The emergence of DeepSeek at the beginning of the year brought artificial intelligence into the era of “deep thinking”, allowing AI to break through the “dimensional wall” and become an exponentially growing national demand. DeepSeek has driven the explosion of reasoning demand, which not only accelerated the penetration speed of the AI industry, but also brought a new round of challenges to computing power.
Related research reports point out that DeepSeek has cracked the long-term path dependence of the global AI industry of “stacking computing power”, and the combination of “low cost + open source” has greatly reduced the threshold for AI application, which is expected to stimulate the wave of AI innovation in the whole society and accelerate the popularization and application of AI in all walks of life.
In addition, in March, AI company Monica released a general intelligent agent named Manus. This product, known as the “world’s first general AI agent”, quickly attracted widespread attention with its ability in “thinking” and “action”. The uniqueness of Manus lies not only in its ability to understand language and analyze data, but also in its ability to operate and solve real problems like humans.
AI technology leap drives computing power demand
This capability breaks through the shackles of traditional large language models (LLM) that are limited to text generation, and demonstrates the potential of “large behavioral models” (LAM). Previously, Google (GOOG)’s Project Astra and Microsoft (MSFT)’s Copilot Studio have already laid out the intelligent body ecosystem. Manus, together with these cutting-edge products, is considered an important node in the process of “AI intelligent body commercialization”.
It is worth noting that when the AI wave sweeps across thousands of industries, the computing power revolution led by AI chips is moving from silence to noise. Driven by this round of DeepSeek and Manus craze, companies are accelerating the layout of cloud and AI hardware infrastructure. For example, Alibaba (BABA) emphasized that it will invest more than 380 billion yuan in the next three years, and ByteDance will spend 90 billion yuan to purchase computing power cards in 2025.
Industry insiders commented that the lower threshold of large model application will activate the reasoning computing power market, especially DeepSeek’s extreme optimization of the hardware end, which will tap the potential of AI chips. At the same time, for the demand for AI chips, training clusters are still the core elements of the development of AI large models. With the popularization of models and the expansion of applications, the inference market will be fully activated, and the demand for inference computing power will be rapidly amplified.
WiMi builds an advantage moat for AI chips
According to data, AI chip manufacturer WiMi Hologram Cloud Inc. (NASDAQ: WIMI) faces the new variable of computing power demand and vigorously develops new quality productivity. Under the collaboration of the industry, it integrates “artificial intelligence +” and deeply integrates with all walks of life, and accelerates breakthroughs in algorithm optimization, expands the research and development of high-performance AI chips, and lowers the threshold for the use of AI, which is expected to further open up the autonomous and controllable space of AI chips.
At present, the reason why WiMi has attracted much attention is due to its deep and excellent foundation in the field of computing power. It is reported that WiMi has established a high-end computing power base for artificial intelligence-the holographic cloud platform. Through massive data supply and high-quality computing power output, it builds a digital base for intelligent computing and releases chip computing power capabilities. The platform supports the construction of a multi-computing power architecture with international advanced chips, supports large model training and inference needs, and adapts to vertical model applications such as embodied intelligence and multi-modality.
AI computing power network is accelerating global layout, optimizing resource allocation and reducing costs. In the future, with the explosive growth of AI large model training and reasoning needs, WiMi will explore customized chips to develop ASIC chips, lower technical barriers, attract developers and enterprises to build an ecosystem, and provide users with flexible and low-threshold computing power solutions, fill the resource gap, promote computing power support stability and diversification, and help the popularization and innovation of AI technology.
Ending
At present, the AI industry is ushering in an unprecedented golden period of development. Computing power is a new type of productivity that integrates information computing power, network carrying power, and data storage power. The rapid development of artificial intelligence has given rise to the explosive growth demand for computing power. Looking to the future, with the continued rise of computing power and the widespread application of AI model technology, chip manufacturers will also play a more important role in the global AI field.
