Artificial intelligence does not abolish scarcity. It moves scarcity from computation to judgment, and in doing so it hands the advantage to whoever has the better judgment about how to use it. That is a contest a small country with deep human capital can win, and Bulgaria is better placed for it than its rankings suggest.
Every serious conversation about artificial intelligence eventually arrives at the wrong question. We ask how powerful these systems will become, how quickly they will improve, and which jobs they will absorb. Those questions are not unimportant. They are simply not decisive. Capability has rarely been the variable that determines which societies thrive during a technological transition. Judgment has.
At the Center for Intellectual Capital, Innovation and Competitiveness we work from one premise: technologies change, but the forces operating beneath them do not. Artificial intelligence is the most consequential leverage architecture of our time, and yet it obeys the same logic as every wave before it. Read that logic correctly and the strategic picture clarifies. Read the moment as a contest of raw capability and a country can win every benchmark while losing the decade.
Consider what AI actually does. Earlier technologies amplified a single activity. Printing amplified the reproduction of text. Electrification amplified physical work. Artificial intelligence amplifies cognition itself, which is to say it amplifies human judgment, sound and unsound alike, at scale and at speed. Amplification is neutral. A good decision becomes more powerful, and so does a careless one. This is why the central question of the era is not how intelligent machines will become. It is whether human institutions can govern the consequences of intelligence that has suddenly become cheap. That question is not abstract for a small economy. It decides whether Bulgaria spends the next decade setting terms or accepting them.
What stays constant when everything changes
Beneath every technology wave, four forces are always operative. They preceded artificial intelligence and they will outlast it, and reading AI through them is more reliable than reading it through any forecast of model capability.
Status comes first. Societies organize themselves around signals of relative standing, and always have. Signals rotate; competition for position does not. Technologies that redistribute status spread quickly, because someone gains. Technologies that threaten established hierarchies meet resistance proportional to the stakes. Artificial intelligence does both at once, which is why its reception is so uneven. It promises to lift outsiders into capabilities once reserved for large institutions, and it threatens incumbents whose advantage rested on scarce expertise. Knowing who gains standing and who loses it predicts adoption better than any benchmark score.
Leverage follows. Appetite for accomplishing more with less is as old as the lever itself. Each technological cycle is a new leverage architecture, and AI is the most general one yet built. What changes is the form leverage takes. What never changes is the demand for it. A society that declines to build with the leverage available to it does not preserve its position; it surrenders that position to a society that does build.
Scarcity is the pivot, and the most misunderstood of the four. Abundance in one dimension does not abolish scarcity. It relocates it. When computation became cheap, energy became the constraint. When information became abundant, the ability to evaluate it became rare, and therefore valuable. Artificial intelligence makes a particular kind of cognition almost free, and in the same motion it makes the complementary human capacities scarce: the judgment to decide what is worth doing, the discernment to recognize when a fluent answer is wrong, and the institutional capacity to act on either. Cheap intelligence does not lower the value of human judgment. It raises it. A country that understands this stops competing on the cost of cognition and starts competing on the quality of judgment, which is a contest it can actually win.
Adaptation closes the set. Systems adapt under constraint, and an organism that cannot adapt does not survive long enough to debate the philosophy of change. History keeps the receipts. Kodak held some of the earliest patents in digital photography and chose to protect the film business those patents threatened. Nokia built touchscreen and internet-capable prototypes as early as 2006, then let internal loyalty to its existing operating system bury them, and was caught flat-footed when the iPhone arrived in 2007. U.S. Steel, assembled in 1901 into the largest company on earth, spent the following century defending an old way of making steel against nimbler rivals and foreign competitors; in June 2025 it stopped trading as an independent American company, absorbed by a Japanese acquirer. None of these institutions failed lacking information. Each failed from an excess of attachment to one expression of a permanent need.
The exception proves the rule. Apple has survived four decades of technological upheaval precisely because it was willing to make its own best products obsolete, cannibalizing the iPod with the iPhone rather than waiting for someone else to do it. Survival was never about having the most advanced technology in the building. It was about the willingness to act on what that technology meant before the choice was forced. Artificial intelligence now puts every incumbent, Apple included, back in front of the same test, and the firms that hesitate will not be spared because they hesitated thoughtfully. Information was never the scarce input. Adaptation was, and adaptation is a decision, not a discovery.
Hold these four together and a working method appears. To anticipate how AI will reshape a sector or a nation, ask where it redistributes status, what leverage it offers, where it relocates scarcity, and who is positioned to adapt. The technology will keep changing. Those four questions will not.
A scoreboard that matters
Apply that method to the present and a familiar pattern appears, one that history has produced before. Capability is expanding faster than the capacity to direct it. Artificial intelligence advances through competitive markets, research breakthroughs, and enormous flows of investment, while the institutions responsible for education, regulation, and public governance evolve far more slowly. An asymmetry of this kind is not new. Industrial production was transformed long before societies built adequate labor protections. Nuclear capability arrived before the governance needed to contain its risks. Social media connected billions of people before anyone understood how algorithmic incentives would reshape public life. Every one of those gaps produced instability, and there is little reason to expect AI to be the gentle exception.
What is genuinely new is the breadth. Because AI amplifies cognition rather than any one task, the asymmetry surfaces everywhere at once: in courts assigning liability for autonomous systems, in schools unsure what to teach once expertise becomes a commodity, in firms automating decisions that no one can fully audit. None of these are engineering problems. They are problems of judgment and governance, which is to say they are problems of leadership. For a small state the breadth is not only a threat; it is an opening, because a country cannot lead everywhere but it can choose a few fields and lead in them decisively.
Most of the current debate measures the wrong thing. Benchmarks measure reasoning. Investors measure valuations. Enterprises measure efficiency gains. Each of those metrics is real, and each misses the one that matters most, which is whether people themselves become more capable. Do researchers make discoveries that were previously unreachable? Do small firms gain capabilities once reserved for large ones? Do citizens gain agency over their work and their institutions, or does a society grow quietly dependent on systems that fewer and fewer of its people understand? A future in which machines perform more tasks is not automatically a better future. A better future is one in which human beings accomplish more, and keep their judgment and their responsibility while doing so.
We frame that distinction through intellectual capital, which has three components, and the relationship among them is where competitiveness is actually decided. Human capital is the knowledge, skill, and judgment held by people. Structural capital is what an institution retains after its people go home: the processes, platforms, standards, and research environments that let knowledge compound instead of walking out the door. Relational capital is the trust and the networks that let an institution act with others. A country can be rich in one of these and poor in the rest, and that imbalance, far more than any gap in AI capability, decides whether it leads or follows. Competitiveness in this era is not a contest over who deploys the most powerful models. It is a contest over who converts human capital into structural and relational capital fastest, so that the value created by talented people accrues to the institutions and the surrounding country rather than dispersing somewhere else. That conversion is the scoreboard that matters. It is also, as it happens, precisely where Bulgaria's opportunity lies.
Bulgaria's move
We refuse to read Bulgaria's situation as a status report, because in a field moving this fast a status report is obsolete before it reaches its readers. We read it instead as a position to be converted into authority. Our raw material is unusually good. Bulgaria sits at three strategic fault lines simultaneously, and almost no other EU member state holds all three.
Energy comes first. Bulgaria is the most structurally central node in the southeastern European grid, with nuclear generation that no neighboring economy can replicate. As Europe electrifies, and as artificial intelligence turns electricity itself into a hard constraint on who can train and run advanced systems, that centrality stops being a legacy inheritance and becomes leverage over how an entire region manages its transition.
Security is the second. As a NATO member on the alliance's most exposed southeastern flank, Bulgaria faces threat vectors more proximate than most members ever have to confront. Proximity to risk, handled with competence, converts into proximity to relevance, and relevance inside an alliance is itself a form of capital.
Critical resources form the third. Bulgaria holds a mineral base whose strategic value to the European Union rises as the union pursues technological autonomy and works to reduce dependence on supply chains it does not control. Standing at any one of these fault lines would be notable. Standing at all three is a strategic asset of the first order, and whether Bulgaria treats it as one is a choice rather than a fact of geography.
Read those three fault lines through the four constants and the opportunity sharpens into something usable. Each is a source of leverage that Europe increasingly needs and cannot easily source elsewhere. Each sits on top of a scarcity the union is now exposed to, in power, in security, and in raw materials. Each could redistribute status toward a member long treated as peripheral, which is exactly why the position will not be handed over and has to be claimed. What remains is adaptation, the only one of the four constants that lies wholly within Bulgaria's own control. Geography supplied the first three. The fourth is a decision, and it is the one this report exists to force.
Every opportunity that follows from this position is mediated by a single variable. Human capital is the binding constraint across every field without exception, and on this point we are candid about the gap we mean to close. Bulgaria generates more knowledge than it activates. Talent is abundant; the structural capital that would let that talent compound at home is thin. The evidence is not subtle. At the 2025 International Olympiad in Informatics, in the discipline Bulgaria itself invented and first hosted in 1989, the national team finished fourth in the world, ahead of both the United States and Japan. A country that can out-rank the two largest technology economies on earth in raw informatics talent, and still ranks roughly twenty-sixth in the European Union for converting research into market value, does not have a talent problem. It has an activation problem. Its best human capital continues to leave, because the infrastructure that would let value flow back does not yet exist at comparable scale. Bulgaria's assets are real: membership in both the EU and NATO, an information and communications sector already worth more than four percent of national output, a flat ten percent tax, internet infrastructure above the European average, and an engineering base accumulated over generations. Yet the constraint is equally real: weak technology transfer, limited participation in the standards that will govern emerging fields, and institutional capacity below what strategy requires. We treat that constraint not as a confession but as the agenda itself. Closing the gap between knowledge generated and value captured is the single act that turns Bulgaria from a participant in the AI transition into a voice that others in the region cite.
Artificial intelligence raises the stakes on this gap rather than lowering them. When the tools to build are available everywhere, talented people no longer need to leave to work at the frontier, which means a country that builds even modest activation infrastructure can now retain value it would have lost a decade ago. That same openness cuts the other way. Talent that stays in place physically can still contribute its entire value to institutions abroad. Whether cheap and ubiquitous intelligence becomes an instrument of retention or an accelerant of loss depends on the structural capital a country chooses to build around its people, and that choice is available now in a way it will not remain indefinitely.
This agenda resolves into three decisions: where to compete, where to govern, and where to build.
We should compete wherever our position confers a natural advantage. Energy intelligence is one such arena, the layer of software, analytics, and applied expertise that makes the regional transition manageable, built directly on the grid centrality Bulgaria already holds. Cybersecurity governance is another, because a flank state that develops the recognition, the career pathways, and the response frameworks to match its exposure becomes the region's natural first call when something goes wrong. European regulatory implementation is a third arena, and an underrated one. Most observers discuss the AI Act, NIS2, and the Chips Act as compliance burdens. We see a market. Whoever builds the hybrid legal and technical expertise to apply those frameworks intelligently turns regulation into an export, sold across a region that needs the same capability and has nowhere obvious to buy it.
We should govern where capacity, not legislation, is the bottleneck. A framework is not the same as the ability to apply it well, and the distance between the two is precisely where AI governance fails. Energy presents the same pattern, where coal phase-out, nuclear decisions, and grid interconnection demand governance robust enough to survive a change of government rather than dissolve into energy politics. Biosecurity belongs here too, because the dual-use character of synthetic biology connects directly to NATO obligations and is the precondition for being trusted as a jurisdiction of responsible innovation. Beneath all of it sits data infrastructure, the intelligence layer underneath every digital field, where the design of governance decides whether European rules function as floors that enable or ceilings that prevent.
We should build the structural capital itself. A small number of genuinely excellent research environments, chosen in strategically selected domains and wired into European networks, will achieve more than funding spread broadly and thinly. Technology transfer infrastructure matters as much: professional people whose institutional role is conversion rather than administration, the connective tissue between knowledge and value that Bulgaria currently does without. Latent industrial assets from an earlier era can be revitalized into a defense industrial base, with NATO demand for autonomous systems, cybersecurity capability, and space hardware supplying the market context. And because the Western Balkans sit on accession trajectories, Bulgarian credibility grounded in EU and NATO membership creates a natural basis for regional intellectual capital leadership, the relational capital that lets a small country shape standards well beyond its own borders. A region learns to trust the institution that shows up early, codifies what works, and shares it openly; that habit, repeated across enough files, is how a small state earns a voice in rooms it does not formally control.
None of this depends on a single technological threshold. The future of artificial intelligence will not be settled by one breakthrough or one decisive model. It will be settled by thousands of smaller decisions about incentives, institutions, and governance, taken by people who either adapt to the new constraint or attach themselves to an old expression of an old need. Kodak had the patents. Nokia had the touchscreens. U.S. Steel had the market. Every one of them had the information, and not one of them had the will to act on what it meant in time. That is the test now facing every country sitting on talent it has not yet learned to keep.
Bulgaria has the ground. Whether it builds on that ground deliberately, or waits for others to build around it, is finally the only question that matters, and the constants beneath the change do not wait for anyone. Artificial intelligence has made one kind of intelligence cheap. In doing so it has made judgment, discernment, and institutional courage correspondingly precious. Those have always been the scarce inputs to leadership, and they are precisely the inputs a country can decide to build. The nations that lead the next decade will not be the ones with the most intelligence. They will be the ones with the most judgment about how to use it. That is a contest Bulgaria can win, and the window in which to enter it is open now.
A note on sources. Bulgarian figures and assessments draw on the European Innovation Scoreboard 2025 and the European Commission's 2025 country report on Bulgaria; the Innovation.bg 2025 report of the Applied Research and Communications Fund; BASSCOM and sector reporting on Bulgarian information and communications technology; and the Institute of Mathematics and Informatics at the Bulgarian Academy of Sciences, which reported the national team's fourth-place finish at the 2025 International Olympiad in Informatics. The corporate cases draw on widely reported business histories of Eastman Kodak, Nokia, Apple, and U.S. Steel, the last following Nippon Steel's completed acquisition in June 2025. The analytical framework of the four constants and the three fault lines is the Center's own.