By Javier Surasky
When people talk about artificial intelligence, discussions usually revolve around models, algorithms, or applications. If we go one step further, questions of ethics, resource consumption, data management, human rights, or even what it means to “think” or “be intelligent” begin to emerge.
All these issues intersect with others, such as human
security in the context of AI, the use of dual-use technologies (civilian and
military), possible forms of AI governance at the national and global levels,
and legal responsibilities arising from its use. Together, they give shape to a
true AI ecosystem whose main actors include companies, governments,
universities, international organizations, research centers, investors, civil
society organizations, and experts who participate, in different ways, in its
development and management.
Understanding artificial intelligence therefore requires looking beyond the technology itself, adopting a position situated in time and space, and asking who has the capacity to develop it, finance it, regulate it, deploy it, research it, and define the rules under which it operates.
AI Does Not Exist in a Vacuum
Artificial intelligence systems are the result of networks of mineral resources, computing power, infrastructure, data, and human talent. Developing advanced models requires all of them, along with enormous amounts of available capital, human labor that is not always properly recognized, the development of markets, and users open to innovation who can incorporate AI into economic and social activities, both public and private.
All of this takes place in a world marked by inequality and crowded with optimistic and pessimistic AI “gurus” who alternately raise their voices about a future that is promising, uncertain, dangerous, or even one of extinction.
Seeking to rebalance the playing field, governments, international organizations, courts, civil society organizations, scientific communities, and other actors step in to try to define rules, standards, and principles. Yet even among them, there are little more than a few broad agreements that remain difficult to translate into concrete action.
From this perspective, a specific system of economic, political, and institutional relations takes shape around AI, filtering into social life through multiple tools used every day by many people: large language models such as ChatGPT, Claude, or DeepSeek; AI-assisted audio and video editing platforms; image and spam filters; traffic predictions; and route recommendations for getting from one place to another are just a few examples.
Those who control strategic resources can influence the trajectory of the technology that accompanies us, just as those who produce knowledge help determine which problems are researched, those who develop infrastructure shape access to advanced capabilities, and those who establish laws and standards define the conditions under which these systems reach society.
Analyzing AI without taking these relationships into account is, at the very least, naive, and can even be dangerous.
From Ecosystem to Power
Talking about an “artificial intelligence ecosystem” helps me convey the idea that this is a field made up of multiple interconnected actors. But not all of them have the same capacity to influence the whole.
In my view, power within the AI ecosystem is therefore distributed, relational, and multidimensional.
In this map, power is understood as something difficult to quantify, but visible in an actor’s capacity to control resources, shape decisions, set agendas, or alter the possibilities for action available to others.
It can stem from economic capital, control over infrastructure, access to data, scientific capacity, regulatory power, knowledge production, political influence, or the ability to define the terms of the debate.
Mapping actors makes it possible to identify them and analyze how they exercise power within the “spaghetti bowl” of the AI field.
Who Is Part of the AI Ecosystem?
The universe of actors is very broad, and its boundaries are sometimes blurred. With that caveat in mind, some categories help me organize the mapping of AI actors.
The private sector occupies a key position. Technology companies, infrastructure providers, semiconductor manufacturers, digital platforms, startups, and investors participate at different stages of AI value chains, making some options viable and others not.
Governments intervene as regulators, funders, buyers, developers, and users of AI systems through the adoption of public policies, especially in areas such as industry, science, education, and the establishment of regulatory frameworks for AI itself. They also set priorities and, at times, national AI strategies that favor one of the technological and/or regulatory development models currently competing at the global level.
Academia and research institutions produce scientific knowledge and train much of the human talent that later circulates among universities, companies, and governments.
Think tanks participate in the production of ideas, assessments, and public policy proposals, helping connect research, governments, companies, and public debate.
Civil society brings in perspectives related to human rights, inclusion, transparency, accountability, social impact, and democratic participation. Some organizations act as watchdogs, while many others have taken more proactive approaches aimed at encouraging change.
Last, but certainly not least—at least in theory—international organizations provide spaces where states and other actors come together and engage in dialogue, and they are among the best-positioned forums for addressing a technology that, by its very nature, crosses borders and cultures.
These categories are not watertight compartments, and the circulation of people, knowledge, resources, and ideas among them is a permanent feature of the AI ecosystem. In addition, the same organization can occupy several positions at once by developing models, funding university research, participating in the construction of AI governance, and providing technological infrastructure.
Mapping Actors Means Mapping Relationships
Identifying actors is only the first step.
A power map becomes truly useful when it also allows us to observe the relationships among those actors. Who funds whom? Who provides infrastructure? Who regulates, and how? Who controls models, data, or computing capacity? Who is excluded from the benefits of AI?
These questions take us from an aggregated list of actors toward the idea of an “ecosystem,” because an organization’s power depends, among other variables, on its position within this network.
The AI map also has a geographical dimension: the capabilities required to develop advanced AI systems are distributed in profoundly unequal ways across countries and regions.
Computing infrastructure, educational and scientific capabilities, capital, and companies capable of developing frontier models tend to be concentrated in a small number of centers, raising the question of “digital peripheries.” A map of AI power should also make these asymmetries visible.
A Necessarily Dynamic Map
The artificial intelligence ecosystem changes very quickly. New companies emerge; acquisitions take place; new alliances appear; specialized public and private institutions and university research centers are created; and new semiconductors keep pushing models forward, while regulatory frameworks struggle to keep pace.
For that reason, an AI power Map is, by definition, always unfinished. Even so, it helps us observe change and follow its evolution, and it may even prove useful in putting names and faces to decision-makers who are not always easy to identify.
From the Map to the Questions
The goal of the Global Radar Analytics AI Power Map is to provide an entry point that helps identify those who participate in the construction of contemporary artificial intelligence and understand what resources they control, how they relate to one another, and what capacity they have to influence the direction of this technology, now intertwined with the fate of our societies.
Every new actor added to the map opens new questions; every actor excluded from access to decisions that will affect their lives raises a warning sign about the world we are building.
Taken as a whole, the map allows us to transform a dispersed set of organizations into an institutional, economic, and political architecture shaped by those who have the capacity to set agendas and narratives that condition and steer the path of artificial intelligence.