AI revenue tripled! What did Alibaba Cloud do behind the scenes?

Author: Zhang Yifan

Source: AI Hard

AI Revenue Surges by Three Digits Year-on-Year?!

At the May 14th earnings conference, Alibaba revealed that in Q1 24, Alibaba's AI-related products achieved a strong growth in revenue by three digits year-on-year.

Group CEO Eddie Wu stated that the growth of the AI business mainly comes from two aspects:

• The demand for model training and inference services from basic model companies and internet companies;

• Growth in demand for vertical applications in industries such as finance and automotive.

Regarding future expectations, the management believes that with the company's strong layout in AI, Alibaba Cloud's commercial revenue will resume double-digit growth in the second half of the fiscal year.

Looking at the cash flow changes in the financial report, Alibaba Cloud remains the company's key capital investment direction in the first quarter.

Why can Alibaba Cloud's AI business achieve such strong growth, and how will Alibaba Cloud's AI strategy proceed? This article attempts to explore this through the business changes of the company in the past six months.

I. Alibaba's AI Layout

In September last year, after Eddie Wu took over as the new CEO of Alibaba Group, he proposed two strategic focuses: "User First, AI-Driven." In the November conference call, Eddie Wu officially clarified three directions: technology-driven internet platform business, AI-driven technology business, and globalized business network.

For Alibaba Cloud, after two months of restructuring, Eddie Wu confirmed that Alibaba Cloud will implement an AI-driven, public cloud-first strategy in the next five years.

As a result, the company subsequently expanded its layout in the AI field more comprehensively, with a clearer strategy. Everything is in line with the growth of public cloud services. Whether it is the original open-source model base, the TongYi series of AI applications, the model community, or further expanding external investments to enlarge Alibaba's "AI ecosystem circle," all contribute to the core of Alibaba Cloud - driving traffic to public cloud services.

AI Hard statistics show the current main AI product ecosystem of Alibaba Cloud:

II. AI Cloud Services

Undoubtedly, Alibaba's public cloud has become the core of the cloud sector, with all businesses revolving around this topic. Focusing on establishing competitive advantages in technology and scale, and reducing the prices of global public cloud products, has become Alibaba Cloud's strength.

Further breakdown, the advantages of AI cloud services are reflected in computing power, communication, model libraries, and other elements.

1. Computing Power/Network

As a leading cloud service provider, Alibaba Cloud has inherent advantages in computing power and network resources. Leveraging its cloud platform, Alibaba has built sufficient AI infrastructure (including computing power and communication facilities) and launched the cloud development platform ModelScope, allowing developers to easily develop and deploy various AI models in the cloud In early April this year, Joseph Tsai, Chairman of the Board of Alibaba, pointed out in an interview that the current computing power reserve is sufficient to support 18 months of large-scale language model training needs. In addition, the company has corresponding deployments in the "inference" phase to support the operation of the models.

This was confirmed by the cash flow consumption of the company in the current quarter.

1. Communication Network

Alibaba's HPN 7.0 network architecture is currently the highest standard for Ethernet AI clusters. It uses Ethernet RDMA low-latency technology, supporting cluster scalability of up to 100,000 cards, achieving fast response and low latency.

On May 14th, Alibaba Cloud's AI high-performance network architecture HPN 7.0 research paper was included in SIGCOMM2024, becoming the first paper in the history of SIGCOMM on AI intelligent cluster network architecture. SIGCOMM is the world's most authoritative computer communication network conference. Previously, Google's Jupiter network was selected by SIGCOMM and developed into a classic architecture. Some experts point out that Alibaba Cloud's HPN 7.0 is expected to become a new paradigm for the next generation of AI high-performance network architecture.

Recently, the large-scale model of Alibaba Cloud's Thousand Questions 2.5 version is based on the high-performance network cluster built on HPN 7.0.

1. AI Cloud Development Platform

Alibaba launched the cloud development platform ModelScope, providing developers with a wide range of model libraries, allowing developers to easily develop and deploy various AI models in the cloud.

III. Model Base

While providing cloud computing services for large model customers, Alibaba Cloud is also developing its own open-source large models. However, this is not to seize the customer market, but to help industry technology grow faster through Alibaba Cloud's technical open source.

As of now, Alibaba's Thousand Questions large model has iterated to version 2.5. Recently, the latest evaluation results released by authoritative institutions show that Alibaba Cloud's Thousand Questions 2.5 has matched the performance of GPT-4 Turbo in Chinese.

The release of the Thousand Questions model is, on the one hand, to establish the company's AI ecosystem (adopting an open-source form), help the industry develop faster, expand the market, and increase the demand for AI cloud computing power. On the other hand, it is also to promote the development of AI applications. Based on this large model, developers can quickly build AI applications suitable for different scenarios, such as intelligent customer service, image recognition, natural language processing, etc., increase internal and C-end user growth, and thereby increase the demand for cloud computing services.

The rapid development of these two aspects can quickly drive the growth of Alibaba Cloud's business

Four, Model Application

Considering the company's AI-driven strategy, models suitable for various different scenarios/businesses are needed to support.

Building on large models, Alibaba Cloud has launched a series of vertical industry applications based on multimodal.

According to the official website, the Tongyi series has launched dialogue AI application Tongyi Qianwen, speech-to-text AI application Tongyi Tingwu, and image generation AI application Tongyi Wanxiang, among others.

In addition, there are vertical applications tailored to specific needs in different business scenarios, including Tongyi Farui (legal), Tongyi Lingma (programming), Tongyi Xiaomi (customer service), and more.

The increasing adoption of AI-driven businesses also means creating more incremental demand for cloud computing.

At last year's performance meeting, Alibaba revealed that through AI capabilities, the company's search and advertising businesses have significant room for improvement. However, the company also mentioned that this functionality is still in the early stages of research and testing. But they believe that with further research and investment, this feature will provide very positive assistance to the future search and advertising of Taotian.

Furthermore, for customers looking to customize AI applications, Alibaba has launched the Bailian Big Model Platform. Developers can develop a large model application in 5 minutes through "drag and drop" and "refine" an exclusive model in a few hours.

Five, External Investment, Establishing Alibaba AI Ecosystem

After Eddie Wu proposed the strategy of prioritizing public cloud, Alibaba's investment in AI large models has significantly accelerated. Currently, all five major domestic large model unicorns have received investments from Alibaba without exception.

From investing in Zhipu AI in September 2023 to completing a new round of investment in MiniMax in March this year, Alibaba has completed investments in China's five leading large model unicorns (Yuezhi Anmian, MiniMax, Zhipu, Baichuan, Zero One Wanwu). These leading large model manufacturers are gradually increasing their influence in the B-end and C-end markets, providing a huge increment for Alibaba's AI cloud services, becoming an important source of triple-digit growth in AI business.

From the strategic perspective of the company's AI-driven and public cloud prioritization strategies, such investments are crucial at present.

Large model manufacturers need stable and cost-effective AI infrastructure to train and infer models. As a leading domestic AI cloud service provider, Alibaba can provide this service of high quality.

This was also confirmed in this quarter's financial report, as Alibaba reduced the prices of its global public cloud products this quarter due to establishing competitive advantages in technology and scale. This further drives the development of the AI industry At the same time, Alibaba seizes the initiative through investments and services, leading major AI model companies to deploy their businesses on Alibaba Cloud. With the rapid growth of users of large AI models in the later stage, the demand for AI cloud computing also increases, thereby driving the growth of Alibaba Cloud's core business.

Alibaba's investment in AI large model companies becomes the capital expenditure of these companies, which then flows back to Alibaba Cloud, forming a closed-loop business.

This investment approach is also seen in similar cases with overseas tech giants, such as Amazon's investment in Anthropic, which is based on a similar logic. Furthermore, using cloud computing power as capital to invest in large model manufacturers strengthens the connection between cloud computing companies and large model manufacturers, while also alleviating cash flow pressure for both parties.

VI. What about the future of Alibaba Cloud?

Currently, Alibaba Cloud's "AI-driven, public cloud-first strategy" is already showing results, as Alibaba Cloud has chosen a focused path. However, for future growth, the most important factor still relies on the significant development of the generative AI industry, and the certainty of this is continuously increasing.

Just as Alibaba predicts in its financial reports: "Alibaba Cloud is expected to return to double-digit growth in the second half of the year."