The AI Race Beyond the U.S. and China
The global race for AI is often described as a contest between the US and China, with the two countries constantly outdoing one another with ever more advanced language models. But what about the rest of the world? One answer is that they should try to develop their own, comparable foundational AI models. However, the ASEAN Future Forum, hosted by Vietnam in June 2026 with the participation of international experts, highlighted just how difficult this path is. This is because both South-East Asia and Europe lack comparably large datasets, computing capacity, energy and cutting-edge research.
Focusing on AI-alternatives and applications
The question for these regions is: what alternative AI pathways are available under these different conditions? One option is to focus on AI techniques beyond language models, such as image analysis. A second option is to focus on the application of language models. The starting point for this approach is the observation that, in many markets, chatbots are not off-the-shelf products, but must be adapted to sector-specific needs and integrated into physical objects. Language models remain an important foundation in this context, but are only one component of the value chain; in some cases, smaller models may suffice. What Clayton Christensen described in relation to earlier foundational technologies may apply here: Technology overshoots demand.i
Language models as a commodity
Furthermore, whilst there are not many, there is more than one large language model. In the long term, this could lead to language models becoming a commodity – that is, an interchangeable basis for a wide range of goods and services. Large language models could therefore become less crucial in many sectors than is currently assumed. This would be very attractive for both Europe and ASEAN. This is because one effect of such AI value chains would be that there would not only be very few winners (the manufacturers of large language models), but a great many of them, including all users across the various sectors.
Shared challenges – incomplete Integration
However, many steps are still required to implement these alternatives. In this regard, the regions can learn a great deal from one another, as they share important common ground: both regions tend to have fragmented economic structures dominated by SMEs and are characterised by their diversity. The EU Member States are just as heterogeneous and incompletely integrated as the ASEAN states. Consequently, both regions face the task of establishing harmonised standards for data protection, cross-border data exchange and interoperability – on the ASEAN side, for example, through the ASEAN Framework on Digital Data Governance, the ASEAN Model Contractual Clauses, the ASEAN Digital Masterplan and the ASEAN Digital Economy Framework Agreement (DEFA). One example of cooperation building on this is the Joint Guide issued by ASEAN and the EU, which assists businesses in complying with regulations on data transfers between EU and ASEAN jurisdictions.
Segmented data sets
However, both regions also face the challenge that data sets are fragmented, difficult to access and categorised inconsistently. Although the EU has established a fair amount of infrastructure in recent years to facilitate data sharing – such as the Common European Data Spaces – there is still a lack of incentives for businesses to share data, for example. This is an area where best practices can be exchanged.
Limited computing capacity
The same applies to computing capacity: both regions have very limited computing capacity, due both to their access to high-performance chips and to their energy situation. Consequently, they are reliant on AI solutions and infrastructures that consume fewer or different resources than American and Chinese solutions – such as forms of distributed machine learning.
Great potential for collaboration
It is worth deepening the transfer of innovation between Europe and ASEAN in the areas of AI infrastructure, AI applications and AI governance. Both regions must find their own path in AI, beyond the US or Chinese models. In doing so, they face very different starting points, particularly in terms of resources and talent. However, this should not obscure the fact that there are similar challenges regarding AI value chains, datasets and energy consumption. There is great potential for cooperation in this area.
The first steps have already been taken, such as the agreement signed between the EU and Singapore at the end of 2024, which aims to strengthen cooperation between the EU’s AI Office and Singapore’s AI Safety Institute. Further steps must follow – in keeping with the motto: ‘Leaving no one behind!’
