The CB Insights report on the state of AI in 2025 reveals a major tipping point: global investments have nearly doubled, but most is concentrated on a handful of players. For European SMEs and mid-caps, the challenge is not to rival these giants, but to understand where the real opportunities for action lie.
Published on 2026-01-20 by Nathalie Lamborghini Dumas.
The widening gap
The numbers are staggering. $225.8 billion invested in AI in 2025, up from $114 billion the previous year. But behind this apparent doubling lies a more nuanced reality: mega-rounds of over $100 million now represent 79% of total funding. Three players (OpenAI, Anthropic, and xAI) alone capture $86 billion, or 38% of the total.
This concentration phenomenon is significant for European leaders. It signals that the race for large language models (LLMs) has become a game of titans where only players with colossal resources can hope to compete on infrastructure. But it also reveals, in contrast, that the bulk of value for traditional companies lies elsewhere.
Robotics and agents: the two waves not to miss
The report highlights a major strategic evolution: robotics now captures 11.4% of AI investments, leading all technology sectors. This convergence between artificial intelligence and physical systems opens concrete prospects for European industry, historically strong in manufacturing expertise.
Even more significant: the fastest-growing sectors concern AI agent infrastructure (+87%), voice platforms (+54%) and multi-agent systems (+50%). In other words, AI is no longer content to analyze or generate content. It is preparing to act, decide and orchestrate complex processes autonomously.
For leaders, this evolution raises a fundamental question: are you ready to integrate AI agents into your value chains, or will you suffer them as new intermediaries imposed by others?
Accelerated consolidation: giants are buying intelligence
782 AI startup acquisitions were completed in 2025, 1.5 times more than in 2024. AI now represents 7.5% of venture capital M&A exits, up from less than 5% the previous year. Meta spent $3 billion on Manus, Workday $1.1 billion on Sana Labs.
This acquisition frenzy reflects a strategic urgency among large groups: rapidly integrating AI capabilities rather than developing them internally. For SMEs and mid-caps, this means that rare AI skills are becoming monetizable assets, and that the window to position yourself as an orchestrator remains open but is progressively closing.
Unicorns born for the commercial battle
The report reveals a qualitative change: 63% of new AI unicorns reach billion-dollar valuations while already operating at commercial scale, compared to only 23% in 2024. The era of speculative valuations on simple prototypes seems over.
This maturation is good news for traditional companies: available AI solutions have reached a level of robustness sufficient for deployment in demanding business contexts. The time of perpetual experimentation is coming to an end.
Europe at the crossroads
Europe represents 22% of global AI transactions, compared to 51% for the United States. But this reading through the lens of fundraising can be misleading. Europe has assets that the numbers don't capture: a dense industrial fabric, unique business data, established trust relationships with customers and partners.
The real challenge for European companies is not to become the next OpenAI. Let's take an example: an agricultural cooperative that uses standard AI software to optimize its harvests remains a customer of a solution. The same cooperative that connects farmers, transporters, buyers and weather data on its own platform becomes the indispensable actor of its territory. The first pays a subscription. The second takes a commission on every ecosystem transaction.
Four questions to take action
Facing this acceleration, I propose that leaders ask themselves four concrete questions.
First, which decision-making processes in your value chain (including among your customers and suppliers) could be orchestrated by AI agents? And who will control them: you or a third-party platform?
Second, are your business data sufficiently structured to become the foundation of a sector platform, or will they remain scattered in tools you don't control?
Third, where can AI and robotics allow you to connect actors that are currently fragmented : and become the indispensable intermediary of these flows?
Fourth, in 3 years, will you be the one orchestrating the interactions in your sector, or an interchangeable supplier listed on someone else's platform?
A closing window, a model to invent
The $225 billion invested in 2025 is not an alarm signal, but a starting signal. The rapid consolidation of the market, massive acquisitions and the emergence of autonomous agents are reshaping value chains in real time. Companies that wait to "see how it evolves" risk waking up in an ecosystem whose rules have been written by others.
But the most important lesson from this report is not in the staggering amounts. It is in what these investments do not fund: business expertise, customer knowledge, the ability to execute on the ground. Tomorrow's winners will be neither pure AI players nor traditional companies ignoring the revolution. They will be organizations capable of hybridizing artificial intelligence and business intelligence, automation and human judgment, technological scale and customer proximity.
For European SMEs and mid-caps, the time has come to transform their assets (data, expertise, trust relationships) into lasting strategic positions. It is this combination they are ideally positioned to invent. Those who act in the next 18 months will have the choice of their place in the ecosystem. The others will have to settle for the one assigned to them.