Price hike wave arrives! Leading stocks hit the 20CM daily limit up, does the token face an inflation cycle?

Computing Power Concept: The morning session on March 19th showed relative strength, with copper information, Lotus Holdings, Jinkai New Energy, Shen Sangda A, Litong Electronics, and Aorui De hitting the daily limit; Kai Tian Gas, Hongjing Technology, Jiechuan Intelligent, and Yihua Recording surged over 10%.

Exponential Surge in Computing Power Demand

Amid the recent “shrimp farming” craze, AI intelligent agents are sweeping across industries, integrating into daily work and life at an unprecedented speed. Behind this trend is an exponential increase in computing power demand.

This can be seen from the sustained high activity in AI data center investments. Guojin Securities believes that major domestic and international companies continue to increase CapEx, with Silicon Valley’s four tech giants expected to reach $650 billion in CapEx by 2026, further intensifying the AI arms race. Specifically: Amazon has become the largest investor among the four, setting its 2026 capital expenditure target at $200 billion; Alphabet’s CapEx plan is between $175 billion and $185 billion, nearly doubling year-over-year; Meta expects its annual CapEx to rise to $135 billion, an 87% increase; Microsoft announced a 66% YoY growth in its second-quarter CapEx, with its fiscal year CapEx approaching $105 billion by June.

Additionally, intelligent computing centers are expanding continuously, with domestic substitution accelerating. According to IDC data, China’s intelligent computing power reached 75.0 EFLOPS in 2020 and is projected to hit 2781.9 EFLOPS by 2028, with a compound annual growth rate of 57.1% from 2020 to 2028.

Guojin Securities states that, based on multi-dimensional data and industry dynamics, the investment in AI computing infrastructure remains high, with the AIDC sector showing sustained high-growth expansion.

Computing Power Enters a “Full-Chain Inflation” Cycle

Recently, price hikes among computing and cloud service providers also reflect the surging market demand for computing power. On March 18, Alibaba Cloud announced that due to global AI demand explosion and supply chain price increases, its AI computing, storage, and related products will see a maximum price increase of 34% starting April 18, 2026. Among these, products like the Pingtouge Zhenwu 810E computing cards will increase by 5%-34%, and the file storage product CPFS (Intelligent Computing Edition) will rise by 30%.

Baidu Smart Cloud also announced on March 18 that starting from midnight on April 18, prices for AI computing-related products and services will increase by approximately 5%-30%, with parallel file storage rising about 30%. The reason is the continuous rise in demand for computing power, along with significant increases in core hardware and infrastructure costs.

In fact, prior to these recent price hikes by Alibaba Cloud and Baidu Smart Cloud, there have been frequent reports of price increases among cloud and computing power providers since 2026:

• January 23: AWS raised the price of its EC2 Capacity Blocks for ML by about 15%, with the p5e.48xlarge instance’s hourly cost increasing from $34.61 to $39.80.
• January 27: Google Cloud officially announced a price increase starting May 1, 2026, affecting Google Cloud, CDN Interconnect, Peering, and AI infrastructure services.
• February 12: Zhipu announced structural adjustments to its GLMCoding Plan pricing, with increases starting at 30%.
• March 11: Tencent Cloud’s AI development platform announced adjustments to some model billing strategies from March 13, 2026, ending free public testing for models like GLM 5, MiniMax 2.5, and Kimi2.5, switching to pay-as-you-go. Additionally, prices for the Hun Yuan series models Tencent HY2.0 Instruct and Tencent HY2.0 Think increased, with input prices for HY2.0 Instruct rising from 0.0008 yuan per thousand tokens to 0.004505 yuan, and output prices from 0.002 to 0.01113 yuan; for HY2.0 Think, input prices rose from 0.001 to 0.0053 yuan, and output prices from 0.004 to 0.0212 yuan. Users can offset some costs through packages.

Under this context, Guojin Securities states that, driven by strong supply and demand logic, the computing power industry chain is expected to enter a “full-chain inflation” cycle by 2026, with industry prosperity spilling over from core chips to AIDC, cloud and computing services, supporting power equipment, and servers.

Is Token Inflation Behind the Scene?

Data shows that tokens are the smallest units for processing text in large models, akin to AI’s workload units. When industry talks about token inflation, it refers to the surge in complexity of AI processing tasks, which results in higher computing resource consumption for the same service.

Domestically, as AI intelligent agents are scaled up, token consumption is skyrocketing. Guolian Minsheng Securities estimates that China’s total daily token consumption has surged from 100 billion in early 2024 to 180 trillion by February 2026. As AI agents evolve toward multimodal and multi-agent collaboration, this number continues to accelerate.

Internationally, low-cost electricity is China’s core competitive advantage for token export. Dongwu Securities notes that electricity cost advantages directly impact the cost per million tokens output: Chinese models require only $0.5-$1.5 per million tokens input, while US models cost about $2.5-$10, a gap of over five times.

This cost advantage is fueling a scale explosion of Chinese tokens abroad: according to OpenRouter data, in the third week of February 2026, China’s token consumption accounted for 61%, surpassing the US (39%) for the first time. Over the past month (until March 13), MiniMax M2.5 with 8.67T tokens and DeepSeek V3.2 with 3.68T tokens ranked first and third globally. From February 2025 to February 2026, China’s token consumption share increased by 421%.

Top 20 Performance Forecasts for the Computing Power Concept

In stocks, under the high prosperity of computing power, some stocks’ institutional forecasted performance growth is also impressive. For example, Zhongke Xingtu is predicted to grow 1610.23% in 2026; Guanghuan Xinwang is forecasted at 521.97%. Runjian Shares, He Dun Intelligent, Jingjiao Micro, and others also show strong predicted growth.

In terms of stock price performance, as of March 18, HJ Technology has doubled since 2026, up 137.3%. UCloud-W increased over 60%, and Dongfang Guoxin rose over 50%.

(Source: Eastmoney Research Center)