MangoBoost launches Mango Inference platform with 1MW Gangnam AI factory planned
MangoBoost launched its AI inference platform, Mango Inference, and began paid service; a 1MW AI factory in Gangnam is targeted for Q1 completion.
MangoBoost launched Mango Inference, an AI inference platform, and has begun its paid service, according to techm.kr. The platform lets customers run open-weight models on MangoBoost infrastructure through an API, without building their own infrastructure.
MangoBoost currently supports five models on Mango Inference and said it aims to add new models within a month of their release.
The company also plans to build a 1MW AI factory in Gangnam, Seoul, targeting completion in the first quarter of next year. It is introducing hundreds of AMD GPUs, with several dozen already secured, and said its software on AMD's MI355X GPU achieved inference performance similar to or higher than Nvidia B300-based systems.
MangoBoost aims to complete its Series B funding round within this year, with proceeds going toward the Gangnam 1MW AI factory. The company plans to expand to AI factories of 1–5MW by next year and up to 500MW long-term, including overseas infrastructure.
Background
MangoBoost has been collaborating with AMD for about three years to improve GPU inference performance, and previously operated by supplying component technologies such as servers and networks rather than directly selling inference services.
Quick answers
What is Mango Inference?
Mango Inference is MangoBoost's AI inference platform, which lets customers run open-weight models on MangoBoost infrastructure via API. Paid service has begun.
How many models does Mango Inference support?
MangoBoost currently supports five models on Mango Inference and aims to add new models within a month of their release.
Where will MangoBoost build its AI factory?
MangoBoost plans to build a 1MW AI factory in Gangnam, Seoul, targeting completion in the first quarter of next year.