MangoBoost teams with AMD on Mango Inference AI service

South Korean startup MangoBoost launched Mango Inference, a serverless AI inference platform running on AMD Instinct GPUs, and deepened its partnership with AMD to challenge Nvidia. Its LLMBooost software cuts costs by at least 20%, with a 1MW Seoul facility due for full completion in Q1 2027.

• Mango Inference is a serverless service billed by usage with a single API key
• LLMBooost software reduces costs by at least 20% via data-transfer optimization
• MangoBoost plans to procure tens of thousands of AMD Instinct GPUs
• 1MW Seoul facility targets full completion in Q1 2027, long-term goal 500MW