---
title: Alibaba's Joe Tsai advocates for open-source AI to achieve European tech sovereignty
url: https://www.elseif.net/alibabas-joe-tsai-advocates-for-open-source-ai-to-achieve-european-tech-sovereignty
published: 2026-10-28T15:14:52+00:00
language: en
section: Models
source: https://www.xataka.com/robotica-e-ia/presidente-alibaba-tiene-claro-unica-forma-que-europa-alcance-soberania-tecnologica-ia-open-source
organizations: Alibaba
publisher: elseif
---

# Alibaba's Joe Tsai advocates for open-source AI to achieve European tech sovereignty

Alibaba's president and co-founder, Joe Tsai, has stated that the only way for Europe to achieve technological sovereignty is through open-source AI. Tsai made these remarks at the Wave event, which took place in Italy. He emphasized that Europe should not rely on closed APIs supplied by external companies but instead adopt open models that can be trained to meet specific needs. Tsai believes that this approach is crucial for Europe to gain independence in technology. He also noted that while Chinese AI companies share their innovations and advancements, US laboratories do not. This stance underscores the importance of open-source AI in ensuring that Europe can control its technological future without depending on foreign entities. Tsai's comments come at a time when Europe is increasingly concerned about its reliance on US and Chinese technology. Recent events, such as the international blockade of Anthropic's most powerful models by the US, have highlighted the risks of dependence on foreign technology. This has led companies like Siemens and Renault to combine models from different sources as a precautionary measure.

However, the challenges do not stop there. China and the US control approximately 90 percent of the global AI computing infrastructure, which means that even if Europe wanted to reduce its dependence, it lacks the necessary capacity. To address this issue, Brussels presented a package of measures on June 3. These measures aim to minimize Europe's external dependence in technology by investing in the construction of data centers, AI, and semiconductors. The package also promotes open-source solutions as an alternative. However, when it comes to open source, China is leading the way. Alibaba's strategy with AI is straightforward: to achieve the same success that Google had with Android. Tsai's proposal involves downloading the weights of an open model, running it on the company's own infrastructure, and training it with internal data. This approach ensures that sensitive information remains within the company. Alibaba recently published the weights of Qwen3.8, with its best version having 2.4 trillion parameters and 95 billion active parameters per token. This requires significant infrastructure to execute, making the smaller version, Qwen3.8-27B, more practical for most companies.

This version allows for commercial use, modification, and redistribution under the Apache 2.0 license. It is important to note that Alibaba has a vested interest in promoting open-source AI, as it is behind the open AI model Qwen. Additionally, Alibaba recently announced the new Zhenwu V900 chip, which will begin production in 2027, and plans for future models. However, Alibaba has not always been a strong advocate for open source. The shift towards open source began with Qwen3.8, partly in response to competitors like Moonshot and DeepSeek, who were already publishing large open models. Open models offer advantages in terms of control and flexibility, which is beneficial in principle. However, relying on Chinese models means shifting dependence from one third party to another, who also has the power to change licenses or withhold the best models. This is a temporary solution but does not address Europe's fundamental problem: the lack of capacity and proprietary models.