The computing services market has entered a price-increase cycle, driven by the proliferation of AI applications and the popularity of OpenClaw. Major cloud service providers, including Alibaba Cloud, Tencent Cloud, and Baidu AI Cloud, have successively announced price hikes for AI computing power and related services.

On March 18, Alibaba Cloud announced on its official website that, due to surging global AI demand and rising supply chain costs, procurement expenses for core hardware have increased significantly. After careful evaluation, the company decided to adjust prices for services such as AI computing power and CPFS starting April 18. Products including the T-Head Zhenwu 810E computing cards will see price increases ranging from 5% to 34%, while the CPFS file storage service will rise by 30%.

On the same day, Baidu AI Cloud issued a notice stating that, effective April 18, it will raise prices for AI computing power and storage products. AI computing-related services will increase by approximately 5% to 30%, with parallel file storage services rising by about 30%.

Earlier this month, Tencent Cloud preemptively adjusted prices for some of its models. Starting March 13, it increased prices for its Hunyuan series models, with some core products surging by as much as 400%. The collective action by these three leading cloud providers signals the beginning of an industry-wide price hike trend.

Since the beginning of the year, major overseas cloud service providers have also raised prices for core products. On January 22, Amazon announced a 15% price increase for EC2 instances used in large model training. On January 27, Google Cloud stated it would raise prices for data transfer services, AI, and computing infrastructure, with some increases reaching up to 100%.

Domestic Chinese AI models have seen a dramatic surge in token usage, reflecting their growing competitiveness. Data from the AI model aggregation platform OpenRouter shows that during the week of February 9 to 15, Chinese large models reached a weekly usage volume of 4.12 trillion tokens, surpassing the 2.94 trillion tokens of U.S. models for the first time.

In the following week, from February 16 to 22, usage of Chinese models further surged to 5.16 trillion tokens, marking a 127% increase over three weeks. Meanwhile, usage of U.S. models declined to 2.7 trillion tokens, indicating a substantial breakthrough in the influence of domestic models within the global developer ecosystem.

The growth in traffic for Chinese models has shown a "cluster-style" explosion. OpenRouter data indicates that OpenClaw has become a major traffic gateway for domestic models. As of March 18, Step 3.5 Flash and MiniMax M2.5 ranked first and second in total usage volume on OpenClaw over the past month, demonstrating strong competitiveness. A recent performance announcement from MiniMax revealed that daily token consumption for the M2 series in February 2026 increased more than sixfold compared to December 2025.

A research report from Guojin Securities suggests that 2026 will be a pivotal year for China's computing power demand, transitioning from "cloud-based training" to a dual-driver model of "training plus inference." The computing power gap is expected to widen rapidly, catalyzed by more modalities and broader application scenarios. Consumer-side traffic growth, combined with the rise of native AI applications such as AI-generated dramas and programming, alongside the preparation of vertical B2B models, will collectively drive a significant increase in real-time inference computing consumption.

On March 18, the computing power leasing sector saw a significant rally in the A-share market. Stocks such as Meili Cloud, YunSai ZhiLian, DataGo, and Aurotek hit the daily limit up, while others like Glodon Company and UCloud Technology saw gains exceeding 10%.

As computing demand surges, the profitability of computing leasing companies has improved markedly. According to statistics, 32 concept stocks related to computing leasing have released performance data for 2025. Based on annual reports, preliminary earnings reports, and the lower bounds of forecasts, 20 of these stocks achieved profit growth. Among them, seven are expected to turn a profit from a loss, eight reported year-on-year net profit growth, and five reduced their losses.

Industry leaders XinYisheng, Runze Technology, and Zhongji Innolight ranked top three in growth. XinYisheng's performance forecast indicates an estimated net profit of 9.4 to 9.9 billion yuan, a year-on-year increase of 231.24% to 248.86%. The company attributed this growth to sustained investment in computing power and rising demand for high-speed products, leading to a significant rise in both sales revenue and net profit.

Runze Technology's forecast shows an expected net profit of 5.0 to 5.3 billion yuan, up 179.28% to 196.03% year-on-year, primarily due to non-recurring gains from the successful issuance of a public REIT. In August 2025, the Southern Runze Technology Data Center REIT was successfully listed on the Shenzhen Stock Exchange, raising 4.5 billion yuan and becoming one of the first data center REITs in China. The company's adjusted net profit increased by 5.71% to 11.33%, driven by rapid AI industry development and breakthroughs in business layout, particularly in AIDC services.

From a capital flow perspective, several computing leasing stocks have recently seen increased financing. Since March, eight stocks have recorded net financing purchases exceeding 100 million yuan, with Tuowei Information, Kehua Data, and XinYisheng leading at 580 million, 301 million, and 256 million yuan, respectively.