Alibaba's $53 Billion Global AI Push: The GPU Crisis and Chip Sovereignty
Originally published in Turkish. Read the original →
Alibaba will build AI data centers in Europe, the Middle East and Latin America with a $53 billion investment over 3 years. US bans on Nvidia chips and data sovereignty concerns stand out as the biggest obstacles to the plan.
Chinese tech giant Alibaba announced a massive $53 billion AI investment plan at its Apsara conference this week. The company will build new AI data centers in Europe, the Middle East, Southeast Asia and Latin America. However, US bans on Nvidia chips and data sovereignty concerns emerge as the biggest obstacles facing this ambitious plan.
Qwen3-Omni: A Multimodal AI Model
One of the most important products Alibaba unveiled at the conference in Hangzhou was the Qwen3-Omni large language model (LLM). This model:
- Can process text, images, audio and video
- Can generate speech as well as text
- Is offered under the Apache 2.0 license
The Apache 2.0 license matters a lot: companies integrating this model into their own systems don't have to stay tied to ChatGPT or other closed ecosystems. This gives it an important advantage against rivals like OpenAI and Google.
A $53 Billion Infrastructure Investment
Alibaba's plans are not limited to a new model launch. The company aims to expand its global data center network by investing $53 billion over the next 3 years.
Current Footprint
- 91 availability zones
- Operations across 29 regions
- Existing facilities in London and Germany
Newly Planned Data Centers
- Netherlands - First data center
- France - First data center
- Brazil - First data center
- Additional facilities in the Middle East and Southeast Asia
Dr. Feifei Li, President of International Business at Alibaba and Senior Vice President of the Cloud Intelligence Group, explained:
> "The strategic expansion of our global infrastructure is designed to meet the accelerating demand from our forward-looking customers."
The Proximity Strategy: The Advantage of Being Close to Users
Neal Riley of the Adaptavist Group comments on Alibaba's strategy as follows:
> "International competition in generative AI is heating up in the model arena with the emergence of strong rivals from China such as Qwen and DeepSeek. Alibaba's move shows that similar competition will also be seen among infrastructure providers for AI tools and systems."
Riley underlines the concept of proximity:
> "Proximity to demand is very important - for Alibaba to compete on a global scale, it needs to position itself as close to its users as possible."
What Will Be Offered?
The services Alibaba is expected to offer from its new European data centers:
- Standard cloud services
- Elastic compute services
- Big data analytics
- Machine learning
- AI development platforms
Developer Incentives
To persuade companies to use its facilities, Alibaba is offering these incentives:
- 2 billion free tokens on Model Studio
- Up to $120,000 in cloud credits
- Resources for building AI agents and applications
The Chip Problem: How Do You Do AI Without Nvidia?
Here is the most critical point: Nvidia GPUs have become the lifeblood of AI infrastructure. However, as a Chinese company, Alibaba is blocked by the US from buying Nvidia's top-end hardware.
The US-China Chip War
- Washington has banned Nvidia from selling its top-end chips to China
- Nvidia has a restricted H20 chip developed specifically for China
- The US wants a cut of these H20 sales
- Beijing has told Chinese companies not to use even Nvidia's H20 chips
The Alternative: T-Head Chips
Alibaba's own T-Head chips are now considered able to compete with the H20. This reduces the company's dependence on US technology.
The Nvidia Partnership Paradox
An interesting detail: while Nvidia is blocked from selling chips to Alibaba, the two companies announced a partnership on physical AI (robotics). It is an interesting example of how complex the tech wars are.
The Data Sovereignty Problem
On top of the chip problem, there are other obstacles in front of Alibaba's European plans: data sovereignty concerns.
Europe's Worry About Washington
European governments and customers are worried that Washington could pressure cloud providers and access data in European facilities. In June, Microsoft France executives confirmed:
> "that they could not guarantee that data belonging to French citizens would never be transferred to the US government without the explicit consent of French authorities."
China's Own Pressure Methods
But the Chinese government also has ways to intervene in its own cloud operators. This leaves Europe facing the following question:
Will Alibaba's plans increase autonomy, or are they a new threat?
The EU's Foreign Direct Investment Protocols
The EU's Foreign Direct Investment (FDI) protocols can come into play when there are concerns about foreign investment in critical national infrastructure. Data centers definitely fall into this category.
The Vodafone Model
To get around these obstacles, Alibaba will likely rent capacity from data center operators. For example, its first facilities in Germany are located in Vodafone facilities in Frankfurt.
This approach:
- Avoids being subject to the EU FDI regime
- Means the operators have already done the heavy lifting on land, power and planning
- Allows faster market entry
The UK's Position
The UK, home to one of Alibaba's existing data centers, is notably absent from the new plans.
A spokesperson for the UK Department for Science, Innovation and Technology (DSIT) made the following statement:
> "We don't speculate on possible investments, but data centers are now regulated as critical national infrastructure. The government can step in to review any investment deal from any country that could affect national security."
The UK's recent $42 billion trade deal with the US (led by Microsoft and Google) includes roughly 120,000 GPUs from Nvidia. This suggests the UK is busy with other things for now.
Technical Challenges
Other factors complicating data center plans:
- Power shortages: AI data centers consume enormous amounts of energy
- Finding land: Suitable locations are limited
- Access to water: Critical for cooling systems
- Local opposition: Environmental concerns and NIMBY (Not In My Backyard) resistance
Conclusion: An Ambitious but Uncertain Future
Alibaba's $53 billion plan shows that the company wants to be a serious player in the global AI market. The Apache 2.0 license of the Qwen3-Omni model and the planned data center expansion are strong steps supporting this goal.
But there are significant obstacles ahead:
1. GPU supply: US bans and China's own directives 2. Data sovereignty: Europe's concerns about both the US and China 3. Regulatory approvals: EU FDI protocols and national security reviews 4. Physical constraints: Difficulties with power, water and finding land
Alibaba's effort to chart its own course with T-Head chips and its strategy of partnering with operators like Vodafone are smart moves to overcome these obstacles. But how much of these plans will actually come to life will become clear over the coming years.
One thing is certain: the AI infrastructure war is not just a technical matter - it is a multi-layered struggle with geopolitical, economic and strategic dimensions. Alibaba's place in this war will play an important role in shaping the future of the global technology balance.
