Opinion
DIGITALISATION
Gigabit connectivity for Europe: bridging the digital divide by 2030
Miryam Aranzadi
European Projects Consultant
ARTIFICIAL INTELLIIGENCE
The upcoming AI Factories will support public and private entities, including SMEs and startups, in the development of cutting-edge AI models
Head of European Projects at the Bilbao office
The European Union has identified Artificial Intelligence (AI) as a key technology for its future, and faces the challenge of positioning Europe as a leader in the development and deployment of reliable and ethical AI. It is one of the key areas for which Europe is seeking strategic autonomy, reducing our technological dependence on third countries and developing our own capabilities.
By way of background, we can go back to 2018, with the publication of the European AI Strategy, which set out the European Commission’s vision of the transformative potential of AI, while also warning of the risks associated with its uncontrolled development and use. This strategy already sensed that these technologies would develop rapidly, driven by increased computational power and the availability of large amounts of data for training, but the pace and scale of advances have exceeded most expectations. In recent years, in addition to the generative AI models that have become immensely popular, we are seeing spectacular advances in fields such as deep learning, natural language processing (NLP), computer vision, reinforcement learning, etc.
As an evolution of this strategy, which established a general framework and a long-term vision, in January 2024 the European Commission published a new AI Innovation Package. It includes specific measures and tools to provide AI startups with access to supercomputing resources; define AI training and capacity building programmes for researchers and professionals; support SMEs to adopt AI and develop AI-based products and services; and encourage public-private collaboration to foster an AI innovation ecosystem. It is also proposed to set up AI Factories to drive technological development.
AI Factories are envisioned as dynamic ecosystems that bring together the key ingredients to support public and private entities, including SMEs and startups, in the development of cutting-edge AI models. The idea is that they offer a centralised access point to assist in the development and validation of large-scale AI models and applications by providing infrastructure and services. Examples of such services could be AI model training, capacity building, technical advice and support, or access to collaborative networks.
In practice, these will be centralised or distributed entities that provide access to supercomputing resources. These may be supercomputers optimised for AI or partitions of an AI-dedicated supercomputer. Around it will be infrastructure for data storage, high-speed networking, big data for training and AI development environments, as well as staff with expertise in AI, supercomputing, software development, and data scientists.
The metaphor of the ‘factory’ is useful for understanding AI Factories. We can imagine large digital factories where researchers, companies and startups (the ‘workers’) can access state-of-the-art supercomputers (the ‘machinery’), large data sets (the ‘raw material’) and cutting-edge development tools (the ‘production process’), with the aim of creating AI applications such as chatbots, virtual assistants, recommender systems, etc. (the ‘final product’).
Although there are similarities, AI Factories should not be confused with sectoral Testing and Experimentation Facilities (TEFs), which have been launched in recent years. While the former focus on providing access to supercomputers to develop and train large AI models, the latter set up controlled environments where companies can test and validate AI technologies in specific sectors, such as healthcare or agriculture, prior to real-world deployment. In other words, while AI Factories prioritise infrastructure and algorithmic development, TEFs specialise in applied experimentation and validation of specific solutions.
Funding for AI Factories will be a key issue if this new paradigm is to be successful. The multiannual financial framework 2021-2027 earmarks significant funds for the development and deployment of digital technologies, including AI. Beyond the Horizon Europe framework programme, which funds the development of R&D&I projects and the scaling up of AI startups, the Digital Europe programme covers investments for the deployment of these technologies and training in their use.
Digital Europe is expected to be the main framework for supporting the creation and operation of AI Factories, which should also receive funding from member states and companies. As a first step, in the context of the AI Innovation Package, an amendment to the regulation of the public-private partnership in supercomputing (EuroHPC) has been approved to establish AI Factories as a new pillar in its activities.
This will provide additional funds (the amount to be specified) to acquire or optimise AI supercomputers, facilitate access to them, train AI supercomputing experts and cooperate with related initiatives. This has already materialised in the publication of two calls for expressions of interest for entities intending to host the first AI Factories in Europe.
AI Factories should contribute to Europe’s strategic autonomy and strengthen Europe’s AI innovation ecosystem. They are expected to support the research and development of European AI technologies and their application in virtually all sectors. However, there are also challenges on the horizon, the main one being the availability of adequate funding: while the mobilisation of public funds for infrastructure investment and the operation and maintenance of AI Factories is already foreseen, it remains to be seen how co-financing from private sources, member states and regional governments will be secured.
Other challenges include the availability of talent for the operation of AI Factories, as there is currently a shortage of skilled AI professionals in Europe. It will also be key to ensure that AI Factories are able to navigate the European regulatory framework, especially the AI Act, without adding significant cost and complexity layers to the development and use of AI. This relates to the challenge of managing the ethical and security impact of the large AI models that will be developed in these facilities, particularly around the protection of fundamental rights and privacy.
The AI Factories are a strategic commitment of the European Union to consolidate its leadership in the field of AI. These factories are expected to become the catalysts of innovation in Europe, driving the development of novel AI technologies and applications, job creation and improving the quality of life of European citizens. Mario Draghi’s report on ‘The Future of European Competitiveness’ underlines the strategic importance of AI)for Europe, as a driver to keep Europe at the forefront of innovation and to radically transform multiple sectors, and already mentions AI Factories as key players to address innovative use cases of AI in multiple sectors.
Bilbao Office
Head of European Projects at the Bilbao office
Opinion
DIGITALISATION
Miryam Aranzadi
European Projects Consultant
Opinion
INNOVATION
Daniel Errea
Consultant
Opinion
Innovation Fund
Juan Sanciñena
Senior Consultant
News
EUHubs4Data
Apply before the 8th of February 2021 to join the EUHubs4Data ecosystem.
Opinion
DIGITALISATION
Miryam Aranzadi
European Projects Consultant
Publication
INTERREG PROGRAMME
In this comprehensive ebook by Kaila you can learn more about the main European territorial cooperation programme.