MangoBoost 推出 Mango Inference 推理平台，目标明年营收超 1000 亿韩元
MangoBoost Aims for 100 Billion Won Revenue Next Year via AI Infrastructure Rental < Semiconductor < 기사본문 - The Elec Inc.
MangoBoost 于 10 月 12 日正式商用其 AI 推理平台 Mango Inference，用户可通过 API 接入按 token 消耗付费，支持 DeepSeek、GLM 和 MiniMax 等模型。公司称定价比竞争对手低 20% 以上，并计划明年一季度在首尔江南建成 1MW AI 数据中心，可扩展至 5MW，GPU 由 AMD 供应。其今年营收预计约 100 亿韩元，明年目标超过 1000 亿韩元。
正文
当前显示原文，中文全文译文尚未就绪。
Company launches AI inference platform 'Mango Inference' Plans to build up to 5MW data center in Gangnam, Seoul
Lee Jun; Published 2026.10.13 14:27; Reply 0
MangoBoost is expanding beyond AI server design into AI infrastructure rental services, combining its proprietary server and networking technologies with AI inference software.
MangoBoost announced on Oct. 12 that it launched its AI inference platform, Mango Inference, at a press conference held at Grand InterContinental Seoul Parnas in Gangnam District, Seoul. The company first unveiled Mango Inference in July and officially launched full commercial service on Oct. 12.
Mango Inference enables AI companies to run AI inference models without building their own data centers. Users can connect their application programming interfaces (APIs) to the platform to process AI workloads, paying fees based on token consumption. Supported AI models include DeepSeek, GLM, and MiniMax, with plans to support new models within a month of release.
MangoBoost emphasized that its business strategy differs from existing neocloud providers. "Neocloud providers purchase off-the-shelf servers, virtualize them, and then rent out computing resources," said Jangwoo Kim, Chief Executive Officer (CEO) of MangoBoost. "In contrast, we build the computers ourselves, integrate our own software, and deliver the complete solution to customers."
The company stated its pricing is more than 20% lower than competitors, with further cost reductions depending on the AI model, performance, and token consumption. MangoBoost explained that its proprietary Data Processing Units (DPUs), graphics processing unit (GPU) servers, storage servers, and networking technology help cut infrastructure costs and boost operational efficiency. "Nvidia and MangoBoost are the only companies with this kind of business model," Kim added.
MangoBoost is also expanding its data center infrastructure to support Mango Inference. It plans to establish a 1-megawatt (MW) AI data center in Gangnam District, Seoul, by the first quarter of next year, with plans to expand capacity up to 5MW. The facility will feature MangoBoost-designed GPU servers, storage systems, and DPU-based networking technology. GPUs will be supplied by Advanced Micro Devices (AMD), while memory will be sourced from Samsung Electronics and SK hynix. MangoBoost currently operates a portion of its data center infrastructure at its headquarters in Gwanak District, Seoul.
MangoBoost expressed confidence in securing GPUs despite global shortages. "AMD has committed to supplying us with hundreds of GPUs," Kim said. "Securing a stable supply while others struggle is one of our key competitive advantages."
The company is targeting revenue of over 100 billion won next year. "We expect around 10 billion won in revenue this year, partly from server and storage system sales," Kim said. "Starting next year, including revenue from Mango Inference, we aim to exceed 100 billion won."
MangoBoost was founded in 2022 by Kim, a former Seoul National University professor. Initially a DPU design specialist, the company has expanded into GPU server and storage server design, as well as AI infrastructure rental services.