Free tools Windows power users keep installed
One-click scans. No signup required.
iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Alibaba chairman Joe Tsai’s argument is that open AI models could give European organizations more choice among suppliers and more control over how they deploy AI. That is a case for diversification—not a promise of technological independence. At VivaTech in Paris in June 2026, when asked whether Europe could trust China not to cut off access to its technology, Tsai’s reported answer was: “you can’t.”
What Joe Tsai argued about Europe’s AI choices
Tsai promoted open-source Chinese models, including Alibaba’s Qwen family, as one way European businesses could diversify their technology stacks and protect data. The point is not that an open model makes its origin country irrelevant. It is that access to model components and the ability to choose where and how to run a model may give a buyer alternatives to relying on a small number of proprietary platforms.
His blunt answer about China underscores the limit: open access does not remove geopolitical risk. A model may be available today while its development, updates, hosting, support, or surrounding infrastructure still depend on organizations and jurisdictions a European buyer does not control. Tsai’s pitch is best understood as an argument for adding options, not eliminating dependence. Caixin Global reported on Tsai’s VivaTech remarks on June 22, 2026.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Why Alibaba is making the case
Alibaba has a direct commercial interest in wider use of open models: it develops Qwen and sells cloud infrastructure. Tsai describes the company’s AI strategy as an integrated stack spanning chips and infrastructure, foundation models, and applications. In Alibaba’s account of his VivaTech remarks, he said, “Having an integrated, full-stack approach is our core strategy.”
#1 Best Overall
Open model access and paid cloud services can coexist. A company may release or make model weights available while earning revenue from the compute, hosting, tools, and services used to run AI. Alibaba has made this point explicitly: its account of Tsai’s February 2026 World Government Summit remarks says open-model developers can create value elsewhere in the stack, including cloud computing. That is Alibaba’s business case, not an independent guarantee that open models will be cheaper or more advantageous for every buyer.
Alibaba said it had pledged more than US$53 billion for AI and cloud infrastructure over three years beginning in February 2025. The company also said it opened its third European cloud hub in France, after Germany and the United Kingdom, and planned agentic AI services for European markets later in 2026. These are company statements; the announced future services should not be treated as launched unless current availability is confirmed. Alibaba’s account of Tsai’s VivaTech 2026 position describes its full-stack strategy and European plans.
What “open” can—and cannot—mean
“Open-source AI” is often used loosely. A model’s weights may be accessible while its training data, training code, documentation, development process, or license terms are not. That distinction matters because access to weights can make deployment or adaptation possible without necessarily allowing outsiders to reproduce how the model was built or fully audit its provenance.
Nor does an open model automatically put a European organization in control of its data. That depends on where inference runs, who operates the infrastructure, what data is retained, and which laws and contracts govern the service. A model downloaded for local deployment offers a different control profile from the same model accessed through a hosted API; neither arrangement answers every question about security, support, updates, or continuity.
Rank #3
The Linux Foundation’s August 2025 report on Europe’s open-source community identifies active work by researchers, grassroots projects, startups, and large enterprises, including efforts on European-language datasets, benchmarks, leaderboards, and models. It names OpenLLM Europe and OpenGPT-X and notes interest in smaller, specialized models. The report also describes barriers: compute and funding constraints, uncertainty about regulatory responsibilities, and cases where important components remain closed beyond model weights. The Linux Foundation report therefore presents openness as a practical opportunity with unresolved constraints, not a completed sovereignty solution.
What European respondents say matters for sovereign AI
In the Linux Foundation’s 2025 Global Collaboration in AI Survey, 89% of the Europe-only sample of 70 respondents selected open-source software as an approach critical to advancing sovereign AI. This is a multiple-selection survey result about respondents’ views, not proof that open source achieves sovereignty or a representative poll of all Europeans.
| Approach selected | Share of Europe-only respondents (n=70) |
|---|---|
| Open-source software | 89% |
| Open data | 69% |
| Open standards | 69% |
| Open governance | 49% |
| Open infrastructure | 37% |
| Open hardware | 20% |
Respondents could select multiple approaches, so these percentages are not competing shares that add up to 100%. The survey wording asked which open approaches were critical to advancing sovereign AI; it did not measure whether those approaches had delivered it. The Linux Foundation’s survey report gives the question and Europe sample details.
Recommended Free Tools
How a European organization should evaluate the options
Choosing between a downloadable model, a hosted open model, and a proprietary AI service is not a simple open-versus-closed decision. The relevant trade-offs depend on the model and deployment arrangement. Before committing, compare:
- What is actually open: Check whether the weights, code, training data, documentation, and license are available, and what the license permits. Do not infer that all components are open because a model is described that way.
- Where inference runs: Establish whether the model runs on infrastructure your organization operates, a cloud provider’s environment, or another hosted service—and who can administer that environment.
- Data handling and jurisdiction: Review retention, logging, access controls, processing locations, contractual terms, and the laws that apply to the provider and infrastructure.
- Fit for the work: Test performance on your actual tasks, languages, and data. European-language support or a model’s availability does not establish that it will meet a specific organization’s quality threshold.
- Compute and operating burden: Estimate the hardware or cloud capacity, engineering time, monitoring, maintenance, and update work required for your intended deployment. A model that can be downloaded is not automatically inexpensive to operate.
- Security, support, and upkeep: Decide who patches and monitors the system, how vulnerabilities are handled, and what support exists if a model or service changes.
- Regulatory responsibilities: Determine which duties fall on your organization, provider, or both. The Linux Foundation report identifies clearer understanding of regulatory obligations as an ecosystem need.
- Continuity and geopolitical exposure: Consider the possibility of lost access to updates, cloud services, expertise, or other dependencies—not just whether weights are currently downloadable.
These are decision criteria, not a product ranking: the cited reports do not provide a controlled comparison of Qwen, European models, or hosted AI services. Alibaba’s European cloud presence may be relevant when assessing hosted infrastructure, but its location alone does not establish that a deployment meets an organization’s requirements for data control or sovereignty.
What open models might make possible for smaller builders
Tsai has also argued that freely available models could widen who can experiment with AI. In Alibaba’s account of his March 2025 Converge Live remarks, he said that when models are freely available, “anyone can build, customize, and apply AI” beyond the largest technology companies. That describes a potential route for smaller businesses, individuals, and entrepreneurs to build applications; it is not a measured finding that open models have already produced equal access or successful outcomes. Alibaba’s account of Tsai’s March 2025 remarks presents this as his view of how open models may spread innovation.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
