Image source: @VisualChina
Text | Liu Kuang
After the "downturn" of the past year, cloud service providers will rekindle their fighting spirit in 2023 due to the outbreak of AI large models. Standing at the current time node, cloud service providers are also rethinking how to occupy a greater advantage in the new situation, so a new war around "technology competition and market competition" has begun.
Following the announcement of the launch of the "largest price reduction in history" by Alibaba Cloud on April 26, Tencent Cloud and Mobile Cloud have also followed suit and announced their participation in this "price war". It is not difficult to foresee that this "price war" provoked by Alibaba Cloud has successfully set off a chain reaction in the industry.
The industry consensus behind cloud service price reductions
As the leading industry leader in China, Alibaba Cloud's every move has always set off a lot of ripples in the industry, and this price reduction is naturally no exception. On May 16, Tencent Cloud and Mobile Cloud successively announced price reductions, officially raising the volume of this wave of "price wars".
Among them, Tencent Cloud announced price reductions for a number of core cloud products, with some product lines reducing by up to 40%, and the price reduction policy will officially take effect on June 1. Specifically, for cloud servers in some regions, the price of Tencent Cloud will be reduced by up to 40%; In terms of cloud networks, Tencent Cloud reduced the LCU fee for CLB by 18% and the standard NAT gateway by 15%. In terms of databases, Tencent Cloud Database (TDSQL-C serverless) released a new resource package payment model, which reduced the publication price by up to 25% compared with the same specification subscription products. On the evening of the day Tencent Cloud announced the price reduction, Mobile Cloud also announced a price reduction for a number of cloud products, with some product lines reducing by up to 60%, and the price reduction policy will officially take effect on May 17.
As various cloud service vendors have shown the weapon of "price reduction", domestic cloud service giants are further deepening around the market competition, and a new offensive and defensive war has slowly begun.
First of all, as the leading boss, Alibaba Cloud's "active price reduction" can effectively transmit pressure, force other manufacturers to follow suit, and make the market competition between each other increasingly fierce. According to data provided by Canalys, Alibaba Cloud still ranked first in 2022, with a year-on-year increase of 7%; HUAWEI CLOUD ranked second with a market share of 19%, with an annual growth of 13%, leading the overall market growth. Tencent Cloud ranked third with a market share of 16%; Baidu Intelligent Cloud, which ranked low, increased by 11% year-on-year, ranking ahead of Alibaba Cloud.
Due to the leading scale, although the price of some products of Alibaba Cloud was higher than that of Tencent Cloud and HUAWEI CLOUD, there were still "customers" willing to pay for it, but the situation of "bottom" growth rate was still difficult to ignore. This time, Alibaba Cloud took the initiative to "vigorously reduce prices", one can level the price gap between it and other "friends", and the other can expand its market advantages. From the perspective of Tencent Cloud, Mobile Cloud and other manufacturers, because they have not yet achieved profitability, they are under more pressure, but if they do not reduce prices, they may face the unfavorable situation of losing market share, so they can only choose to follow up.
Secondly, with the gradual popularization and application of the cloud computing market, cloud vendors hope to attract more users to enter the cloud computing field, and the price reduction can just achieve this goal. After years of development, the penetration rate of cloud services into various industries has increased, and the industry has increasingly entered the era of "deep cloud use". In this context, through large-scale price reduction, one can improve the stickiness of existing cloud service customers and increase the renewal rate of cloud service vendors; Second, it can lower the threshold for cloud migration in various industries and increase the enthusiasm of customers in various industries to use the cloud.
From this point of view, although Alibaba Cloud took the lead in provoking the "war", other cloud service vendors chose to respond positively, not only out of defense, but also for the consideration of following the trend and seeking new space. Especially considering the decline in the growth rate of the entire industry last year, if this situation wants to continue to maintain high growth, there needs to be a certain "stimulus" measure, based on this, there is a consensus in the industry on cloud service price reduction.
Cloud service vendors accelerate transformation
In fact, whether it is to B or to C, price is always the most important means of competition, in the middle and late stage of the industry's development, this means can also be used as a "strategic weapon" to help enterprises establish a new internal mechanism as soon as possible.
First, through active price reduction, we can force internal business change to accelerate, thus forming a new development path driven by scale and efficiency. From the perspective of the entire cloud service industry, the shift from IaaS to PaaS and SaaS is the general trend, but so far domestic IaaS and PaaS spending is still the mainstream, PaaS and SaaS expenditure is still relatively small, and overseas Amazon Cloud, Google Cloud and Microsoft Cloud, etc., have achieved PaaS and SaaS-based cloud revenue sources, so the overall profitability is very strong.
In this context, the cloud giants represented by Alibaba Cloud and Tencent Cloud take the initiative to "reduce prices", which will inevitably further reduce their overall gross profit margins and force the internal transformation to PaaS and SaaS with higher gross profit margins, while IaaS businesses with more serious homogeneity and lower gross profit margins will further move towards internal self-research to seek breakthroughs with differentiation. At the same time, price reductions can also accelerate the scale of cloud business and drive the scale of cloud vendors to increase, thereby further strengthening the scale effect and driving cloud services to scale and efficiency.
Second, behind the cloud giant's active price reduction, it is also an active concession of its role from "integrated" to "integrated", and its strategic value is more prominent. With the role of cloud service vendors, from "integrated" to "integrated", the role of industry partners has become increasingly important in the ecosystem of cloud service vendors. Cloud service vendors actively reduce prices, which is essentially cloud computing vendors actively giving up the pricing power of some projects and giving benefits to ecological partners to facilitate "integration", so as to expand their industrial influence.
From this point of view, whether it is Alibaba Cloud, Tencent Cloud, or Mobile Cloud, the price reduction is to facilitate "integration", accelerate industry penetration, and actively promote the strategic move to the role of "being integrated".
The new variable affects the weight increase
It is worth noting that behind the "price reduction" of many cloud service vendors, the industry changes brought about by the AI large model are also an important technical consideration.
First of all, the expansion of computing power driven by the development of AI big models may drive a new round of cloud service market growth. According to research data from investment agency A16Z, almost everything in generative AI will be through a cloud-hosted GPU (or TPU) at some point. Whether it's a model provider and research lab training a model, a hosting company performing inference and fine-tuning tasks, or an application company doing both, floating-point operations per second (FLOPS) is the lifeblood of generative AI.
A16Z estimates that companies with future generative AI applications will spend an average of about 20%-40% of their annual revenue on inference and customized fine-tuning. This is typically paid directly to the cloud service provider for instances or to a third-party model provider, who invests about half of the revenue in cloud infrastructure. According to this calculation, 10%-20% of the revenue of generative AI will go directly to cloud service providers.
From the current point of view, this predictable performance growth is undoubtedly a growth opportunity worth long-term bet on for many cloud service vendors whose growth has stalled in the past two years. After all, compared with the "uncertainty" of AI large models, the core value of cloud service vendors as "shovel sellers" as computing power providers is relatively more certain (after all, no matter what the results of AI training large models are, computing power is always just needed).
Second, the convergence of AI technology and the cloud is accelerating, and model-as-a-service (MaaS) will become an important prerequisite for this. As the industry enters the era of "deep cloud use", AI capabilities based on cloud computing have gradually been verified by enterprises and industries, and the combination of AI and cloud can lower the threshold for enterprises to go to the cloud, and at the same time play the value of data with AI's deep learning capabilities to build a better "cloud" path for enterprises.
In short, with the maturity of AI large model technology, the innovation threshold of the entire AI application in the future is lower (lower cost, faster iteration), and start-ups can derive new technologies based on large model innovation instead of underlying chip layer and framework layer innovation. At the same time, the large model that focuses on applications has gradually attracted more attention from more enterprises and has become an important consideration criterion for enterprises when choosing cloud services. This is also the core reason why many manufacturers, including BATH, have released AI large models in a "hurry".
After all, for many cloud service vendors, it is not important whether the current product is mature enough. It is important for cloud service vendors to demonstrate to downstream vendors that they have the ability to change technology and launch new solutions. This important "background board" is also another technical consideration for Alibaba Cloud, Tencent Cloud, Mobile Cloud, etc., who have announced "price reductions".
Intelligent inclusion becomes a new focus of cloud services
As major cloud service vendors embrace the AI model, a new cloud era with intelligent cloud as the main feature and deep inclusiveness as the core feature is slowly unfolding.
On the one hand, the combination of cloud services and AI models is further catalyzing the business value of intelligent clouds. Previously, cloud service providers only needed to charge customers according to the rental of quantitative computing power and storage, but in the AI era, in addition to computing power and storage, customers also need AI general capabilities. Therefore, cloud service providers need a completely different business plan than before. For example, Baidu's enterprise-level "Wenxin Qianfan" large model platform pays according to the total number of words called input and output, and the charging standard is 0.012 yuan/1000 tokens.
Coincidentally, Zhou Jingren, CTO of Alibaba Cloud Intelligent Group, also said that in the future, Alibaba Cloud will launch a variety of parameter versions of Qianwen big models, customers can choose training services of different model scales according to their needs, and charge according to tokens, "exclusive large model cloud inference can choose the postpaid mode of billing according to token metering, or the prepaid mode of account permissions + resource packages." It can be seen that as the commercial value of AI models is increasingly seen, cloud service vendors, including Alibaba Cloud, have become clearer about the application of intelligent technologies and the adjustment of new business models, and this style may also be used as a reference for Tencent Cloud in the future.
On the other hand, with the combination of AI big models and cloud services, AI general technology inclusion will open up space for cloud technology inclusion. To some extent, the reason why the industry attaches so much importance to the breaking of the AI model is that it has a greater influence on ordinary consumers on the C-end, which represents a big step towards inclusive application of AI, which is also the real meaning of the so-called "iPhone moment" in the industry.
From the perspective of cloud services, it is mainly for the industry to do digital transformation, its final service customers or to face C-end consumers, AI large model is undoubtedly a good tool to reach C-end users, it provides more possibilities for cloud services to reach C-end users, making cloud services more concrete.
With the successive price reductions of Alibaba Cloud, Tencent Cloud, and Mobile Cloud, this round of "inclusive" battle related to cloud services will also officially kick off. Whether other cloud vendors that have not yet announced their product policies will follow suit is unknown, but there is no doubt that other cloud vendors will never remain indifferent.