EB SECURITIES: Optimistic about domestic power operators entering the "algorithm electricity" collaborative business and token going global.
This line suggests laying out the "calculate electricity" business, valuing power operators with cost-effective configurations.
EB SECURITIES released a research report stating that in terms of global Token usage, China's open-source AI models have achieved comprehensive overtaking. In terms of pricing, compared to their American counterparts, Chinese large-model Token output prices have a significant advantage. The power sector is at a phase of valuation bottoming out, and the catalyzing effect of Token going abroad under the domestic electricity price advantage is an opportunity for sector valuation repair. The firm suggests focusing on the "algorithmic electricity" business and valuing power operators that have configurational cost-effectiveness.
Key points from EB SECURITIES include:
Event
Recently, the online AI hosting platform OpenRouter's latest data shows that Chinese open-source artificial intelligence models led by MiniMax, Moonshot AI, and KNOWLEDGE ATLAS have topped the global Token usage rankings. Based on the latest data from this platform, China's open-source AI models have achieved comprehensive overtaking in global Token usage.
Significant price advantage of Chinese large models compared to mainstream American AI models
The biggest advantage of Chinese models lies in achieving performance comparable to top international models at extremely low commercialization costs: currently, mainstream American AI model output prices are all above $10 per million Tokens, while the output prices of mainstream Chinese AI models are generally in the range of 10-20 RMB per million Tokens, providing nearly 7 times the price advantage compared to American models.
With the electricity price advantage combined with industry trends driving, Chinese large models are highly competitive
In the future, as the "East Computes West Numbers" project continues to deepen, the national integrated computing power network accelerates its formation, and clean energy and intelligent computing industries collaborate, combined with domestic companies continuously breaking through in reasoning architecture optimization and computing power scheduling efficiency. In terms of electricity prices, China's overall electricity price is highly competitive, with settlement electricity prices in Xinjiang/Gansu/Ningxia/Shaanxi/Qinghai in the first half of 2025 at 0.202/0.265/0.276/0.312/0.228 RMB per kilowatt-hour; as the "East Computes West Numbers" project gradually lands, not only can it improve the level of green electricity consumption in western regions, but also the electricity price advantage of Chinese computing power is expected to expand. The firm believes that this cost advantage has extremely strong long-term sustainability and will continue to be transformed into the core price competitiveness of Chinese models competing in the global market.
Development of the "algorithmic electricity" business is steadily advancing, with multiple power operators accelerating their layouts
Under the promotion of the "East Computes West Numbers" project and a series of industry policies, China has already built a top-level design and landing path for the synergy of electricity and computing, and is gradually improving a new energy supply architecture with "direct supply as the mainstay and the main grid as the foundation," driving the synergy of computing power hubs and new energy bases. In this context, multiple power operators are actively laying out the "algorithmic electricity" synergy to explore a second growth curve: 1) directly entering the data center field, such as NYOCOR investing in the construction of the Xinjiang Changji 5000P intelligent computing center and signing the Ulanqab intelligent computing center project, and Henan Yuneng Holdings planning to participate in the equity of Sentian computation power company and acquire the shareholding of Zhengzhou Heying data control company; 2) some operators are involved in the "algorithmic electricity" business by providing power to data centers, such as China Longyuan Power Group Corporation collaborating with Shanghai Xintianweng Digital Technology Co., Ltd. to successfully deploy distributed computing power nodes and regional industry computing power networks, and Jinko Power Technology's Tencent Haila East Garden data center demonstrating the "wind storage" new energy microgrid project.
Investment recommendation: suggested focus on: Fujian Funeng, Gepic Energy Development, NYOCOR, Henan Yuneng Holdings, Chongqing Fuling Electric Power Industrial, etc.
Risk analysis: Low expectations for the implementation of "algorithmic electricity" synergy policies, low expectations for AI development progress.
Related Articles

MNSO (09896) spent 1.9132 million Hong Kong dollars to repurchase 56,600 shares on March 3rd.

Angang Steel (00347) appoints He Yongya as its legal process agent.

XINYI SOLAR (00968) will distribute a final dividend of 0.8 Hong Kong cents per share on July 3rd.
MNSO (09896) spent 1.9132 million Hong Kong dollars to repurchase 56,600 shares on March 3rd.

Angang Steel (00347) appoints He Yongya as its legal process agent.

XINYI SOLAR (00968) will distribute a final dividend of 0.8 Hong Kong cents per share on July 3rd.

RECOMMEND





