MangoBoost Unveils AI Factory Plan, Targets 500 MW of Servers Across US, Japan, Southeast Asia
MangoBoost, a startup developing data processing units (DPUs), has launched Mango Inference, a platform for accessing artificial intelligence services. The company plans to make "AI factories" its main business, operating computing servers it has designed. MangoBoost aims to secure 5 MW of data center capacity by the end of next year, with a roadmap for 500 MW across the U.S., Japan, and Southeast Asia.
MangoBoost's decision to directly operate "AI factories" for its Mango Inference platform indicates a strategy to control the full technology stack. By running its inference services on its own DPUs and LLM optimization software, the company claims to offer better AI performance per dollar than existing services, pricing it "tens of percent cheaper" than competitors. This approach aims to differentiate its offering through cost efficiency.
The company's data center expansion plan begins with a 1-megawatt facility in Seoul, expected to be completed within the first quarter of next year. MangoBoost then intends to grow its total capacity to 5 MW in markets including South Korea and the U.S. by the end of next year. These centers will house MangoBoost servers equipped with AMD graphics processing units (GPUs).
MangoBoost's chief executive, Kim Jang-woo, stated that AMD will be a key partner in the AI factory business and that the company has a "roadmap to receive GPUs from AMD in the tens of thousands of units." While the initial 5 MW build-out is reportedly within MangoBoost's financial means, the more ambitious plan to install 500 MW of servers in the U.S., Japan, and Southeast Asia suggests a need for further funding beyond its current capabilities.
Share this article
Related reading
6 stories
IREN Climbs 4% on Report of Unconfirmed Anthropic Cloud Deal; TeraWulf and Applied Digital Hold Flat
We previously covered Anthropic's cloud deal, another example of companies building out AI infrastructure.

Anthropic's IPO bank closed position meant to stabilize stock, removing source of support
Apple develops Mac AI agent, plans third-party control safeguards - TechGig

Gemini 4 Neon Costs More Than Carbon Until December 1. Here’s Which Gemini Model You Should Actually Use

How to deploy an app you built with AI to your own server

