The graphics processing unit (GPU) and the central processor unit (CPU) handle different sorts of data. All electronic gadgets require CPUs, including computers, TVs, smartphones, tablets, and other mobile devices. It deals with processing the device’s overall data and ensuring the software functions properly. The GPU works with the CPU to perform sophisticated workloads like supercomputing, artificial intelligence (AI), machine learning, and in-depth data processing.
To maintain a consistent supply of semiconductors, the COVID-caused chip scarcity prompted international chipmakers to look for alternatives to conventional chips. At the same time, businesses are always looking for more effective methods to save expenses and boost chip-level efficiency. Data processing units, or DPUs, have been developed to support CPUs and GPUs. To optimize workload for cloud and data centers and to save costs, the DPU offloads communication networking from the CPU or GPU. It also helps to minimize the stress on more costly GPUs and CPUs.
MangoBoost, a DPU developer, has secured $55 million in Series A funding for its DPU hardware and software solutions that assist businesses and data centers in managing enormous volumes of data and streamlining workloads. MangoBoost did not disclose its value in today’s round, but sources familiar with the situation believe it to be in the $300 million range. Currently, the business has raised $65 million.
The firm, with offices in Seattle and Seoul, claims that its DPU solution enables data centers to reduce energy use and improve performance while maintaining cost-effectiveness and security. When coupled with Samsung’s Petabyte SSD storage system, MangoBoost says its DPU can produce performance three times greater than competing products and a 95% reduction in CPU utilization.
The one-year-old startup will use the money from the investment to speed up the creation and use of its products, such as DPU hardware intellectual property (IP), DPU software, DPU based on FPGA, ASIC, and AI, as well as custom DPU solutions and systems that use DPUs more effectively. The company presently employs 58 people, with 71% of them working on the R&D team. It will use the money to quadruple that number by the end of the next year.
Jangwoo Kim, the CEO of MangoBoost, claims that more than nine years of research and development at the lab of Seoul National University into the operation of the DPU technology for data centers helped the firm achieve efficiency.
As Big Tech firms and semiconductor behemoths like Intel, Nvidia, AMD, Amazon, and Microsoft invested in DPU providers to strengthen their optimization technologies for data center services, the competition in the DPU industry is escalating. Many businesses provide DPU-like technologies, including Amazon Nitro and Microsoft FPGA Smart NIC.
According to Kim, Intel, Nvidia, and AMD commonly provide their DPUs as PCIe cards with a custom DPU processor. The fact that MangoBoost’s solutions offer a comprehensive and adaptable range of DPU features that may satisfy any customer’s demands is one of their main differentiators, he continued. That includes the DPUs for cloud, large data, and AI servers.
Our objective, according to Kim, is to offer full-stack DPU hardware and software solutions to meet the different demands of clients. The company presented at OCP Summit 2023 regarding its collaborative work with Samsung in utilizing MangoBoost’s customized DPU to speed up Samsung’s Peta Byte storage. The company is now in talks with several potential clients for partnerships.
IMM Investment and Shinhan Venture Investment, two South Korean venture capital firms, jointly led series A. The most recent round included participation from the Korea Development Bank, KB Investment, Hong Kong-based IM Capital, and Premier Partners.