When AI Firms Need Neutral Ground, Ask What They Are Routing Around
Hard-news summary
Reuters reported that Singapore is becoming neutral ground for the artificial-intelligence sector as Chinese startups try to operate beyond Beijing’s reach and U.S. firms seek foreign talent without stricter U.S. visa friction. The report described Singapore as moving from East-West gateway to a place companies can use to keep both China and the United States at a distance while the two powers fight over export controls, talent controls, data, chips, and AI leadership.
Reuters cited Kerry Goh, CEO of Kamet Capital, saying that locating intellectual property in Singapore gives international clients comfort that a startup is not directly subject to either Chinese or U.S. controls. The report also said Singapore is using AI-talent visas and tax breaks for registered intellectual property to attract companies from multiple geographies. Firms with ties to China or the United States named in the article include Workato, Addepar, Plaud AI, Harvey AI, Anthropic, OpenAI, Meta’s Superintelligence Labs, and Google DeepMind.
Commentary
The accountability story is not simply that Singapore is good at attracting technology companies. It is that the AI stack is learning to route around national power. When firms move talent, intellectual property, headquarters, revenue, and data into a jurisdiction marketed as neutral, the public should ask what is being avoided and who still has authority when something goes wrong.
A neutral hub can be legitimate. Singapore has real infrastructure, talent, legal capacity, bilingual advantages, and a long record as a business platform. Companies do not automatically become suspect because they choose a stable jurisdiction. But AI is not ordinary software. It carries strategic value, labor consequences, privacy risk, and model-control questions that governments are now treating as national-security terrain.
That is why the wording matters. Reuters reports that Singapore is no longer just a bridge between East and West. It is increasingly a place to keep both sides at bay. That is an elegant business strategy. It is also a governance problem. If Washington cannot fully see the stack, Beijing cannot fully claim it, and local clients are told the product is no longer really Chinese or American, then the next question is obvious: whose rules actually bind the system?
The same pattern keeps appearing across the AI economy. Data centers route around local power constraints until ratepayers notice. Model companies route around copyright norms until courts intervene. Chip supply chains route around export controls until governments tighten the map. Talent systems route around immigration bottlenecks until entire hubs emerge around the gaps. The technology is fast, but the institutional question is old: power looks for the most flexible jurisdiction.
Singapore’s appeal could also make it a pressure point. Reuters quoted a National University of Singapore political scientist warning that the city-state could be seen as a grey space for technology transfers that one or both major governments disallow. That is the risk with neutral ground: it can be a bridge, a buffer, or a loophole depending on who is using it and what they are moving through it.
For citizens, the issue is not whether Singapore should attract AI companies. The issue is whether the AI industry is building a global accountability shell game. A company can say the intellectual property is on the island, the founders are no longer under one passport regime, the engineers are spread across borders, the data is not where the critics think it is, and the revenue is booked somewhere else. Each fact may be defensible. Together, they can make responsibility harder to locate.
This is where governments need better questions. Where is the model trained? Where is the data stored? Where are the key engineers? Which government can compel access? Which court can hear a dispute? Which regulator can audit the system? Which customers are being reassured by the corporate structure? Which countries are being reassured, and which are being routed around?
The United States and China are both trying to protect strategic technology. Both also risk pushing companies into more complex structures as firms chase capital, customers, talent, and operational flexibility. Washington tightens chips and visas. Beijing tightens data and talent movement. Companies respond by finding places where the rules are friendlier, faster, or less politically radioactive. That is rational private behavior. It is not automatically a public-interest architecture.
Singapore may be one of the winners of this moment. It may also become one of the places where the contradictions of AI globalization become visible first. The city-state can attract talent, capital, and headquarters functions. But if the superpowers decide it is becoming a transfer corridor rather than a neutral platform, the very neutrality that made it valuable could draw restrictions.
The lesson for readers is simple: AI is becoming geopolitical infrastructure. It does not just live in code repositories and product demos. It lives in visa systems, tax rules, IP registries, sovereign wealth funds, export-control maps, and corporate domiciles. If companies say they need neutral ground, the public should ask neutral from whom, accountable to whom, and protected by what rules.
The answer cannot be a lazy suspicion of every cross-border startup. It also cannot be blind trust in corporate geography. Neutral ground is useful only if it clarifies responsibility. If it mainly obscures who controls the model, who can access the data, and who answers for harm, it becomes another layer in the stack that citizens are expected to accept without receipts.