News
To "change the world" in the AI era, should we buy electricity, computing power, or build a base?
2 min read
Source: zhidx.com
Zhixi Author | Chen Junda Editor | Mo Ying Zhixi reported on October 13 that recently, "Fortune" magazine released the 2026 "50 Companies Changing the World" list, focusing on companies that provide solutions to major global problems through innovative products, business models or technical solutions. This year, important companies in the AI field such as Alphabet and Anthropic have appeared on the list, reflecting that AI is becoming an important force in promoting changes in the industry and the world. However, the way AI changes the world is not just through breakthroughs in model capabilities and improvements in chip computing power. As large model training and inference continue to consume more computing power, the energy supply and infrastructure supporting AI operations have begun to become a part of this transformation that cannot be ignored. Envision Technology Group is one of the few Chinese companies that entered the list from this dimension. With its "Mission Gobi" and its Ulanqab Galaxy Base, which has the world's largest AI computing supercell, Envision has been selected for the third time on the list of "50 Companies Changing the World". As companies such as Alphabet and Anthropic continue to explore the upper limit of intelligent capabilities, Vision is trying to answer another question: When AI continues to break through the boundaries of capabilities, where does the energy to support its continuous operation come from? Behind the inclusion of several companies on the same list, an industry change worthy of attention is emerging: the competitive dimension of the AI competition is extending from chips and models to energy infrastructure. 1. After buying electricity and computing power, AI giants enter the deep water area of energy. Electricity is becoming an increasingly scarce resource in the AI computing power arms race. Faced with the rising demand for computing power, Alphabet, Anthropic and Vision have different solutions. Alphabet chooses to “buy green electricity”. It relies on the existing power grid, purchases clean electricity on a large scale, and locks in renewable energy through long-term power purchase agreements (PPA) to supply energy for data centers. Anthropic choose "Buy computing power". Through in-depth cooperation with infrastructure providers, it locks in GW-level computing resources and quickly obtains the requirements for model training and inference.