The global race for dominance in artificial intelligence is entering a period in which technological superiority may prove far more expensive than the economic benefits it can generate.
Gartner’s warning that more than $3.5 trillion in AI infrastructure investment could be exposed to impairment risk by the end of 2030 is not merely a troubling forecast for financial markets.
It reveals a fundamental contradiction at the heart of the emerging digital economy: governments and corporations are committing enormous amounts of capital to secure strategic power, without any certainty that the commercial exploitation of these investments will justify their cost.
According to Gartner’s analysis, published on October 14, 2026, AI providers are expected to spend nearly $1 trillion on infrastructure in 2026 alone, with an additional $6 trillion in investment projected between 2027 and 2030.
This represents an unprecedented mobilization of capital, transforming data centers, advanced semiconductors, energy networks, and telecommunications infrastructure into fundamental components of international power. The problem is that the economic returns on these investments remain highly uncertain.
The $4.9 trillion revenue black hole
Gartner estimates that AI data center owners will need to generate cumulative revenues of $8.5 trillion by 2030 to cover announced capital expenditures, financing costs, and a required 12% return on invested capital.
However, total revenues from enterprise and consumer AI applications attributable to these data centers are projected to reach no more than $3.6 trillion. This leaves a $4.9 trillion gap between required and projected revenues, with the latter covering only approximately 42% of the amount needed to meet investors’ financial objectives.
This gap does not mean that the entire amount will be lost. It does, however, suggest that the current investment model assumes returns that projected demand appears unable to support.
Gartner further estimates that, if the current investment trajectory continues, the value of assets exposed to impairment risk could exceed $4 trillion by 2032.
The distinction is important. This is not a prediction of inevitable capital destruction, but an assessment of the scale of investments that may fail to deliver their originally anticipated returns.
The high cost of American supremacy
The geopolitical dimension becomes particularly evident in the United States, where a substantial share of the artificial intelligence technology ecosystem is concentrated.
American technology giants are not competing solely for market share. They are seeking to control computing power, digital platforms, and the infrastructure upon which a significant portion of the future global economy will operate.
This strategy creates a distinctive conflict between economic efficiency and national security.
From a business perspective, an investment that fails to generate adequate returns should be reduced or restructured. From a geopolitical perspective, however, abandoning critical infrastructure may mean surrendering technological influence to competing powers.
The United States may therefore face a difficult dilemma: allow market forces to correct excessive investment or continue developing strategic infrastructure even when its short-term financial performance is unsatisfactory.
It is far from certain that this will lead to direct government financing. Nevertheless, it could increase the political importance of public procurement, tax incentives, and measures designed to secure energy and technological capacity.
China sees opportunity in Washington’s vulnerability
For China, the possibility of a substantial correction in the AI infrastructure market presents both risks and opportunities.
Beijing regards artificial intelligence as a strategic sector closely linked to industrial production, technological self-sufficiency, defense capabilities, and economic competitiveness.
American restrictions on access to advanced semiconductors have already made the development of domestic technological capabilities a national priority.
A potential crisis of overinvestment in the West could provide China with an opportunity to strengthen its relative position, particularly if American companies are forced to reduce their spending.
On the other hand, China is not insulated from the same economic realities. Excessive infrastructure development, rapid equipment obsolescence, and insufficient commercial demand can create similar problems within the Chinese technology ecosystem.
The real difference may lie in how the two economies manage the cost of their technological strategies.
In the United States, pressure to deliver returns to shareholders and financiers could accelerate restructuring. In China, greater capacity for state coordination may allow certain investments to be maintained for strategic reasons, although this does not eliminate their economic cost.
The great shakeout: fewer players, greater power
One of Gartner’s most significant predictions concerns so-called neocloud providers, companies specializing in the provision of computing infrastructure for artificial intelligence applications.
The research firm estimates that more than 80% of these providers will have ceased operating independently by 2030.
This does not necessarily mean that these companies will go bankrupt. A substantial proportion may be acquired, merged, or absorbed into larger corporate structures.
This is precisely where the geopolitical implications become particularly significant.
If the market becomes concentrated among a small number of major providers, control over computing power could become even more restricted to a handful of corporations and, by extension, a limited number of countries.
Large cloud companies benefit from diversified revenue streams and a greater ability to absorb investment costs. Smaller providers depend much more heavily on the rapid commercialization of their equipment.
As new generations of semiconductors replace existing ones, the available period for recovering investment costs becomes shorter. Gartner estimates that growth in spending on AI-optimized Infrastructure as a Service, from $66 billion in 2027 to $174 billion in 2030, will not be sufficient for most providers to recover their investments before today’s equipment becomes technologically obsolete.
This development could lead to a new form of digital dependency, in which governments, businesses, and public institutions increasingly rely on fewer suppliers of critical computing services.
Europe risks becoming a digital dependency zone
For the European Union, Gartner’s forecasts highlight a particularly complex strategic challenge.
Europe is seeking to strengthen its technological autonomy, yet it remains heavily dependent on non-European companies for advanced semiconductors, major computing platforms, and cloud services.
If the AI infrastructure market undergoes substantial consolidation, Europe’s negotiating position could weaken further.
At the same time, however, a correction in excessive valuations and computing costs could create opportunities for European companies that do not seek to compete with American technology giants in terms of scale, but instead focus on developing specialized applications.
Europe’s strategy does not necessarily have to be based on replicating the American model of massive infrastructure investment.
It could concentrate on sectors where Europe already possesses a strong industrial base, including manufacturing, energy, transportation, shipping, and industrial automation systems.
The central question is whether Europe will remain primarily a consumer of American and Asian technologies or develop its own capabilities in selected strategic sectors.
The next AI battleground is electricity
The development of artificial intelligence infrastructure is directly connected to another issue that is often underestimated in financial analysis: access to large quantities of reliable and competitively priced electricity.
Data centers are not merely digital facilities. They are energy-intensive industrial infrastructures requiring highly reliable electricity networks, cooling systems, specialized equipment, and substantial investment in physical installations.
Consequently, the geography of artificial intelligence will not be determined exclusively by the presence of technology companies.
It will increasingly be shaped by energy availability, electricity costs, grid capacity, and political stability.
Countries with competitive energy costs and well-developed infrastructure may gain an advantage in attracting data center investment. Conversely, economies with limited energy capacity will face additional obstacles.
When the technology boom meets geopolitical reality
The history of major technological transformations demonstrates that periods of overinvestment are far from unusual. Railways, telecommunications, and the internet all experienced phases in which infrastructure development significantly outpaced actual commercial demand.
In many cases, early investors suffered substantial losses, yet the infrastructure they created ultimately became the foundation of a new economic era.
The same may happen with artificial intelligence.
A potential impairment of trillions of dollars in investment would not necessarily signify the failure of the technology itself. It could instead mark the failure of particular business models and the redistribution of assets toward financially stronger economic players.
What distinguishes the current technological cycle from previous ones is that artificial intelligence is now directly connected to military power, intelligence operations, industrial production, and the ability of states to control critical functions of their economies.
A potential financial crisis in AI infrastructure would therefore not necessarily remain confined to stock market valuations.
It could accelerate the concentration of technological power, reshape relationships of dependency between governments and corporations, and influence the balance of power among the United States, China, and Europe.
Who will own what survives?
The real issue is not simply who will develop the most powerful artificial intelligence, but who will possess the financial resilience to sustain its infrastructure until it becomes economically viable.
More importantly, who will ultimately control the infrastructure that remains once the current cycle of overinvestment has run its course?
In this process, financial losers will not necessarily be technological losers. Equally, those who gain control of infrastructure following market restructuring may acquire geopolitical influence disproportionately greater than the capital they ultimately invest.
This is perhaps the most important message emerging from Gartner’s analysis. The next phase of competition in artificial intelligence will not be decided exclusively in laboratories or through algorithms, but also on corporate balance sheets, across energy networks, and through the ability of governments to sustain their strategic choices over the long term.
George S. Skordilis is Editor-in-Chief of geo-trends.eu.