Why is Alibaba raising $10.2 billion now?
Alibaba is raising money because its AI plans extend far beyond building a model. The company announced an 80 billion Hong Kong dollar share placement, worth $10.2 billion, with every dollar of net proceeds earmarked for “full-stack AI capabilities.” That includes chips, computing infrastructure, and model development.
The timing matters because these parts depend on one another. Better models need more computing power. More computing power depends on access to chips and infrastructure. Funding only the model would leave Alibaba dependent on the systems underneath it. This share sale gives the company capital to develop the stack as a connected whole.
The scale of the deal also signals how expensive that effort can be. Alibaba is selling 710 million shares at HK$112.70 each, a 3.6% discount to Friday’s closing price. It is the largest-ever primary follow-on offering by a Hong Kong-listed company, and the third-largest share sale globally in 2026, behind Alphabet’s $80 billion raise and Intel’s $15 billion sale.
That does not prove Alibaba will win the AI market. It does show that the company is treating AI infrastructure as a major, long-term capital requirement rather than a small addition to its existing business. The offering was structured offshore of US securities registration, with Morgan Stanley, HSBC, UBS, and CICC serving as joint bookrunners.
The main takeaway is simple: Alibaba is raising $10.2 billion now because its AI strategy requires ownership and investment across chips, computing, and models. The model is only one layer of the bill.
What makes this Hong Kong’s largest follow-on share sale?
Alibaba’s planned HK$80 billion, or $10.2 billion, share sale is not an initial public offering. Alibaba is already listed. Instead, it is issuing 710 million new ordinary shares to raise fresh money from investors. That makes it a primary follow-on offering, because the proceeds go to the company rather than to existing shareholders selling their holdings.
The size is what sets the transaction apart. If completed as planned, it would be the largest primary follow-on offering ever undertaken by a Hong Kong-listed company. It would also be the world’s third-largest primary follow-on share sale this year, behind offerings from Alphabet and Intel.
Alibaba plans to sell the shares at HK$112.70 each, a 3.6% discount to the stock’s latest closing price. That discount gives investors a financial reason to participate, while the new shares give Alibaba a large pool of capital for its AI plans. The company said all net proceeds would support its “full stack” AI capabilities, including semiconductors, computing infrastructure, and the development and deployment of AI models. It has not provided a breakdown of how much will go to each area.
Investor demand was strong enough for Alibaba to increase the offering after it became oversubscribed. Sovereign wealth funds were among the interested investors, according to people familiar with the transaction. Morgan Stanley, HSBC, UBS, and China International Capital Corp. are acting as joint bookrunners.
The important distinction is simple: this is not just a large stock trade. It is a company raising new money at unusual scale, with investors backing an expansion of Alibaba’s entire AI infrastructure stack.
Where will the HK$80 billion actually go?
Alibaba has attached a clear purpose to the HK$80 billion share sale: all proceeds will be invested in its full-stack AI capabilities. That means the money is not being raised for a general corporate fund, at least according to the announcement. It is tied to the company’s push to build AI across the systems needed to develop and run it.
The phrase “full-stack AI” is broad, and the available details do not provide a precise split between model development, computing infrastructure, or other AI work. The announcement does, however, place the fundraise in a specific contest: Alibaba is responding to an intensifying race in China and the West to build frontier models and the computing infrastructure behind them.
That distinction matters. A large language model is only one part of an AI business. Training and operating advanced models also requires substantial computing capacity. Alibaba’s statement connects its investment to both sides of that equation, the models and the infrastructure that supports them. The funding therefore appears designed to strengthen the company’s complete AI system rather than finance a single product launch.
The share sale was made to investors outside the United States and grew after the offering became oversubscribed, according to sources. Morgan Stanley, HSBC Holdings, UBS Group, and China International Capital Corp. acted as joint bookrunners.
Still, investors do not yet have a public line-by-line budget for the HK$80 billion. The important takeaway is both what Alibaba has said and what it has not: the money is reserved for full-stack AI, but the announcement does not reveal how much will go to models, computing, or other capabilities.
Why do AI companies need chips, infrastructure, and models together?
An AI model is only useful if a company can train it, run it, and deliver its answers to customers. That requires more than model research. It requires computing chips, data-center infrastructure, and software that connects these pieces into a working service.
The surprising part is how closely these layers depend on one another. Chips provide the processing power for training and operating models. Infrastructure supplies the storage, networks, cooling, and systems needed to use those chips at scale. Models turn that underlying capacity into products and services. Weakness in any one layer can limit the others.
A better model does not automatically create a better business if it is too expensive or slow to run. Likewise, access to powerful chips has limited value if a company lacks the systems needed to train models efficiently or the software to put them in customers’ hands. The value comes from coordinating the full stack.
That helps explain why Alibaba is directing the proceeds from its new share placement to its full-stack AI capabilities. The investment is not limited to developing a single model. It supports the broader set of technologies required to make AI reliable and commercially useful.
This approach also changes how AI spending should be judged. Large outlays may affect profits before they produce visible returns, because companies are building several connected layers at once. The payoff depends on whether those layers reinforce one another over time.
The takeaway is simple: AI progress is not just a model problem. It is a systems problem. Companies that control, connect, or efficiently operate the full stack can turn computing capacity into usable AI products.
What does “full-stack AI” mean in practical terms?
“Full-stack AI” sounds precise, but Alibaba’s disclosure leaves the phrase deliberately broad. The company said it will use 100% of the net proceeds from its HK$80 billion, or US$10.21 billion, share placement to invest in its “full stack” AI capabilities. It did not say how much would go to any particular part of that effort.
In practical terms, the phrase signals that Alibaba does not intend to fund only an AI model. Its business spans e-commerce and cloud computing, so the investment may support the wider set of capabilities needed to build and run AI services across those operations. That can include the computing capacity and software required to develop AI, as well as the cloud systems used to deliver it. The important point is not a single product. It is the supporting system around AI.
This distinction matters because models are only one part of the cost. AI also requires infrastructure, engineering work, and the ability to operate services at scale. Alibaba’s decision to commit the entire placement’s net proceeds to this area shows that it is treating AI as a company-wide investment rather than a small research project.
There is still a firm limit to what can be concluded. Alibaba did not provide an investment breakdown, and it made no further comment beyond its regulatory disclosure. The company had also already spent nearly half of its three-year capital expenditure plan by the April-to-June quarter.
The takeaway is simple: “full-stack AI” describes the complete operating foundation around AI, but Alibaba has not yet shown investors exactly which layers will receive the money.
Why accept a 3.6% discount on the shares?
Alibaba accepted a lower price for newly issued shares because raising money quickly can matter more than protecting the stock price in the short term. The company priced 710 million new shares at HK$112.70 each, while its Friday closing price was HK$123. That is an 8.4% discount based on the figures disclosed, not 3.6%. The difference is important because issuing shares below the market price immediately puts pressure on existing shareholders.
The unusual part is where the money is going. Alibaba said it would use all of the net proceeds from the HK$80 billion, or $10.2 billion, placement to expand its full-stack AI capabilities, including AI infrastructure. This is not a small research budget. The company had already spent nearly half of its three-year capital expenditure plan, and capital expenditure rose 75% to 67.7 billion yuan in the June quarter. Profit fell 75% over the same period as that spending weighed on results.
The placement therefore gives Alibaba fresh funding without relying only on operating cash flow while AI costs are rising. It also spreads the financing across non-U.S. investors, according to the disclosure. But the tradeoff is immediate: current shareholders own a smaller percentage of the company, and the lower issue price helped trigger a sharp market reaction. Alibaba shares fell as much as 10% in Hong Kong after the announcement.
The broader lesson is simple: a discount can be the price of speed and certainty. Alibaba is asking investors to accept dilution now in exchange for building the infrastructure it believes will support its AI business later.
How could this reshape Alibaba’s cloud and AI businesses?
Alibaba’s $10.2 billion share placement could change AI from one part of its cloud business into the organizing principle for the whole company. Alibaba says 100% of the net proceeds will go toward “full stack” AI capabilities, covering chips, infrastructure, and the development and deployment of AI models.
That allocation matters because AI services depend on more than a strong model. They also require computing capacity, specialized hardware, data center infrastructure, and software that can put models into everyday use. By funding each layer, Alibaba can build a tighter connection between the systems it operates and the AI services it sells.
The spending also comes as Alibaba’s capital expenditure rose 75% to 67.7 billion yuan. The share placement gives the company additional funding for a business that can require large, continuing investments before customers generate enough revenue to cover them.
For Alibaba Cloud, the likely effect is a stronger role as the company’s main delivery channel for AI. Cloud customers need somewhere to run models, process data, and deploy applications. Alibaba’s investment in chips and infrastructure could support that work, while its model development could provide the software layer customers use.
The structure may also help Alibaba compete through integration rather than through models alone. Vey-Sern Ling, a senior equity advisor at UBP, said Alibaba was well-positioned to pursue AI growth because of its full-stack advantages.
The key takeaway is simple: Alibaba is not treating AI as a single product. It is financing the hardware, infrastructure, models, and deployment tools together, which could make its cloud business more central to every stage of an AI workload.
What can other companies learn from funding the entire AI stack?
Alibaba’s financing plan points to a practical lesson: AI spending is not limited to training a model. The company said it would use proceeds from its share placement to build “full stack” AI capabilities, including chips, infrastructure, and the development and deployment of models.
That matters because model quality depends partly on the systems underneath it. Alibaba CEO Eddie Wu said the company needed to make capital investments first, so it could build the compute capacity required to capture future growth. In other words, demand cannot be served by software alone when the available computing infrastructure is too small.
The unexpected part is how large and broad the funding effort became. Alibaba planned to sell 710 million ordinary shares at HK$112.70 each, a 3.6% discount to its most recent closing price. The HK$80 billion placement would be the largest primary follow-on offering by a Hong Kong-listed company. Strong investor demand led Alibaba to increase the offering after it was oversubscribed, with sovereign wealth funds among the reported participants.
Other companies can draw a clear boundary from this example. AI investment should match the full path from hardware to usable products, rather than treating model development as a standalone project. But that also means committing capital before the resulting growth is certain. The funding has to support capacity that can serve future demand, not only current workloads.
The broader market signal is important too. MiniMax’s Hong Kong IPO, which raised capital for an AI company, shows that public markets are also willing to finance the compute-heavy infrastructure behind China’s AI expansion. The takeaway is simple: companies planning serious AI growth may need a financing strategy for the whole stack, not just a budget for model training.
Will more AI investment move from software budgets to infrastructure?
AI spending is starting to look less like a normal software purchase and more like a capital program. A company can pay for an AI application, but the system behind it also needs models, cloud capacity, chips, data centers, data, and specialized talent. Those costs do not disappear when a model becomes available through an API.
MiniMax’s Hong Kong listing shows how this funding need is widening. The company raised about $619 million in January after already raising more than $850 million privately. Its valuation rose from roughly $2.5 billion in its last major private round to about $6.5 billion at the IPO price. MiniMax builds models that handle text, audio, images, video, and music. It also runs consumer products such as Hailuo AI and Talkie, along with business API services.
The important point is not simply that one AI company went public. Public investors are being asked to finance more of the stack, including the expensive infrastructure needed to train and serve models. That gives AI companies access to a broader pool of money than private rounds and Big Tech balance sheets alone. It also gives investors exposure to much larger operating costs and more uncertain returns.
For enterprise buyers, this may change how AI budgets are planned. Software subscriptions will remain visible, but they may represent only one part of the bill. Usage, compute, storage, and the infrastructure supporting internal deployments can become equally important.
The takeaway is simple: AI investment is moving closer to infrastructure economics. The winners will not be judged only by model quality or user growth, but by whether they can turn costly compute into durable business value.