In recent years, with the rapid development of AI, major domestic internet giants have been actively deploying large AI models. Especially recently, the popularity of Lobster has allowed various AIs to make a fortune. However, as expected, the AI price hike wave has arrived on schedule. With overseas giants like Amazon and Google, as well as domestic giants like BAT collectively raising prices, domestic large model companies such as Zhipu have also followed suit. This makes one wonder: with the price hike wave here, can we still afford Lobster employees?
I. Giants Collectively Raise Prices
According to a report by Haibao News, recently, concept stocks related to optical modules and computing hardware have surged against the trend. Behind this is the OpenClaw (commonly referred to as "Lobster" domestically) craze sweeping from abroad to China this year, causing a surge in Token calls and directly driving up computing power demand. The Token industry chain has become a new core focus.
Alibaba Cloud and Baidu Intelligent Cloud have successively announced price increases for AI computing power-related products. Alibaba Cloud's official announcement stated that due to the global AI demand explosion and supply chain price hikes, Alibaba Cloud's AI computing power, storage, and other products have seen price increases of up to 34%. According to insiders, Alibaba Cloud's MaaS business platform, Bailian, achieved its highest growth rate in history from January to March this year.
Baidu Intelligent Cloud stated that due to the rapid development of global AI applications, computing power demand continues to rise, and the costs of core hardware and related infrastructure have significantly increased. Therefore, it decided to raise the prices of AI computing power-related products and services by 5% to 30%, with parallel file storage increasing by about 30%.
Earlier this year, Amazon AWS and Google Cloud announced price increases for some services. On January 22, AWS announced a 15% price increase for EC2 used for large model training. On January 27, Google Cloud significantly adjusted prices for data transmission services such as CDN Interconnect, Direct Peering, and Carrier Peering, with a 100% increase in North America.
Domestically, Tencent Cloud announced a price increase for large model services on March 11, adjusting the billing strategies for some models. For example, the input price for the Tencent HY2.0 Instruct model increased sharply from 0.0008 yuan per thousand Tokens to 0.004505 yuan per thousand Tokens, a rise of 463.13%.
Moreover, the price hike wave is not over. On April 8, Zhipu announced its third price increase plan this year, raising prices by 10% with the release of its new flagship model GLM-5.1, just one month after its last price increase of 30% or more. Compared to the price cuts and free trials in 2025, the current clear signal is that domestic AI large models are entering an era of collective price increases.
II. Is the AI Price Hike Wave Coming? Can We Still Afford Lobster Employees?
Recently, major large model giants collectively announced price increases for some of their cloud services and related products, giving those who were worried about being distilled a glimmer of hope. Compared to expensive Tokens, we seem cheaper. So, what is the industrial logic behind this price hike wave?
First, the explosion of AI agents has led to a surge in computing power demand. From the perspective of structural changes on the demand side, the popularity of new-generation AI agents like OpenClaw has completely reshaped the underlying logic of computing power consumption. In the early days of large model applications, user behavior was mostly limited to single-round dialogues or simple text generation, where Token consumption was relatively limited and predictable. However, with the maturity of autonomous agent technology, AI is no longer a passive Q&A machine but a digital employee capable of autonomous planning, tool invocation, and executing complex tasks.
The daily per capita Token consumption of a mature agent is often dozens or even hundreds of times that of traditional chat users. This exponential demand surge is not linear business growth but a dimensional leap. When massive numbers of agents are online simultaneously, engaging in high-frequency logical reasoning and data interactions, the computing power infrastructure originally designed for human interaction suddenly faces enormous throughput pressure.
This explosive growth in demand directly breaks the old supply-demand balance, rapidly shifting computing power resources from "relatively abundant" to "extremely scarce." When marginal utility rises sharply and supply elasticity is insufficient, price increases are not only an inevitable reflection of market laws but also a necessary means to screen high-value application scenarios and curb inefficient computing power waste.
Second, the supply-demand imbalance of core hardware leads to computing power tension. From the perspective of hard constraints on the supply side, the supply-demand imbalance of high-performance computing chips and HBM high-bandwidth memory, among other core hardware, forms the physical foundation of this price hike wave. Although domestic cloud vendors have heavily invested in building a domestic computing power ecosystem over the past few years, capacity bottlenecks in high-end training and inference chips remain severe globally. Particularly, HBM memory, the "blood" of large models, has high technical barriers and long expansion cycles, becoming a key bottleneck restricting computing power release.
Currently, computing power is no longer merely about stacking servers but a sophisticated system composed of advanced process chips, high-speed interconnection networks, and high-bandwidth storage. The shortage of core hardware has significantly increased the marginal cost of computing power supply, and cloud vendors can no longer dilute costs through simple economies of scale. This rigid constraint on the supply side forces the industry to re-examine the pricing mechanism of computing power. When "computing power is power" becomes a consensus, vendors with stable, high-performance computing power supply capabilities naturally have stronger pricing power. The current price increase is actually a reasonable revaluation of the value of scarce hardware resources and an inevitable result of cost pressure transmission from the upstream industry chain to the downstream.
Recently, many of my friends in tech companies, especially CTOs, have been complaining that the prices of storage chips and servers are now unaffordable. The battle for computing power has instantly become a battle for costs, which is the most noteworthy aspect at present.
Third, the industry's pricing logic of "exchanging price for volume" has fundamentally changed. Looking back at the development of the cloud computing industry over the past decade, it's not hard to see a vicious cycle:恶性价格战 (malicious price wars). To compete for market share, major vendors have wielded the "price knife," not only squeezing competitors' survival space but also greatly compressing their own profit margins. At times, the price of cloud services even fell below operating costs, resulting in typical "involution." This "exchanging price for volume" model might have worked in the mobile internet era because marginal costs were接近零 (close to zero), and traffic monetization paths were clear.
However, the arrival of the AI era has completely broken this logic. Computing power is no longer a cheap commodity but an expensive specialized means of production. If low-price strategies continue, cloud vendors will be unable to cover the high costs of GPU procurement and power operations, let alone invest huge R&D funds for model iteration. Healthy industry development must be based on reasonable profits. Only when prices return to value can enterprises have the ability to reproduce and innovate.
The collective price increases by tech giants are actually a "collective rational return" for the industry. This marks the Chinese cloud computing market moving away from the "burning money subsidies" era and entering a value competition era centered on technical strength and service quality. This is extremely beneficial for building a良性生态 (healthy ecosystem) for the entire industry, shifting the focus of competition from "who is cheaper" to "who is more stable, who is smarter, who can solve problems better," which is undoubtedly a positive signal of industry升级 (upgrade).
Fourth, Token economics is emerging, and tiered pricing will become the norm. Once, data centers were seen as "warehouses" for storing data, with their value mainly体现在 (reflected in) space leasing and data保管 (custody). In the AI era, data centers have evolved into "factories" producing intelligence, with their core output being high-value Tokens. This role change directly催生 (gives rise to) a new pricing logic. Future AI services will no longer follow traditional annual/monthly subscriptions or pay-as-you-go models but will adopt tiered pricing based on dimensions such as Token throughput, response speed, and推理复杂度 (reasoning complexity).
This refined pricing strategy can more accurately match computing power需求 (demand) in different scenarios, allowing high-real-time, high-complexity tasks to pay higher premiums while offline batch processing tasks enjoy lower costs. This is not only an innovation in business models but also a great improvement in resource allocation efficiency. Through price leverage, the industry will guide computing power resources to areas that create the greatest social value, avoiding resource misallocation and waste.
In such a context, many companies even directly use the provision of Tokens as a new employee benefit. Of course, we have been discussing whether Tokens are means of production or employee compensation and benefits, but there is no doubt that in the current era, Tokens have become an important computing power bottleneck restricting the development of AI companies.甚至 (Even more), a friend of mine complained to me that at the beginning of the year, the tech giant where my friend works required all employees to use Lobster and their digital twins, but recently, due to massive consumption, they imposed流量限制 (流量限制).
Finally, faced with increasingly expensive computing power resources, what should be the future of AI Lobster? Can we still afford it?
This article is from the WeChat public account "Jianghan Vision Observation," author: Jianghan Vision Observation
