Developments in and around AI have progressed in highly disruptive ways in the years up to 2030 – both in closed and open-source models. Major and rapid advances in agentic AI have enabled AI’s integration across a wide range of applications and hardware, such as smart devices and robots. As a result, AI is omnipresent and plays an all-encompassing role in political, societal, economic and military spheres. The US and China dominate in closed military AI systems, while some countries, including the Netherlands and Gulf states, possess second-tier but competitive closed models.
Description
The team behind rAIdar[18] is shocked to see that their scale-up has made the national news overnight. Alphabet had launched a hostile takeover bid for its patents and intellectual property, prompting the Dutch government to panic and block the deal. Initially focused on using AI to improve radar systems for applications in commercial cars, the military dual-use opportunities led the Dutch Ministry of Defence to place several large orders with rAIdar last year. The move was welcomed by many, as it meant avoiding yet another source of dependency on American Big Tech and alternatives from the United Arab Emirates. No European government wants to repeat the mistake of the UK way back in 2015, when it let Google acquire DeepMind – the most promising AI lab in the world at the time.
rAIdar is not the only Dutch company that manages to compete on the global stage. Similarly, it is not unique in finding itself entrapped in a tense dynamic involving ambition, research, the global tech race and power politics. The Netherlands and the EU have long been indecisive and insecure in their industrial and innovation policies. They failed to rescue the tie-up between ASML and Mistral, which seemed promising just a few years earlier. With the return on investment from Mistral’s LLMs in the red and ASML in the shadow of China’s lithography boom, the collapse of two European champions is imminent.
Meanwhile, the US is overconfident and unpredictable. China still overly depends on advanced chips from the West, which are subject to strict export bans. Several other countries, such as Saudi Arabia, have moderately successful AI in different fields, but they dominate none.
Then, unexpectedly, the balance shifts. The United Arab Emirates unveils a compact AI-enhanced directed-energy weapon that is capable of disabling incoming missiles at a fraction of current costs.[19] Within weeks, China adapts the design and transfers it covertly to Russia, shifting the still ongoing war in Ukraine: Ukrainian offensive drones and missiles suddenly become obsolete. Western air defences suddenly look outdated. Yet the fragility cuts both ways: a rapid European breakthrough – such as secure photonic communications integrated with rAIdar’s stack – could flip the board just as quickly.
While some AI applications, such as rAIdar’s models, still partly rely on relatively traditional machine-learning applications, there has also been a boom in the use of agentic AI applications. The Netherlands has secured a leadership position in HealthTech. At the first-aid department of Dutch hospitals, patient triage has been fully automated. The decision-making process for where to send respective patients has sped up, halving the average waiting time for initial diagnosis while uplifting accuracy rates to over 98 per cent. In addition, AI agents are quicker and more accurate at drafting reports and patient dossiers, allowing doctors to focus on healing patients instead of doing paperwork.
Not all is well with agentic AI, however. In the US, the administration has introduced automated AI judges in federal criminal courts. The government deemed this necessary to ensure a fair and efficient judicial branch, while moreover saving money on an expensive institution. Conversely, human rights activists point at leaked government documents that reveal that the administration has knowingly and purposefully used an underlying LLM that increases racial inequality and bias in court rulings. The ‘black box’ nature of AI systems allowed the government to do so under the guise of introducing ‘an innovative and neutral computer program’.
Scenario Analysis 2: National Security Implications for the Netherlands in the EU Context
In this Plausible Tomorrow, agentic AI has rapidly evolved, driving not only economic transformation but also geopolitical reconfiguration. Global complexity has surged beyond the capacity of most states to navigate technological disruption independently. Competition is fierce in all layers of the technology stack, now mostly on algorithms and applications. Rivalry continues to exist in the hardware layers because of decreasing access prices and innovative breakthroughs that help to elevate those with more limited resources.
Companies struggle to stay afloat in this highly competitive and rapidly evolving environment. A fear of missing out on tech leadership and economic competitiveness reinforces the strategic alignment between states and their tech champions, as both sides increasingly feel the need to protect and assist each other amid intensifying geopolitical rivalry.
Governance models fail to keep up with rapidly developing and commercialised AI technologies. Global coordination on AI ethics and safety standards collapses under the weight of divergent state interests and corporate influence. Regulation remains largely symbolic and fragmented at the national and regional levels, as all power blocs aim to gain an edge over others.
Meanwhile, the economic landscape is marked by massive disruption. AI-induced automation drives deep restructuring of the labour market. The widespread availability of open-source models and agentic AI makes many knowledge-based professionals superfluous, including junior lawyers, graphic designers and programmers. Deemed indispensable just a few years ago, many of them are now replaceable by armies of AI agents controlled by a small number of specialised AI prompters. This puts huge pressure on the social security and cohesion of European welfare states, and the debate around policies like the Universal Basic Income now takes centre stage in the Netherlands. As some studies predict that up to 50 per cent of jobs may disappear across white-collar sectors before 2035, questions about social protections and income models are central. With this wave of unemployment coming, governments face starker choices than ever: as most profits leave the EU, they either have to increase income tax, and so risk crippling consumption; or reduce expenditure on social welfare or healthcare, thus increasing social unrest.
Truth decay further accelerates in a world dominated by agentic AI. Autonomous agents can generate, personalise and distribute tailored propaganda at a scale beyond human moderation capacity.
The penetration of agentic AI into critical services also generates risks for physical safety in a whole number of domains: failures in automated triage or biased decision systems in healthcare, for instance, could endanger patients rather than protect them. At the same time, the climate impact of mass-scale AI and its necessary energy-intensive data centres puts pressure on the European Green agenda and electricity grids. EU Member States and regions race to build their own Silicon Valleys – often without sustainable energy strategies.
Militarily, the blurred boundary between private innovation and strategic weaponisation becomes a source of risk. Highly innovative tech companies benefit ever more from first-mover advantages, offering dual-use systems that create new threats – and then proprietary solutions. The number of companies following the surveillance-based business model of Palantir has surged and there is little democratic control over them. The proliferation of cheap, autonomous and destructive technologies outpaces legislation and control regimes.
Disruptive AI is no longer an anomaly – it is the default, and geopolitical blocks are fighting to outflank each other.