As a business form based on silicon-based chips, the trend of the cloud computing industry's ** decline is actually far from the scope limited by Moore's law in the broad sense, that is, a 50% decline every 18-24 months.
The reasons for this, the author has interviewed a number of industry insiders, which can be described as complicated.
However, Alibaba Cloud's price reduction horn sounded this time seems to have launched a new round of the industry's best war, but in fact, it may not only expand its own business scale, but also a new effort for the industry to pursue the best development paradigm.
It is not so much a strangulation between cloud vendors as a declaration of war between public clouds and private clouds.
The author presses. After clarifying the "AI-driven + public cloud first" strategy, on the last day of February, Alibaba Cloud sounded the horn of attack. The official announcement of the largest price reduction in history, covering the core product line, with an average reduction of 20%.
The average reduction is 20%, and the actual price reduction is much lower than this, and some products have a maximum reduction of 55%.
In all the previous ** battles, there was a problem - it would backstab old customers.
Therefore, the most extreme thing about Alibaba Cloud this time is that even if you bought it according to the old ** but did not consume the part before, it is also within the scope of discount.
This kind of strength and angle has dropped a bombshell in the cloud computing industry.
But what people didn't expect was that the industry followed up quickly, which was also surprising.
In the evening of the same day, JD Cloud released a message announcing the opening of the price comparison activity. JD Cloud said, "Drop it casually, compare it to the end!" We continue to compare prices across the network! Breakdown low price! another 10% lower", and promised to "buy expensive and pay for it".
In fact, not only JD Cloud will follow up, but I believe that the industry will also follow up.
There is a precedent for this, such as in 2023, after Alibaba Cloud launched a price cut, other vendors will follow suit. Some of Tencent Cloud's product lines have dropped by up to 40%, all mobile cloud products have dropped by up to 60%, and some of e Cloud's products have discounted as little as 1 percent.
Apparently not, for two reasons.
First, although the price reduction is very large, the author has learned from some industry players that in terms of the industry average, this price reduction has not broken through the cost bottom line, in other words, even if the price reduction or follow-up price reduction, it is not a loss-making business.
More importantly, it is forcing the industry to enter the positive rhythm of the "first flywheel".
The so-called flywheel is that enterprises obtain a larger market share and user scale through price reduction, and once the scale and share of the enterprise are expanded, the cost is diluted, and the more confident the price reduction, the more the price will be reduced.
Every business has its own starting point for the "flywheel", provided that a certain scale is reached. Therefore, whether it is Alibaba Cloud or the various enterprises that follow up. Its price reduction is not a price reduction for the sake of price reduction, but to keep up with the starting rhythm of the ** flywheel.
Second, the war will force technological progress.
The cloud computing industry is a typical resource-intensive + technology-intensive industry, but although there is still room for price reduction of pure IaaS layer resources (servers, chips, bandwidth), there is still an upper limit for transparency after all.
On the contrary, only through technological progress, better flexible application of resources, and finally the formation of benign cost controllability, can there be room for continuous price reduction.
It's like running a commercial real estate. Among them, the cost of building a building is not much different from everyone. But if you really want to make a profit, you have to pull up the occupancy rate of the house through the combination of long-term rental, daily rent or hourly rent, so that the vacancy rate is kept low to the extreme, so that you can outperform the market and outperform your opponents.
The way to reduce the vacancy rate in the cloud computing industry is mainly to improve the elasticity of computing power through technical means, and what is ultimately reflected in market competitiveness is the highest level of elastic computing power.
Therefore, no matter from the perspective of expanding the scale of the industry or forcing technological progress, the effect of Alibaba Cloud's price reduction this time is positive.
In recent years, with the emergence of the "one cloud and multiple cores" architecture of private clouds, as well as the development trend of hybrid multi-cloud and distributed cloud deployment, the largest proportion of enterprise private cloud storage market size is quite obvious.
But in fact, compared to the proportion of the more mature cloud computing market, this ratio has a lot of room for optimization.
Liu Weiguang, senior vice president of Alibaba Cloud Intelligence Group, said that China's cloud computing has developed for more than a decade, but the penetration rate of public cloud is still significantly lower than that of mature markets in Europe and the United States. There are 20 million servers in the Chinese market and about 21 million servers in the United States, but the computing power provided in the form of public clouds in the United States accounts for 60%, while in China it is only 28%.
After consulting with a number of industry insiders, they gave two strong reasons for the development of private cloud or hybrid cloud.
On the one hand, in addition to Internet companies, the early cloud users are basically large government and enterprise customers, who are highly sensitive to data security and have strong demand for private cloud or hybrid deployment.
On the other hand, the characteristics of this type of customer are, relatively speaking, ** not sensitive enough;
I admit that the above statement is very reasonable, but the core problem is still - at present, our public cloud is not cheap enough, the cost of private cloud is not far from that of public cloud, and the cost advantage of public cloud has not formed an absolute comparative advantage.
In addition, even for large customers, there may be problems where the cost of using the cloud is not accounted for.
The account is that the private cloud is more expensive and the public cloud is cheaper and cheaper, and the larger the scale, the more so.
Alibaba Cloud's price reduction and the follow-up of obvious competitors will have two great impacts on the game between public cloud and private cloud.
In fact, with the extension of the time for enterprises to use private clouds, the cost must be clear.
Perhaps the cost difference between private and public clouds is not so obvious when first deployed.
The real cost is reflected in the efficiency of using the cloud, and the disadvantages of private clouds are prominent in this regard:
First, resources such as computing power and storage are inelastic, so you can only buy more, not less. In fact, this loses one of the biggest meanings of cloud computing - resources can be elastically scaled, and what is the difference between such a cloud and IDC?
The elasticity provided by cloud computing essentially allows customers to adopt a combination of normalization and elasticity to meet the resource requirements of different periods of time. For example, if 1,000 servers are held offline, 500 servers will be held on the cloud and 300 servers will be used flexibly during peak hours.
Through flexible use, the most obvious effect is to improve resource utilization and reduce costs. For example, after the full-stack migration to Alibaba Cloud, the CPU utilization rate increased by 125%, but the total cost of ownership was reduced by more than 30%.
Second, the increasing development and management costs of private clouds will not decrease, but will only increase. In particular, cloud computing needs to develop a variety of applications and suites in order to give full play to its capabilities at the PaaS and SaaS layers.
For example, Alibaba Cloud's latest serverless products can support enterprises to achieve process development, and the cloud provides rich products such as containers, databases, data analysis, and data development governance, which can help enterprises apply new technology stacks at a faster pace. Himalaya uses Alibaba Cloud products to accelerate the speed from data insights to business decisions, with the average duration of offline tasks reduced by 42% and the duration of long-tail tasks reduced by 70%.
These development costs are very high for private clouds, while public clouds have a large number of users to dilute R&D costs, so various applications will emerge in an endless stream, and private clouds have no way to dilute the costs well;
Third, the elastic invocation of resources and the performance of the underlying hardware of the public cloud are based on the premise of the common use of tens of millions of users, and the underlying optimization capabilities honed at this scale can never be surpassed by the private cloud.
Therefore, Alibaba Cloud's price reduction is equivalent to calculating an account for the enterprise
When you use a private cloud, the cost includes the ever-expanding hardware cost + labor cost + management cost, as well as the shortcomings that cannot give full play to the full performance of the hardware;
When you use the public cloud, you can not only relatively avoid the overhead of these two aspects, but also automatically enjoy the dividend of the public cloud market's price reduction year by year - if Alibaba Cloud comes to do this once a year, how much can you save?
At a deeper level, the value of public cloud is not only that the infrastructure is cheaper than traditional IT, but also that it helps enterprises seize the dividend period of digital transformation.
Take left-handed technology as an example, they have gone through it - it takes at least a few months from planning to use of a self-built IDC, while using a public cloud can be used out of the box. Therefore, left-handed doctors quickly completed resource deployment based on Alibaba Cloud public cloud, and the project launch time was shortened by 67%.
Therefore, Alibaba Cloud's lead is to add an important weight to the game between public cloud and private cloud.
We don't want to eliminate private cloud or hybrid cloud, after all, the need for this does exist, and we need to adjust the ratio of public cloud and private cloud to a reasonable space through price reduction. At the same time, it also increases the global competitiveness of China's cloud computing.
The market competition of the public cloud is currently reflected in the form of price reduction, which is inevitable, but how to properly use the first war to the extreme, and at the same time avoid the negative effects of the first war, what is needed is "controllable nuclear fusion".
Nuclear fusion has long been achieved by mankind, it is nuclear, and war is nuclear. But what human beings are pursuing is "controlled nuclear fusion", which has both the power of nuclear fusion and is controllable in nature.
For the cloud computing market, every seemingly turbulent price reduction is behind the controllable scale and trend of the cost affordability + ** war, and this must ultimately be achieved through the technological competitiveness of enterprises.
Technology is the first thing for the controllability of the best war, and Alibaba Cloud has full-stack technology.
In the field of industrial Internet, "all steak" is a very magic word, which can be translated as "omnipotent".
The starting point of Alibaba Cloud comes from the determination to "go to IOE" more than ten years ago, that is, to get rid of the control of IBM, Oracle and EMC at the server, database and storage levels, and achieve full-stack self-development.
It is precisely because of this motivation that Alibaba Cloud has achieved one of the best technology leaders among domestic cloud computing vendors - taking the operating system as an example, the Feitian operating system, which was developed ten years ago, can bring stronger performance under the same specifications and resources with the cooperation of the CIPU architecture, and at the same time achieve extreme elasticity and greatly improve resource efficiency.
For another example, in order to replace expensive EMC storage, Alibaba Cloud's self-developed Pangu storage system greatly improves storage resource utilization through EC and compression technologies.
As another small example, in 2021, Pingtouge, a semiconductor company under Alibaba, released the self-developed cloud chip Yitian 710, which you can understand as a CPU designed for cloud computing.
In terms of hardware itself, the continuous optimization of self-developed chips such as Yitian 710 can increase the cost performance of databases, big data, AI, high-performance computing, and first-class codecs by more than 80%.
From the perspective of reliability and sustainability, data centers and cloud services are the core commercial scenarios of Yitian 710, and the current mainstream is that since AWS has continuously increased the proportion of self-developed ARM architecture processors GR**ITON in order to reduce the EC2 cost of its basic cloud computing services, powerful cloud computing manufacturers are pushing their own chips, and Yitian is one of the leaders.
Based on this full-stack self-developed concept that has been promoted since ten years ago, Alibaba Cloud has been able to achieve large-scale pooling of cloud computing resources today, and then multi-tenant cutting, which can achieve extremely high resource utilization.
You can imagine how much absolute advantage does the latter have in terms of hardware, network and other resource procurement costs based on 10,000 and 1 million?
In addition, another important confidence of Alibaba Cloud is that it has enough customers to share the above resources, and the more customers, the lower the dilution cost.
Alibaba Cloud is ahead of other cloud computing competitors in China, and its biggest advantage is that it has seized the window period of development at the earliest, and after having a certain scale, it has eaten the initial dividends, and then launched the infinite game of scale-cost-scale.
Therefore, anyone can participate in the first battle, and they will definitely participate, but the final winner must be a player with both a comparative advantage + absolute technical advantage + a larger and larger scale, which is also the purpose of industry competition - to screen the most powerful players through the first battle——— and then continue to output the competitiveness of the public cloud, and ultimately ensure that the industry develops in accordance with a healthier market structure.
Cloud computing is entering the deep water area of popularization, and the difficulty is increasing, and it is positively correlated with the digital level of the enabling objects.
Therefore, the sharp price reduction of cloud computing will not only bring about the reduction of existing users, but will help accelerate the replacement of traditional fields that are very under-digitalized, and will benefit the operational efficiency of the entire economic system.