Kenneth Tiong warns ESR omits critical frontier AI access risk facing Singapore

A Workers' Party Member of Parliament has published a pointed critique of Singapore's Economic Strategy Review (ESR), arguing that amid extensive discussion of artificial intelligence (AI), the report is silent on the most consequential risk facing the country's AI ambitions: whether Singapore will continue to have access to frontier AI models at all.

MP for Aljunied GRC, Kenneth Tiong published the piece on his personal website on 14 May 2026 — the day after the ESR's 32 final recommendations were released at the Singapore Business Federation's (SBF) Future Economy Conference. 

Those recommendations were grouped into eight thrusts covering economic growth, workforce transition, and resilience, with AI adoption featuring prominently throughout.

Tiong declared his interest as a director of a company that makes AI-enabled applications and consults on the same.

A critique rooted in parliament

Tiong's website piece builds directly on a speech he delivered in Parliament on 6 May 2026, during a motion on an AI transition with no jobless growth — a week before the ESR's final recommendations were published.

In that speech, Tiong had described two personal moments of awe and dread since the release of ChatGPT in November 2022: the first when he observed that coding ability, once considered essential, had become cheap and abundant within three years; the second in November 2025, when he began using an AI coding agent that he said could complete tasks overnight without supervision.

"It is an exhausting world it heralds," Tiong told Parliament, describing software engineers running multiple AI agents around the clock and recent graduates applying to hundreds of jobs without receiving a single interview.

Tiong had put forward three propositions: that access to premium AI and AI agents must be universal rather than gated behind course enrolment or union membership; that the handful of companies building frontier AI should be engaged with the same strategic seriousness Singapore brings to bilateral relations with countries; and that the retrenchment framework must be upgraded to match the speed of AI-driven displacement.

What the ESR misses

Returning to those themes after the ESR's release, Tiong argued in his post that the review's treatment of AI rests on a flawed foundational assumption.

"The ESR treats Singapore as a consumer and deployer of frontier capabilities produced elsewhere, not as an orchestrator of AI capability flows," he wrote.

He noted that the ESR's resilience thrust — which identifies energy, supply chains, food, and climate as domains requiring strategic buffers — does not extend that logic to AI compute and model access. He described this as a major omission.

"Its Thrust 8 identifies energy, supply chains, food, and climate as domains requiring strategic buffers and diversification — but does not extend this logic to AI compute and model access, which, if we are to take the AI Hub framing seriously, is this country's critical emerging dependency," Tiong wrote.

He argued the review is written entirely within the assumption that the current era of abundant access will continue and that frontier AI is a commodity input Singapore can procure, with the differentiator being what it does with that input. If that assumption breaks, he said, the entire three-tier deployment model — from leading companies as AI champions down to economy-wide diffusion — is compromised at its foundation.

Access already constrained

Tiong said the access problem is not theoretical.

He cited the geoblocking of certain AI products in China — noting in his parliamentary speech that he had learned first-hand during a recent visit to China that one major AI coding tool is blocked there entirely, covering mainland China, Hong Kong and Macau — as well as limited previews of advanced AI capabilities and restricted global product rollouts.

He also referenced what he described as the selective distribution of advanced AI security tools, including models reportedly capable of autonomously identifying previously undiscovered security vulnerabilities, to a restricted set of United States firms under named internal programmes.

He said OpenAI had taken a similar approach with a separate initiative.

Tiong attributed these characterisations to his own research and assessment.

"These are not just productivity questions — they include national security," he wrote.

In his parliamentary speech, Tiong had gone further, noting that even AI engineers at the world's leading frontier AI laboratories are reportedly concerned about falling behind because they cannot access one another's tools — suggesting access constraints operate even within the industry itself.

A subsidised moment that may not last

Tiong also raised the possibility that current AI pricing reflects an unsustainably subsidised market, drawing a parallel with ride-hailing platforms that offered cheap fares underwritten by venture capital before raising prices once market share was consolidated.

He noted that he currently pays approximately US$200 per month for an AI coding subscription he considers cost-effective, but said the pricing appears heavily subsidised relative to underlying compute costs.

If frontier AI pricing corrects to its true cost and remains substantially elevated, he argued, Singapore's diffusion model — which assumes affordable, widespread access — would price out everyone below the enterprise tier.

In his parliamentary speech, Tiong had cited figures suggesting that Anthropic's chief executive reported 80 per cent of revenue comes from enterprise customers driven by usage-based pricing.

If AI agents remain enterprise-grade by default, he argued, individual citizens including jobseekers, freelancers and retirees would be locked out of the tier where real productivity gains are made.

Four responses and Singapore's leverage

Tiong drew on analysis by Anton Leicht of the Carnegie Endowment, who he said identifies three compounding forces that will constrain frontier AI access: security concerns that motivate withholding, compute scarcity that makes serving frontier models genuinely zero-sum, and the eventual use of access controls for broader strategic purposes by the United States government.

He set out four responses he said he finds broadly correct.

The first is to reduce the justification for restriction by hardening cyber and biosecurity infrastructure. 

The second is to build data centres at scale to alleviate the compute shortage.

The third is to offer hyperscalers favourable terms for regional infrastructure buildout in exchange for contractual frontier access guarantees.

The fourth is to retain some independent AI development capacity as a fallback.

Tiong said Singapore holds meaningful cards on the third point, and that this is where he and Leicht converge.

In his parliamentary speech, he had argued that Singapore is among the world's leaders in water reuse and integrated water management — a binding constraint on data centre expansion across water-scarce Southeast Asia — and that this gives the country genuine leverage.

"If we position ourselves as the infrastructure partner of choice for this region, that is real leverage — something we bring to the table in exchange for access, for pricing, and for presence," he wrote.

He also called in his parliamentary speech for Singapore to attract genuine development offices from frontier AI companies rather than predominantly sales offices, and for the government to actively map how Singaporeans can enter the tight global labour market for frontier AI researchers — noting that annual salaries excluding equity for AI researchers in the United States are commonly in the range of half a million to one million US dollars.

On the broader geopolitical dimension, Tiong said it will likely remain a goal of middle powers to proliferate frontier AI capabilities, and that Singapore will need continued foreign policy engagement with both the United States and China to maintain ecosystem access.

"Many AI researchers and AI users I have spoken to, both Singaporean and foreign, have inequality of AI access as their number one worry," Tiong wrote. "If we don't have a frontier AI access resilience plan — sitting alongside the existing frameworks for energy, food, and supply chains — we are building an AI hub on an assumption we do not control."