Artificial intelligence (AI) has become one of the most important arenas of global economic and technological competition by 2025. It is now clear that the world’s three major poles — the United States, China and the European Union — are pursuing fundamentally different strategies. The US builds AI development on fast, market-driven innovation; China on a state-steered ecosystem; and Europe on a regulated and safety-oriented model. These three paths represent not only three regulatory philosophies, but three distinct economic and societal logics. This comparison is particularly timely now, as Hungary’s new AI Act enters into force on 1 December 2025.
Over the past year, the United States has visibly entered a new phase with the launch of the AI in Science programme, which connects federal laboratories, national databases and supercomputing capacities to accelerate drug discovery, energy technology and other strategic fields through AI. This approach continues to follow the logic of “innovation first, regulation later”, where speed itself is the key competitive advantage — and it is precisely this direction that the EU’s own AI in Science strategy seeks to emulate. Indeed, the most advanced large language models emerge first within this ecosystem. Analyses of major scientific infrastructures show that where governments coordinate laboratories, data assets and computational resources, the pace of scientific breakthroughs measurably increases.
However, this model also has a price: social, privacy and security concerns often surface only after technologies are deployed on the market, increasing corporate risk exposure and creating complex, sometimes opaque responsibility chains.
China’s AI strategy is even more complex and cannot be reduced to state-driven infrastructure building. Since the National AI Plan of 2017, Beijing has been operating a hybrid model: the government defines strategic priorities, while major technology companies — Baidu, Alibaba, Tencent, Huawei — execute them in close partnership. These companies develop language models, deliver AI-based innovations in automotive systems, e-commerce and fintech, and participate in state-supported projects. Urban management systems, autonomous vehicle test zones and industrial automation are all areas where state coordination and private resources operate jointly.
China’s advantage lies in scale and coordination: it can mobilise enormous datasets, substantial computing capacity and centralised investment flows. At the same time, the model is far less transparent to external observers. Details of state–industry cooperation and the sources of training data are often not public. The societal embedding of AI development also follows a different logic: constraints on the use of personal data are far weaker, which is ethically and politically incompatible with European standards. For these reasons, the Chinese model is not only difficult to adapt in Europe — it is practically impossible.
In this global environment, Europe has consciously positioned itself differently. It has recognised that if it attempted to compete with the US or China in speed or risk-taking, it would almost certainly lose. This is not a criticism but an acknowledgment of structural realities: Europe has smaller capital markets, slower-developing AI infrastructures and lacks the type of major technology players needed to scale rapidly. Therefore, the European Union has deliberately chosen a “third way”: it does not aim to be the fastest, but the most trustworthy.
The AI Act seeks to create an ecosystem that simultaneously protects fundamental rights, supports innovation and provides a high-quality regulatory framework for companies. This is complemented by the EU’s new reporting system, which enables anonymous, safe notifications of potential breaches of the regulation. Franco-German proposals — including President Macron’s Paris speech on a “28th regime,” European preference, and a new AI IPCEI — indicate that alongside regulation, Europe is now increasingly building its competitiveness pillar as well. This is precisely the missing link without which the “third way” cannot become a genuine alternative to American speed or Chinese coordination.
Yet the greatest strength of this strategy is also its greatest risk. Regulation only has value if it can be enforced. Auditing how a model functions, how it was trained, on what data, and what risks it carries is extremely difficult in the case of large technology companies. Due to the complexity of foundation models, traditional conformity assessments face significant limitations.
The EU relies heavily on its GDPR experience: billions of euros in data-protection fines demonstrate that the Union can enforce its rules. But AI is a different domain, evolving far more rapidly. The AI Office’s current staffing and resources are modest relative to the scale of the systems it must supervise. Moreover, member states are establishing their oversight structures at different speeds and capacities, which can lead to uneven implementation. This is why the AI Office’s investigative powers, mandatory risk-management documentation and built-in oversight mechanisms are critical: they aim to ensure that regulation works in practice, not only on paper.
The key question remains whether European and national authorities will be able to keep pace with technological developments — and what happens if regulatory advantage fails to compensate for technological lag.
Hungary’s new AI Act (Act LXXV of 2025) fits into this European architecture. The law establishes the domestic compliance and oversight framework, creating two types of authorities: the notification authority, which designates conformity assessment bodies for AI systems, and the market surveillance authority, which oversees the lawful use of AI systems. The government will designate the specific institutions in a separate implementing regulation. Data-protection matters will continue to fall under the competences of the National Authority for Data Protection and Freedom of Information (NAIH).
The structure of the law follows the EU AI Act: operators of high-risk systems face registration, documentation and risk-management obligations, while the list of prohibited practices mirrors EU rules. For Hungarian companies, this creates a more predictable legal environment, but also introduces administrative burdens. The law will not accelerate technological development; in some cases, it may even slow it down. However, by reducing legal and reputational risks, it can provide longer-term competitive advantages.
But this still does not fully answer the question many Hungarian companies are asking today: if the US is racing ahead, how can we compete?
The reality is that Hungarian firms would not be able to follow the American trajectory in any case. The speed of model development, capital market dynamics and infrastructural scale of US startups and tech giants cannot be replicated. The Hungarian and European approach follows a different logic: risk minimisation, operational stability, and legal predictability.
For a Hungarian company, the true competitive advantage does not lie in matching the pace of American development, but in the fact that the solutions it offers already comply with European regulations. The stringent requirements of the AI Act create a market in which compliant domestic firms can outperform foreign players that do not prioritise European compliance.
However, it is also essential to recognise the risks of this strategy. If most Hungarian firms remain merely users of AI systems developed abroad and fail to build their own development capacities, the compliance advantage may eventually become hollow. Compliance alone does not create technological knowledge, nor does it secure positions at higher levels of the value chain.
Thus, the core question is not only whether Hungarian companies comply with regulations, but whether they can develop the domestic competencies needed to become developers, not just users, within the AI economy. Regulation can be a competitive advantage — but only if genuine innovation and knowledge stand behind it.
Ultimately, the success of Europe’s “third way” depends on whether this balance can be achieved.
Picture source: jypix / depositphotos.com
