Nuclear analogy fails for AI control
The debate over nationalising artificial intelligence rests on a neat analogy: AI, like nuclear technology, is a dual-use tool that can power a city or level one. The argument, as presented by DBS’ chief economist Taimur Baig and former IMF economist Anthony Annett in The Business Times, is that if AI carries a comparable dual-use risk, perhaps it deserves comparable treatment Baig & Annett, The Business Times
The problem is that the analogy does not hold. Uranium is a physical substance. You can fence a reactor, guard an enrichment facility, and inspect nuclear sites for diversion to weapons-grade material. That is roughly what the non-proliferation regime does. AI is code, learning models, and know-how. It has already diffused across thousands of researchers, universities, and open-source repositories. As we have seen with the rise of freely downloadable Chinese models, it has crossed national borders that no treaty currently governs. You cannot ring-fence what has already leaked into millions of laptops.
The Nuclear Mirage
The nuclear analogy was always a rhetorical shortcut, not a structural map. It sounds persuasive in a policy paper because it borrows the gravity of Cold War-era arms control. But the material reality is different. When the United States nationalised uranium enrichment during the Manhattan Project, it controlled a finite resource that could be physically secured. When the Soviet Union built its nuclear complex, it sealed entire cities behind barbed wire. AI has no equivalent of a reactor core. Its dangerous capabilities are not locked inside a single facility; they are distributed across servers, research papers, and open-weight models that anyone with an internet connection can download.
Baig and Annett’s first concern is safety. Advanced AI systems could, in the wrong hands, disrupt financial systems, erode wealth, or interfere with utilities such as power grids and water systems. Rogue actors could use AI to spread disinformation at industrial scale, run scams indistinguishable from genuine communication, or design a lethal pathogen. These are not science fiction scenarios; they are the stated concerns of the very labs behind the technology Baig & Annett, The Business Times
But the response to these risks cannot be the same as the response to nuclear proliferation. You cannot inspect AI models the way you inspect uranium centrifuges. The capabilities are in the weights, which are numbers. You can copy them infinitely. You can hide them on a USB drive. The International Atomic Energy Agency has no equivalent for software because software does not glow in the dark.
The second concern is about power. A handful of firms — OpenAI, Anthropic, Google, and a few others — are building the foundational systems on which entire economies may soon run. This concentration does not stay economic; it translates into political power. Public ownership would blunt that conversion. There is also a fairness argument: the breakthroughs underpinning today’s AI trace back to decades of publicly funded research, and the models are trained on data generated by all of us. If the public funded the seed and supplied the soil, perhaps it deserves a share of the harvest Baig & Annett, The Business Times
This is a persuasive case. But it is also too broad to survive contact with how AI actually works. The AI ecosystem has layers, and the safety and power arguments apply to them unevenly. At the very top sit the foundational models — the GPT-, Claude-, and Gemini-class systems capable of catastrophic misuse. This is where the nuclear analogy holds some water. Beneath that sits infrastructure — chips, power supply, data centres. This is a resource allocation and supply chain problem, not a control-of-catastrophic-capability problem. Taiwan does not need to nationalise TSMC to manage the risk of a rogue chip; it needs export controls and standards for testing. And at the widest layer sit startups and application developers — thousands of companies building narrow, specific tools on top of models they do not own or train themselves. This is where nationalisation makes the least sense because this is precisely the layer where competition, speed, and trial-and-error matter most, and where the catastrophic-risk argument barely applies.
The Bureaucracy Trap
Even narrowing the target to the handful of firms building frontier models does not make state ownership costless. Governments are rarely the fastest or most creative builders of cutting-edge technology. Bureaucracies optimise for caution and accountability, not the rapid, hit-and-miss experimentation that produces breakthroughs. Nationalise the labs building frontier models and you may get safer AI, but you will very likely also get slower and less capable AI Baig & Annett, The Business Times
History is littered with examples of state-led technology projects that failed because they could not match the speed of private-sector innovation. The Soviet Union built a space programme that put the first man in orbit, but it could not build a personal computer that anyone wanted to use. France’s Minitel system was a state-run digital network that preceded the internet but died because it could not evolve. The United States’ attempt to build a government-run health insurance website in 2013 was a technical disaster that took months to fix. These are not arguments against all state intervention; they are arguments against the assumption that state ownership automatically produces better outcomes.
The advocates of nationalisation might respond that safety is more important than speed. That is a legitimate position, but it comes with a trade-off that must be acknowledged. Slower, less capable AI is not necessarily safer AI. If a government-run lab falls behind the frontier, the most dangerous capabilities may emerge elsewhere, in jurisdictions with weaker oversight or no oversight at all. The race to the bottom in AI safety does not stop because one country nationalises its labs; it accelerates because the gap between the regulated and the unregulated widens.
The Singapore Exception
For a country like Singapore, nationalisation makes even less sense. Singapore’s AI strategy has never been about building the next frontier model; it is about applications — tools built by private firms, often small ones, solving specific problems in logistics, finance, healthcare, and language. The country’s economic model runs on foreign investment, open competition, and a reputation for being an easy, predictable place to build a business. A state-dominated AI industry would cut directly against all three Baig & Annett, The Business Times
It would signal to foreign investors that the state intends to dominate the sector in which they plan to invest. It would narrow competition in exactly the layer — applications — where competition drives quality. And it would very likely produce worse products, not safer ones, since the safety case for nationalisation was always about frontier models, not about the delivery app bolting a chatbot onto its customer service line. Singapore’s interest is not in owning AI. It is in making sure AI, mostly owned and built elsewhere, is governed well when it arrives.
This is not a uniquely Singaporean insight. Every small, open economy faces the same calculus. You cannot nationalise what you do not have. You can only regulate what arrives at your borders. The question is not whether to own AI; it is whether to build the infrastructure for safe adoption.
The Referee Role
If outright ownership is the wrong tool for nearly the whole AI economy, that does not leave governments with nothing to do. It leaves them with the more difficult job of designing the rules of the game rather than becoming the main player. The first lever is guard rails around sensitive sectors. National security, finance, and critical utilities deserve tight, specific regulation because they are exposed to systemic risk. The second lever is incentive design. Tax breaks and subsidies can be effective instruments, and governments can use them to nudge AI development towards job-augmenting tools rather than job-displacing ones; to slow the roll-out of AI in elementary education so it does not blunt children’s cognitive development; and to reward innovation aimed at public good in areas such as public health, urban planning, and disaster response, rather than purely private profit .
The third lever is governance standards with real teeth. This is where regulation earns its keep: mandatory testing and certification before high-risk AI systems go live, just like how new drugs are approved before they hit the market; explainability requirements for decisions that materially affect people’s lives, in health, credit, hiring, and insurance; independent audits before deployment, not after a failure; mandatory incident reporting when AI systems cause harm; and provenance standards — both for the data used to train a model, and for labelling AI-generated content, so disinformation carries a fingerprint. Singapore has already begun building some of this through AI Verify, the world’s first government-built AI testing framework IMDA, Singapore.
The question is not whether to regulate. The question is whether regulation can be effective without ownership. The answer is yes, but only if the rules are enforced. The history of financial regulation shows that rules without enforcement are worse than no rules at all because they create a false sense of security. The history of nuclear regulation shows that ownership without rules is equally dangerous because it concentrates risk in the hands of the state. The middle path is the hardest to walk, but it is also the only one that leads somewhere.
The Sentence Left Unspoken
The advocates of nationalisation raise legitimate concerns. AI can disrupt financial systems, erode wealth, and spread disinformation. It can concentrate power in the hands of a few firms. The public funded the research that made it possible. These are real problems that demand real solutions.
But the solution is not nationalisation. The solution is governance. The solution is regulation. The solution is the slow, unglamorous work of building institutions that can keep up with the technology without strangling it. The sentence left unspoken in the room after everything has been said is this: nationalisation is a confession of failure. It is the admission that we cannot govern what we have built, so we must own it instead. But ownership without governance is just another form of capture. The goal is not to own AI. The goal is to live with it.
Sources
1. DBS
3. Anthropic
4. Google