en.Wedoany.com Reported - Chinese AI startup DeepSeek has launched an in-house AI chip project, focusing on improving inference workload efficiency and reducing costs. The company aims to lower inference costs through custom-designed processors while reducing reliance on overseas suppliers such as NVIDIA. It has not yet publicly commented.
The project is still in its early stages, primarily targeting AI inference rather than model training. As generative AI applications accelerate, inference has become one of the fastest-growing segments in the AI market. Unlike training, which requires concentrated bursts of computing power, inference must continuously serve a large number of user requests, making cost efficiency, power consumption, and system reliability increasingly important.
Sources say DeepSeek began developing the chip about a year ago and has ramped up hiring efforts in recent months. Rather than open recruitment, the company has quietly recruited experienced chip engineers through targeted headhunting. The team covers chip architecture, verification, and software enablement.
DeepSeek is one of China's fastest-growing foundation model companies, gaining attention with its open-source model DeepSeek-V3 and inference model R1. As service demand continues to rise, computing infrastructure has become one of the company's largest operating expenses. For many AI companies, computing-related costs can account for more than half of operating expenses, while the limited supply and high prices of advanced GPUs are driving more companies to seek custom chip development.
Custom AI chips have become a strategic priority for several leading AI companies worldwide. For DeepSeek, in-house AI chips can help reduce long-term costs, improve deployment efficiency, and strengthen its competitive position. However, chip development is a capital-intensive, long-term endeavor, with the process from architecture design to tape-out and mass production typically taking over a year, meaning this effort is unlikely to immediately impact the competitive landscape of the AI chip market.
Additionally, DeepSeek is also seeking its first round of external financing. Previous media reports indicated that the company plans to raise approximately $7 billion at a valuation of $52 billion to $59 billion. If this round of financing is completed, investments in chip development and other AI infrastructure are expected to become key priorities.
