Artificial Intelligence: How does innovation at every level transform industries?

Miracle for some, threat for others, artificial intelligence has become an essential technology for our economies. From hardware to services, it permeates the business world, transforming many industries and creating new opportunities for states, individuals, and companies.

Intelligent assistants, chatbots, autonomous cars, generative AI: artificial intelligences are making their way into our daily lives in multiple forms. But while some uses are very visible, they are only the tip of the iceberg. The AI value chain is much richer and teeming with diverse actors. Broadly speaking, the AI value chain can be divided into four main categories⁽¹⁾⁽²⁾:

- Chips: the first link in the chain, they are essential for developing, deploying, and training complex artificial intelligence systems.

- Infrastructure: necessary throughout the deployment of AI, the infrastructure includes data centers to provide computing power and store data, networks to transport data, as well as cloud hosting for applications.

- Foundation models: as the name suggests, these are the models on which AI applications, especially generative ones, are built. This segment includes learning models, AI libraries, and model performance optimization.

- Applications: the last link in the chain and also the most familiar to the general public, applications include interfaces and services for end users.

Each of these links plays an essential and complementary role in the AI value chain. Chips provide computing power, infrastructure offers necessary resources, foundation models bring intelligence, and applications make this intelligence usable and accessible.

These four links are interdependent, but each represents a separate market with its own challenges and distinct actors. Only a few companies operate across the entire chain - this is the case, in particular, for Google and Microsoft, present at every level³.

The demand for chips is exploding

GPUs, or Graphics Processing Units, have become essential for the development of AI. And while some American giants currently have a quasi-monopoly, the demand should naturally open it up to new players. Indeed, the GPU market is growing exponentially. It is expected to reach the size of a trillion dollars within 4 to 5 years⁴, including ancillary activities related to improving chip performance, bandwidth, memory, etc.

Major players have already announced colossal investments, such as OpenAI, the creator of ChatGPT, which stated this year that it is seeking $7 trillion for a project that should boost global chip production capacity⁵. In Europe, the European Chips Act is also set to accelerate funding for both public and private projects in this key field⁶.

The data centers are growing to meet the needs of AI

Just like the chip market, the data center market is growing at the pace of AI. The market is estimated at $416 billion in 2024 and could reach $624 billion in 2029⁷. Among the causes of this growth is the increase in the amount of data collected, but also the processing and analysis work required by artificial intelligence solutions. These often require access to phenomenal computing power.

The data center market is experiencing a massive influx of private capital. Institutional investors are ready to inject colossal sums into this rapidly expanding sector. At the same time, states are welcoming these projects, considering the presence of data centers on their territory as a strategic sovereignty issue.

In Europe, for example, Amazon has announced plans to invest over €15 billion in Spain⁸ and nearly €8 billion in Germany⁹, particularly to strengthen its cloud infrastructure.

An increasingly diverse range of AI applications

For end users, the future of AI promises an increasing diversity of applications. Foundational models - generative AI learning models like ChatGPT - will continue to improve to create increasingly ambitious solutions.

In almost every industry, beyond the use cases already in service, professionals are now imagining how AI will change their profession in the years to come. In construction, for example, AI could allow contractors to accurately predict building maintenance needs, thereby improving long-term energy efficiency¹⁰.

In healthcare, AI-assisted diagnostics could enable better patient care¹¹. In the banking world, AI is already being used in many areas, including advanced fraud detection, such as detecting deepfakes¹².

Logistics also shows promising advancements, particularly in the use of intelligent programs to optimize delivery routes or assist in supply chain management, especially when it is fragmented¹³. Finally, retail offers a wide range of AI applications: inventory management, purchase suggestions, demand forecasting, fraud detection and in-store surveillance, dynamic pricing, etc¹⁴. AI-assisted solutions seem endless, and even the risks posed by the technology, particularly in terms of data protection, are proving to be opportunities for certain actors, such as those in the cybersecurity sector¹⁵.

Future prospects

Throughout the AI value chain, from primary components to end applications, numerous actors work in synergy to develop transformative solutions that will shape the future.

While companies drive innovation, states have a role of support as well as control. In Europe, the EU Artificial Intelligence Act, which came into effect in 2024, aims to ensure the protection of collected data and their proper use by AI applications on the continent¹⁶.

The future prospects for this sector are impressive, as it continues to grow at all levels, with an increasing number of relevant applications every day.

All brands and logos used for illustrative purposes in this document are the property of their respective owners. The information provided is not intended to serve as investment advice or a recommendation to make (or refrain from making) any investment decisions and cannot be invoked as such.