Guizhou: Forging AI Production Capabilities for the Era of the Token Economy

Deep News
Aug 26

On August 17, Guizhou Suanjia Computing Services Co., Ltd. sealed a cooperation agreement to jointly operate computing power with Beijing-based SiliconFlow and Guizhou Shujia Technology. The three parties will leverage the computing resources of the Guian New Area to efficiently convert computing power into token services and build a token factory. "The resource fit is highly compatible and the cooperation is highly complementary, making the joint construction of the token factory a natural step," said Yang Jiawei, General Manager of Suanjia Computing.

SiliconFlow is a leading AI infrastructure service provider in China, and its token supply platform has accumulated over 10 million registered users. This partnership deploys its full suite of AI computing operation solutions to local enterprises in Guian. "Every step of the foundation has been laid, and it just happened to coincide with the AI boom," Yang Jiawei told reporters. Suanjia's revenue has grown from less than 10,000 yuan in 2022 to over 10 million yuan in 2025, with contract amounts reaching 50 million yuan in the first half of this year and an expected annual contract value exceeding 100 million yuan.

On August 28, the 2026 China International Big Data Industry Expo will open in Guiyang, themed precisely on "Tokens: A New Path for Unleashing the Value of Data Elements." This collaboration comes against the backdrop of exponential growth in China's token economy. At the start of 2024, the national daily token call volume was only 100 billion; by the end of 2025, it had surged to 100 trillion, and by June of this year, it had skyrocketed to nearly 175 trillion.

In Guizhou, the brand direction of "For tokens, look to Guizhou" is now clear. A special support policy was released in July, offering subsidies of up to 500,000 yuan for enterprises purchasing large-model token call services and rewards of up to 5 million yuan for building high-quality datasets. The "Guiyang Guian Token Economy Three-Year Action Plan (2026-2028)" also clarifies efforts to strengthen intelligent computing services, data element supply, scenario application traction, and industrial ecosystem cultivation.

Refining Computing Power: From Full-Service Leasing to Pay-Per-Token

Suanjia Computing was jointly incubated in January 2022 by the Guian Supercomputing Center and Guizhou University. Its Suanjia Cloud Platform went live in October 2024, and its Suanqiao API token platform was officially opened in May of this year. "The Suanjia Cloud Platform has nearly 60,000 registered users, with over 90% coming from provinces and cities with strong computing power demand, such as Guangdong, Shanghai, and Beijing," Yang Jiawei noted. The Suanqiao API now hosts over 60 mainstream large models, including DeepSeek, Zhipu, and Kimi, and within half a month of launch, daily token call volume exceeded 1 billion tokens.

Suanjia packages computing power into token services, charging customers based on the number of tokens consumed. "You buy what you use, which lowers the entry barrier." Computing power is now sliced and billed by time and token volume, with a more precise and flexible billing logic. Two years ahead of schedule, Suanjia had already deployed a distributed computing scheduling system, boosting comprehensive GPU utilization from the industry average of 30%-40% to 60%-70%. Combined with optimizations at the model inference layer, platform pricing can be maintained at around 50% of the market average.

The high proportion of out-of-province users on Guizhou's AI data factory is, in Yang Jiawei's view, both a result of the "East Data, West Computing" initiative and a reflection of the local market's current stage. "Demand for computing power outside the province is higher than our local demand, and computing costs in the east are also higher than in Guizhou," he said. However, the company is also actively expanding cooperation within the province. "Based on the open-source ecosystem and combined with industry data, we can build small industry-specific models, and these can fully utilize our own computing power." He also observed that some AIGC companies in the province still tend to directly call overseas or out-of-province computing interfaces of official large models, "but customers using open-source ecosystems to build industry agents can cooperate deeply with us, which is exactly the opportunity for Suanjia to expand its local ecosystem next."

The powerful computing infrastructure is accelerating its transformation into token services. The computing scale in Guiyang and Guian has exceeded 174 EFLOPS, with intelligent computing accounting for over 98%. The province has 50 data centers under construction or in operation, with 400G high-speed computing channels directly connecting to the Yangtze River Delta and the Greater Bay Area. Computing usage rates exceed 96%, with out-of-province customers accounting for over 90%. In 2025, the province's computing operation service revenue reached 82.4 billion yuan.

Aggregating the Ecosystem: From Selling Computing Power to Selling Tokens

"Computing power is just the entry ticket," said Deng Mingsen, Dean of the School of Information at Guizhou University of Finance and Economics and Director of the Provincial Key Laboratory of Computing Network Integration. He believes Guizhou's real advantage lies in the vast amount of real, vertical, and professional industrial and public data scenarios accumulated through over a decade of industrial digital transformation.

In this leap from computing power to tokens, local Guizhou enterprises are quickly staking their claims. "The platform's daily token call volume has grown rapidly from less than 3 billion at the start of 2026 to hundreds of billions now," said Peng Hailin, Director of DataBao and Rotating General Manager of its AI Token line, on August 20, showcasing the latest data from TokenBao.

In May of this year, DataBao's TokenBao was officially launched, spinning off its core token business. Leveraging the parent company's over 20 years of data governance experience, it has built a comprehensive service system of "intelligent routing and scheduling + industry-level tuning + compliant and stable supply." "DataBao's core team has over two decades of legally authorized data systems as its foundation, allowing us to create 'data-native' tokens that others can't replicate in the short term," Peng Hailin explained. DataBao has built a nationwide network for public data, idle computing scheduling, and full-scenario applications.

TokenBao is now building China's first "Data API + Token API + Computing API" three-in-one full-chain platform. The data API connects directly to over 50 ministries and commissions, more than 10 provinces, and over a hundred state-owned enterprises. The computing API connects to a national intelligent computing infrastructure of millions of PFLOPS, while the token API provides both general and vertical tracks. Previously, companies building AI applications had to negotiate three separate interfaces and contracts for data, computing, and tokens, often taking weeks to get everything running. Now, with "one access, full-chain connectivity," developers can connect all three in just a few minutes.

Connecting Scenarios: From Technology to Experience

On August 3, on Taiping Road in Guiyang's Yunyan District, the AI travel product "Banzhi Mengxing" from Gansheng Tengwei Technology saw over 2,000 users on its first day. "Scan the QR code and upload a photo, and within seconds you get a cyberpunk-style photo of yourself on the old street," said a tourist from Guangdong, Ms. Chen, who immediately posted her unique image on social media. The conversion rate was about 50%.

"Cultural tourism is the easiest scenario for ordinary people to perceive the value of AI," said Feng Yuwei, General Manager of Gansheng Tengwei. "We need to turn abstract tokens into products that tourists can experience, purchase, and share." "Banzhi Mengxing" has already transformed the "Red Ribbon" experience, allowing visitors to shift from "observing history from the sidelines" to "participating in the creation of digital keepsakes that can be saved and shared." In Yunyan District, it has been deployed at locations such as Taiping Road, Minsheng Road, Shengfu Road, Ditai Street, and Wenchang Pavilion. "Taiping Road uses a cyberpunk style, but Minsheng Road doesn't, because it's a snack street with a more down-to-earth vibe," said Liu Ziyun, the project's executive lead. The company is also collaborating with the Guiyang Grand Theater's "Colorful Guizhou Style" show. "AI plus tourism is about capturing the emotion you feel in that very moment."

AI short dramas are also gaining traction. At the Yunyan AI Data Factory, Guizhou Yalu Yinghua Digital Technology Co., Ltd. is transitioning from contract manufacturing to original production. "Before the Spring Festival this year, a one-minute contract drama was priced at 2,000 yuan, but after April it dropped to under 1,000 yuan," said General Manager Zou Hailong. A B-grade drama can now be completed in two days, and an A-grade drama in a week, with comprehensive costs at about one-tenth of traditional film production. The company is now working with the Guizhou Cultural Equity Exchange to explore local cultural IPs.

"Three to five years from now, Guizhou will not just be an industry that sells tokens. Instead, it will form a new AI industrial ecosystem integrating computing supply, token production, model services, scenario applications, and value settlement. In the era of the token economy, Guizhou should leap from selling computing power to selling AI production capabilities," Deng Mingsen predicted.

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Most Discussed

  1. 1
     
     
     
     
  2. 2
     
     
     
     
  3. 3
     
     
     
     
  4. 4
     
     
     
     
  5. 5
     
     
     
     
  6. 6
     
     
     
     
  7. 7
     
     
     
     
  8. 8
     
     
     
     
  9. 9
     
     
     
     
  10. 10