and Minister of the Ministry of Science and ICT Bae Kyung-hoon visited the booth together, drawing attention simultaneously from global big tech and the Korean government.
On the 23rd (local time), at MOREH's booth inside the Advancing AI venue, MOREH CEO Cho Kang-won explained the company's distributed inference solution to CEO Lisa Su and Minister Bae Kyung-hoon. The demo ran on a system composed of 32 AMD MI300X GPUs (four nodes).
Through the MoAI inference framework, MOREH directly demonstrated to CEO Su and Minister Bae that the large language model (LLM) GLM-5.1 delivers responses in real time. Su closely examined the processing speed displayed on the screen, and Minister Bae Kyung-hoon was said to have focused on the overseas expansion results of a domestic AI infrastructure corporations.
On the booth screen, performance metrics such as GPU utilization, tokens per second (TPS), time to first token (TTFT), and time per output token (TPOT) were displayed in real time. Other visitors also operated the chatbot themselves to experience the response speed.
In the industry, some view this visit as an opportunity for MOREH to emerge as a key collaboration case for the AMD camp amid efforts to reduce reliance on the Nvidia-centric GPU ecosystem. The government is also said to be focusing on expanded cooperation with AMD as a case that demonstrates the overseas competitiveness of Korea's AI infrastructure technology. In fact, on this day, Minister Bae Kyung-hoon also met with reporters and said, "Dependence on Nvidia is risky," and added, "We need a more open AI ecosystem."
MOREH CEO Cho Kang-won said, "This event was a venue where global customers directly verified that top-tier inference performance can be achieved even in an AMD GPU environment," and added, "We will continue to advance our AI infrastructure software technology so that corporations can operate AI services most efficiently regardless of GPU type."