Visual China.
Text |New knowledge of science and technology.There is no minimum, only lower. In 2024, the main line of China's cloud market will still be low prices.
On the last day of February, Alibaba Cloud announced the largest price reduction in history, covering more than 100 products such as computing, storage, networking, and databases, and more than 500 product specifications, with an average reduction of 20% on the official website of cloud services**, up to 55%. Immediately afterwards, JD Cloud announced on the same day that it would continue to increase the activity of "price comparison on the whole network, and compensation if you buy expensive".
Tencent Cloud announced on March 5 that it would launch a series of AI products** activities, including speech recognition, speech synthesis, AI painting, and digital intelligent humans, with new users purchasing a minimum of 04% off. Many industry insiders said that the first battle in China's cloud market, which is the same as in 2023, is restarting.
At the Alibaba Cloud Partner Conference in April last year, Alibaba Cloud also announced the "largest price reduction in history", and the preferential efforts at that time mainly included a 15% to 50% reduction in core products** across the board, and a maximum reduction of 50% for storage products. After that, Tencent Cloud, JD Cloud, Huawei Cloud, and the three major operator clouds have all followed up the battle to varying degrees.
As the main promoter of two consecutive first-class battles in less than a year, Alibaba Cloud said: "Price reduction is not a short-term market competition behavior, but a long-term strategic choice, which is determined by the business model of public cloud." "Exchanging price for volume can promote manufacturers to further expand their market share and achieve economies of scale.
The same ** war has also been staged in the US market. In 2014, at an event in San Francisco, Google, USA, announced a series of significant price cuts for its services that run software applications and** and store large amounts of data. A day later, Amazon's AWS announced that it would follow suit, with major cloud service products being 30%-40% cheaper and up to 61% lower.
The first battle in the Chinese market is often initiated by the industry boss, which belongs to the defense battle to consolidate the market position; In the United States, the initiator of the price reduction is often the challenger who covets the throne, and the ** battle is a kicking game for novices to enter the game.
Behind the similarities is the different self-confidence of the two cloud market leaders.
AI + cloud computing, what will happen?
In last year's large-scale model boom, CEO Robin Li said at the earnings conference: "Wenxin Yiyan will be a game changer that changes cloud computing. In his view, in the past, cloud computing mainly sold computing power, and in the future, the AI large model will subvert the entire cloud computing market, and the combination of large models and cloud computing will bring opportunities for cloud market players to reshuffle.
There is nothing wrong with this judgment. However, due to various factors, this change is not the first to occur in the Chinese market, but is being staged in the United States.
According to global cloud computing data released by market research agency Canalys, in the fourth quarter of 2023, the three major cloud service providers AWS, Microsoft Azure and Google Cloud together grew by 21%, accounting for 66% of total spending. Among them, Microsoft Azure and Google Cloud revenue growth recovered strongly, both exceeding 25% again.
Microsoft leads the market by a significant margin with a 30% growth rate and continues to close the market share gap with AWS. AWS, the leader in the global cloud market, has seen growth rise from previous quarters, but its 13% year-over-year growth still lags behind the trajectories of Microsoft Azure and Google Cloud.
Cloud computing is the first soil for the commercialization of large models, and AI+ cloud computing is becoming the main force driving the U.S. cloud market.
Over the past year, Microsoft has launched Copilot for Microsoft 365, embedding its generative AI platform into Word, Excel, and other office applications. Google, on the other hand, has introduced Gemini's large language model into the Workspace app.
In Microsoft's earnings report disclosed in early February, Microsoft's revenue from Azure and other cloud services increased by 30% in Q2 of fiscal year 2024. Among them, AI contributed 6% growth to the Azure cloud, which doubled from the previous quarter, and the vast majority of the increase came from OpenAI's inference calls on the Azure cloud.
In addition, Microsoft has added support for OpenAI's latest models in Azure operations, including GPT-4 Turbo, GPT-4 with Vision, DALL-E3, etc. Yi Zhang, an analyst at Canalys, said: "AWS is progressing more slowly in AI than its main competitors, which may explain why it's not growing as fast as Azure. ”
Also slower than its competitors is Alibaba Cloud. According to the latest data from IDC, similar to the fate of the global cloud giant AWS, Alibaba Cloud won 26The 7% share ranks first in the market, but the share is **5 year-on-year8%；Ranked first.
The second, third and fourth Huawei Cloud, e Cloud and Mobile Cloud have increased their share.
According to the report "China Public Cloud Service Market (First Half of 2023) Tracking", the domestic IaaS + PaaS market will increase by 15% year-on-year in the first half of 20239%, a new low year-on-year growth rate in the past three years.
As a leader in the domestic market, Alibaba Cloud cannot grow against the trend without the general environment. One is in the government and enterprise market, where the three major operator clouds have become the new major leaders; The other is that the dividends of the domestic Internet wave have been exhausted, with small, medium and micro start-ups reducing costs, and large enterprises moving towards self-development.
In 2021, under the influence of geopolitics, TikTok no longer uses cloud services from domestic manufacturers, and transferred orders worth about $800 million per year belonging to Alibaba Cloud to AWS and Oracle. This change directly led to Alibaba Cloud's revenue growth in the first quarter of that year hitting a record low.
With the development of Byte's own cloud and AI product Volcano Cloud, important customers of Alibaba Cloud such as Douyin are also returning to Byte's internal digestion. The sluggish government and enterprise market, the loss of leading Internet customers, and the search for new outlets and cloud computing markets under the dual pressure have become the key to Alibaba Cloud's return to first-line growth.
An industry insider said: "It may be that seeing the impact of AI on the US cloud market pattern, Alibaba Cloud launched a price cut again in less than a year, which is the scale effect determined by the business model from the perspective of long-term strategy, but from the perspective of short-term dividends, it is also to attract more small, medium and micro AI start-ups with lower ** in the AI entrepreneurship tide." ”
Looking for new increments in the domestic cloud market, cloud computing dividends under the AI tide are obviously not to be missed.
In the past year, Alibaba Cloud first clarified the strategy of "AI-driven, public cloud first" around the AI tide in China, and then carried out drastic organizational adjustments, establishing a dedicated public cloud business division for the first time, with the goal of giving priority to scale and expanding market share, and rapidly promoted the implementation.
At present, Alibaba Cloud has no more innovative way to break through the new strategy, so it has to come up with the killer feature of "** war" again.
However, the cloud market under the AI tide may be far from Alibaba Cloud's comfort zone. According to the Economic Observer, Alibaba Cloud said that the feedback was very good, but it did not meet the expected goal.
From the perspective of the U.S. cloud market, in the face of the reshuffle moment of declining growth rate in AI competition, for AWS, which is also a leader, in the context of product competition driven by technological innovation, the first-class war is also becoming invalid.
According to the latest data from market research firm Synergy Research Group, global enterprise spending on cloud infrastructure services in the fourth quarter of 2023 was close to $74 billion, an increase of more than $12 billion from the fourth quarter of 2022, and a year-on-year growth rate of 20% in the fourth quarter, significantly higher than the same period last year.
Gartner, another market research firm, predicts that global IT spending will reach 5$1 trillion, up 8% from 2023 levels, with more IT spending coming primarily from cloud inflation and generative AI-fueled pricing of cloud service offerings**.
Economic data and business research from the United States show that cloud costs are on the rise. The January Producer** Index (PPI) report released by the U.S. Bureau of Labor Statistics showed that data processing and related services, including cloud computing, grew by 06%, a year-on-year increase of 37%。
Industry observers say the increase in labor costs has led to the rise of the cloud. Another factor is the rise of generative AI, which cloud providers are embedding into their offerings, which is already starting to impact the pricing of cloud service offerings.
For example, Salesforce announced an average list price of 9% for selling cloud, service cloud, and other products**. This is the company's first product price increase in seven years, and among the thousands of new product and service features it offers, "generative AI innovation" has become a central highlight.
In addition, ServiceNow plans to introduce generative AI capabilities in advanced versions of its IT service management, customer service management, and HR service delivery offerings. CJ Desai, president and chief operating officer of ServiceNow, said at its July earnings conference** that the products are expected to be 60% higher than the previous Pro version.
The increase in pricing comes from cloud service providers believing that customers who adopt them will get more value from more advanced AI products and services. Sid Nag, a VP analyst at Gartner, said the pricing adjustment was to be expected, and the cost of interacting with generative AI apps like ChatGPT could be about 10 times higher than standard keyword searches.
Clouds are getting more expensive, and it's also happening in Europe.
According to Canalys, the cost of public cloud services in the U.S. and Europe will be due to energy costs and other factors, with one-fifth for Americans and nearly one-third for Europe. Starting in early April last year, Microsoft increased its Microsoft cloud offerings in Europe by 9% to 15%, and Google also increased some of its European offerings in March.
In the moment of generative AI reshuffle, it is not low-cost cloud services that gain market share and revenue, but cloud vendors that can provide more advanced AI productivity products. This also leads to the fact that in the context of competition driven by technological innovation, AWS can only bet on technology product research and development, rather than fighting the best war.
Guided by this strategy, AWS's parent company, Amazon, previously invested $4 billion in Anthropic and held a portion of its equity. In return, Anthropic is committed to providing access to its future foundational model to AWS customers worldwide through Amazon Bedrock.
Recently, Swami Sivasubramanian, vice president of AWS, announced in a blog post that Anthropic's newly released Claude 3 series models will be available on Amazon Bedrock, making it easy for customers to build with the latest models.
Amazon Bedrock is a fully managed service released by AWS in April 2023, which allows enterprise users to access Amazon's self-developed and many third-party large models through APIs and other means.
According to the results of the current public test in the market, the release of the Claude 3 series model, which claims to be ahead of GPT-4 in an all-round way, may also be the beginning of AWS's counterattack against Microsoft Azure.
Compared with AWS, which may have to take advantage of Claude3, Alibaba Cloud is still in the incubation stage of AI + cloud computing.
The day before the announcement of the biggest price cut in history at the end of February, Alibaba Group's Institute of Intelligent Computing launched a new AI-audio-model technology (eMo), which officially called it "an expressive audio-driven portrait generation framework."
You only need to provide a ** and an arbitrary audio file, and the emo can generate AI that can talk and sing, as well as a dynamic small ** that realizes seamless docking, and the maximum time can be about 1 minute and 30 seconds.
This is the only benchmark product that has been launched by major domestic Internet manufacturers after OpenAI launched SORA to detonate the network. Compared with the trend of chatgpt in 2023, I have eaten the first crab in the domestic market. Ali in 2024 wants to eat the first crab of Wensheng's ** model.
In terms of general large models, Ali also shot continuously. A month ago, Alibaba led a new round of funding of $1 billion for Moonshot AI, a large domestic AI model team, valuing the company at $2.5 billion, making it the largest single round of funding for an AI startup in China.
Recently, he led the new large-scale financing of minimax, a general large-scale start-up project. Last year, Alibaba also invested in a number of AI industry chain companies such as Baichuan Intelligence and Zhipu AI, and continued to bet on this round of AI boom.
In addition to group-level investment, Alibaba Cloud itself has also launched a number of cloud service-related products around AI in the past year, including a number of AI large model products that are benchmarked against OpenAI, such as Tongyi Qianwen and Tongyi Wanxiang, and announced that it will access all of Alibaba's products for comprehensive upgrades.
Today, Alibaba Cloud has used large models to transform products such as DingTalk and Tmall Genie. In addition, product lines such as DingTalk knowledge base, brain map, whiteboard, teambition, and mailbox have been connected to large models to realize multiple scenario applications. However, at present, AI capabilities are still in the early stage, and they cannot support Alibaba Cloud's technical confidence.
For Alibaba Cloud, the AI model is a speed-increasing disk. The establishment of the speed increase disk is based on the maintenance of the public cloud basic disk. This also means that before a major breakthrough in AI capabilities, Alibaba Cloud still needs to wield the killer weapon to defend the existing stock market.
In addition to technical confidence, whether or not they can enjoy the dividends of the AI market depends on the business structure of cloud service providers. Generally speaking, the main business of cloud vendors can be divided into three parts: IaaS, PaaS, and SaaS.
Taking the European and American markets, which are the first to enjoy the dividends of AI cloud computing, as an example, the pricing increase of core cloud vendors is mainly concentrated in the SaaS layer, and the representative enterprise Salesforce is a platform enterprise with SaaS products as the core. Microsoft Azure is a balanced development of three businesses, which can not only provide computing power to OpenAI, a large model provider, but also eat the biggest cake with the help of office SaaS products.
Compared with the SaaS maturity of the global market, in the domestic market environment, IaaS in software and services is much larger than PaaS and SaaS, while Alibaba Cloud is mainly IaaS-based. Therefore, if you want to enjoy AI dividends, how to change the revenue structure is also the second problem for Alibaba Cloud.
In the first quarter of Alibaba's 2022 fiscal year, DingTalk's performance was merged into Alibaba Cloud for the first time, but Alibaba Cloud's revenue declined quarter-on-quarter in the current quarter, and DingTalk did not have a direct effect on Alibaba Cloud's revenue for the time being. In March 2022, DingTalk began to clarify its PaaS strategy, and formulated three commercialization paths of customized DingTalk, paid subscription, and commission extraction.
Subsequently, Yunding was split, and some market participants speculated that it was to prepare for the listing of Alibaba Cloud. However, under the AI tuyere, Alibaba Cloud, which is affected by various factors, is now suspended from listing. In addition to the re-launched battle, what changes will be ushered in in the technical strength and business structure are the key answers to whether Alibaba Cloud can achieve a new journey in 2024.