Note: This is a research note supplementing the book Unscarcity, now available for purchase. These notes expand on concepts from the main text. Start here or get the book.

Singapore Smart City: The World’s Living Laboratory for AI Governance

How a city-state smaller than New York proved that AI can run civilization without becoming Skynet

The Petri Dish of the Future

Here’s a question that should keep you up at night: can we actually build the systems described in this Blueprint, or are they just pretty pictures drawn by optimists who’ve never wrestled with a government database?

This article answers that question. Singapore has spent a decade and billions of dollars running the world’s most thorough experiment in AI-integrated governance. What works? What fails? What can we learn from a country that had to figure this out because geographic constraints left no alternative?

Singapore answers the “is this possible?” question. Not perfectly. Not completely. But actually.

On November 24, 2014, Prime Minister Lee Hsien Loong stood up and said Singapore would become the world’s first “Smart Nation”—not a city with some clever apps, but an entire country rewired for the digital age. More than a decade later, with over S$1.6 billion in government AI funding alone, Smart Nation 2.0 in full swing, and a new S$1 billion National AI R&D Plan (2025-2030) announced, Singapore offers the closest real-world approximation to the governance systems we’re proposing.

Why Singapore? Because nature forced the experiment.

The city-state occupies 728 square kilometers—smaller than New York City, about the size of Austin. It has zero natural resources. One percent of its land is agricultural. It imports over 90% of its food. Its population of six million generates 16 million daily journeys on roads that cannot expand because there’s nowhere to expand to.

Other countries can sprawl their way out of problems. Singapore must engineer its way through them.

This constraint-as-opportunity dynamic created the world’s deepest integration of AI into governance. Singapore didn’t adopt smart city technology because it was fashionable or because a consultant sold them a PowerPoint deck. It adopted it because the alternative was gridlock, food insecurity, and collapse.

By 2017, Singapore topped the Global Smart City Performance Index, ranking first in all four measured categories: mobility, healthcare, public safety, and productivity. Smart Nation 2.0, launched October 1, 2024, kicked off a wave of acceleration: a S$1 billion National AI R&D Plan through 2030, a National AI Impact Programme targeting 10,000 enterprises and 100,000 AI-trained workers by 2029, a new Digital Infrastructure Act for cloud and data center resilience, and Asia’s first AI governance framework, introduced back in 2019. Budget 2026 doubled down with a national AI council chaired by PM Lawrence Wong, expanded tax deductions for AI spending, and autonomous bus trials launching mid-2026.

But rankings are vanity metrics. The interesting question is: what actually works when AI meets governance at scale?

Traffic: The AI That Already Rules You

Let’s start with something everyone can understand: traffic lights.

You probably think traffic lights run on timers. Red for sixty seconds, green for forty-five, repeat forever. That’s how most of the world works. It’s also insanely stupid when you think about it—a system designed in the 1920s, applied to 2020s traffic patterns, optimized for precisely nothing.

Singapore doesn’t do stupid.

The GLIDE System: Every Light Listens

The Green Link Determining (GLIDE) system controls every traffic signal in Singapore—approximately 2,700 intersections managed by 18 regional computers, each capable of handling 250 junctions. Wire sensors beneath road surfaces detect vehicles in real-time. Local controllers adjust signal timing second by second to favor directions with higher traffic volume.

The result is what traffic engineers call “the green wave”: if you travel at optimal speed, you can pass through multiple consecutive intersections without stopping. It’s not magic—it’s network optimization at scale, something no army of human traffic cops could achieve.

What makes this interesting for governance theory: the system operates on a hierarchy:

Local controllers make immediate intersection decisions (milliseconds)
Regional computers coordinate multiple intersections (seconds)
Central monitoring oversees network-wide patterns (minutes)
Human operators intervene only for exceptional circumstances

Fast AI decisions at the local level. Slower human judgment at the systemic level. This is the “AI as referee, humans as philosophers” principle in operation—not in a Blueprint, but on actual roads with actual cars.

The Numbers That Matter

Between 2017 and 2022, Singapore’s daily travel demand nearly doubled—from 9 million to 16 million journeys. The country made a deliberate decision not to expand its existing 3,300 kilometers of roads. Instead, it bet everything on intelligent management of existing infrastructure.

Did it work?

Metric	Result
Peak-hour delays	↓ 20%
Average rush hour speeds	18 → 21 km/h (+15%)
Commuting time savings	$500 million/year in economic value
Public transport ridership	↑ 25% since 2020
Bus/train waiting times	↓ 2-3 minutes at peak hours
Total congestion savings	~$1 billion/year

These aren’t projections from a pitch deck. They’re measured outcomes from a decade of operation.

The Unexpected Bonus

Here’s the part that should make you rethink infrastructure entirely.

The IoT-enabled cameras deployed for traffic management also reduced crime in high-traffic public spaces by 8%. That wasn’t the plan. The system was designed to count cars, not catch criminals. But once you’ve deployed general-purpose sensing infrastructure, it serves purposes the designers never imagined.

The same sensors that detect traffic jams can spot accidents, track pollution, identify infrastructure failures, and—yes—deter crime. Infrastructure that serves multiple purposes simultaneously is infrastructure that scales economically. This is the foundational logic of the Foundation economy: build once, enable many.

The Next Step: Autonomous Vehicles (2026)

Singapore is now pushing this logic further. In mid-2026, the Land Transport Authority will launch autonomous bus service on Routes 191 and 400 in the financial district—16-seat driverless buses connecting Marina Bay and Shenton Way. In Punggol, Grab and WeRide are running autonomous shuttle routes that opened to the public in Q2 2026, linking residents to supermarkets, clinics, and transit hubs. The government aims for 100-150 self-driving vehicles by year-end, expanding island-wide over five years.

The driver shortage forced this. Singapore can’t hire enough bus drivers, so it’s replacing the bottleneck rather than fighting it. The GLIDE traffic management backbone that coordinates 2,700 intersections now serves a dual purpose: optimizing flows for both human-driven and autonomous vehicles on the same roads.

Vertical Farming: Where Reality Checks the Blueprint

Now let’s look at a case where Singapore is failing—because honest analysis requires it, and because the failure teaches something that matters.

The Audacious Goal

In 2019, the Singapore Food Agency set what looked like an inspiring target: produce 30% of nutritional needs locally by 2030, up from less than 10%. For a nation that imports over 90% of its food from a land area the size of a large airport, this required science fiction to become agricultural policy.

The solution? Grow up instead of out.

Sky Greens Farms opened the world’s first commercial vertical farm in 2012. The engineering is genuinely elegant: 2,000 rotating aluminum towers standing 9 meters tall, hydraulically driven, requiring only 60 watts per tower daily—roughly equivalent to a single light bulb. Annual yield reaches 800 tonnes per hectare, 5-10 times traditional farming output.

Singapore now holds 19.2% of the Asia-Pacific vertical farming market. Companies like ComCrop, Citiponics, and Sustenir Agriculture have built multi-story farms using hydroponics and LED lighting. The government has poured S$28 million into R&D through the ACT fund and established a S$60 million fund to help producers scale.

The Sobering Reality (2025)

And yet.

As of late 2024, only 3% of vegetables consumed in Singapore were locally sourced—down from previous years. Seafood production dropped 14%, with sea-based farms contracting from 98 to 72 facilities. The bright spot? Egg production surged to 35% of national consumption, exceeding its target.

In November 2025, the government formally killed “30 by 30.” Environment Minister Grace Fu announced Singapore Food Story 2, a replacement strategy built on four pillars: targeted local production of protein and fiber, import diversification, stockpiling, and global partnerships. The new targets are more modest: 20% of fiber and 30% of protein locally produced by 2035—a five-year delay with narrower scope.

Translation: we set a goal that technology could achieve, but economics wouldn’t cooperate.

Why This Failure Matters

Here’s the governance lesson that makes this case essential.

Vertical farming works. The technology produces more food per square meter than any farming method in human history. Sky Greens proved the concept a decade ago. The farms are running. The vegetables are growing.

But Singapore exists within a global economic system optimized for cost minimization, not resilience or locality. When Malaysian farms can sell lettuce for half the price of Singaporean vertical farms—because they have land, labor, and energy cost advantages—the market chooses Malaysia.

The lesson for abundance infrastructure: Technology alone doesn’t create transformation. You can have the most advanced vertical farms in the world, but if economic incentives point toward cheap imports, the technology sits unused. This is exactly why the Unscarcity framework doesn’t just propose new technology—it proposes new economic structures that change what’s “rational” to choose.

This is precisely why the Unscarcity Framework proposes separating baseline provision (the 90%) from market incentives (the 10%). If food security is a constitutional right rather than a market outcome, different choices become rational. You don’t ask “which lettuce is cheapest?” You ask “which food system ensures we can eat if shipping lanes close?”

Singapore’s vertical farming struggle illustrates what happens when transformative technology exists but economic incentives point elsewhere. The technology was never the bottleneck. The system was the bottleneck.

Singpass: Digital Identity Without the Dystopia

“But wait,” I hear you saying. “Sure, traffic lights and vertical farms are one thing. But identity? Government ID systems are surveillance systems. That’s Orwell territory.”

Singapore’s Singpass system is the counterargument.

Adoption That Wasn’t Forced

Genuine adoption looks like this:

4.5+ million users covering 97% of residents aged 15 and above
Access to 2,700+ services across 800+ agencies and businesses
350 million+ transactions per year
A new document wallet letting users store and present digital versions of official IDs
Economic impact of $385 million (per Deloitte study)—potentially over $1 billion at full adoption

This isn’t marginal uptake in a pilot program. It’s near-universal adoption, achieved without legal mandate, through design excellence.

The Architecture That Prevents Surveillance

What makes Singpass remarkable isn’t the adoption numbers—it’s the privacy architecture.

The system operates on what technologists call a “privacy firewall”: Singpass holds your identity and attributes but never sees them in action. When you authenticate to a service, your biometric verification happens separately from your personal information. No customer biometric data crosses to service providers. The bank knows you’re you. It doesn’t get your fingerprint.

This separation isn’t a policy promise that could be revoked. It’s architectural—built into how the data flows. You’d have to rebuild the system to violate it.

The implications matter enormously:

Single identity, multiple services: One credential for everything eliminates the fragmented patchwork of logins that creates security vulnerabilities
User-controlled data sharing: You authorize what gets released to whom
Continuous improvement: Unlike physical IDs, the digital-first design enables constant security updates
Legal framework alignment: 2024 amendments to the Computer Misuse Act made it illegal even to voluntarily share credentials for fraud

What Singapore Proves

The common objection to digital identity—“it enables surveillance”—assumes that identity systems must be surveillance systems. Singapore’s architecture proves otherwise.

Verification can happen without tracking. Authentication can protect without exposing. The Relational Identity concept in our Framework isn’t a fantasy—it’s an architecture choice. Singapore made one set of choices. Other nations could make the same choices.

The technology doesn’t dictate the outcome. The design dictates the outcome.

Five Lessons for the Commons

Singapore’s decade-long experiment offers concrete lessons for post-scarcity governance:

1. AI as Infrastructure, Not Authority

Singapore doesn’t use AI to make policy decisions—it uses AI to execute policies more effectively. GLIDE doesn’t decide whether to prioritize cars over buses; humans make that choice. GLIDE implements the choice at superhuman speed and consistency.

Traffic lights don’t govern. They coordinate. The distinction preserves democratic accountability while capturing efficiency gains.

2. Measure Everything, Fail Publicly

Singapore publishes its failures alongside its successes. The embarrassing 3% vegetable figure appears in official communications. The decline in local food production gets acknowledged, not buried.

Systems that cannot admit failure cannot learn. The Five Laws’s commitment to “Truth Must Be Seen” (Axiom II) follows this principle: governance that hides its mistakes eventually becomes ungovernable.

3. Architecture Beats Policy

Singapore’s digital identity system doesn’t promise privacy—it enforces privacy through technical design. The privacy firewall isn’t a regulation that could be changed; it’s infrastructure that separates data flows structurally.

For the Diversity Guard and other Unscarcity mechanisms: design systems that cannot violate principles, rather than systems that promise not to.

4. Universal Infrastructure Enables Innovation

Singapore’s universal connectivity, identity, and transportation create a platform on which private innovation builds. The 800+ businesses using Singpass didn’t have to create their own authentication systems. They built on government infrastructure.

This is the Foundation economy at work: universal provision of coordination infrastructure frees resources for Frontier innovation. The 90/10 Framework generalizes this from identity and transportation to all Foundation goods.

5. Constraints Produce Solutions

Singapore’s innovations emerged from scarcity, not abundance. Limited land forced vertical farming experiments. Limited roads forced traffic optimization. Limited resources forced efficiency.

Post-scarcity systems face the opposite problem: when constraints disappear, what drives innovation? Singapore suggests the answer: ambitious targets and meaningful challenges can substitute for survival pressure. The Ascent economy’s mission-based incentives follow this principle: purpose, not desperation, drives human excellence.

The Caveats (Because Intellectual Honesty Requires Them)

Singapore is not a model to copy blindly.

Democracy: Singapore operates as a one-party-dominant state with limited political competition. Its ability to implement wall-to-wall digital systems reflects both technical competence and concentrated authority. Whether similar systems can deploy in more pluralistic democracies remains unproven.

Surveillance risk: The “Lamppost-as-a-Platform” initiative with extensive facial recognition has drawn criticism. The privacy firewall of Singpass doesn’t extend to every AI system deployed citywide.

Public buy-in: Critics note the Smart Nation initiative “lacks a clear success story” in public imagination. Technical success doesn’t automatically produce popular understanding.

Inequality: Singapore’s income Gini coefficient fell to a record low of 0.379 (after taxes and transfers) in 2025, but its wealth Gini is estimated at 0.55. Smart systems optimize existing distributions; they don’t automatically create equitable ones.

These limitations don’t invalidate the lessons—they contextualize them. Technology-augmented governance requires not just technical architecture but social legitimacy, democratic accountability, and attention to distributive outcomes.

The Laboratory Continues

Smart Nation 2.0 launched in October 2024. Since then, the pace has only accelerated. In 2025, Singapore committed S$1 billion to AI R&D through 2030 and opened the Punggol Digital District—the country’s first fully sensor-integrated smart district, where an Open Digital Platform manages everything from air conditioning to autonomous robots across 50 hectares. Budget 2026 created a national AI council chaired by PM Wong, launched the National AI Impact Programme to reach 10,000 enterprises and 100,000 workers, and started autonomous bus trials in the financial district. SME AI adoption jumped from 4.2% in 2023 to 14.5% in 2024; among larger firms, it hit 62.5%.

Singapore continues iterating, experimenting, and publishing results.

For the Unscarcity Framework, Singapore demonstrates three points that matter:

AI governance scales. A nation of six million can be coordinated through intelligent systems without collapse, chaos, or rebellion.
Privacy and efficiency coexist. Technical architecture can deliver both coordination benefits and individual protection—if designed correctly from the start.
The Foundation economy works. Universal infrastructure enables, rather than constrains, innovation. Shared systems don’t eliminate markets—they provide the foundation on which markets operate more effectively.

Singapore isn’t a utopia. It’s a laboratory. The experiments continue, the data accumulates, and the results inform what comes next.

The question isn’t whether AI can govern. Singapore proves it already does—millions of times daily, at every traffic light, in every digital transaction, across every optimized system.

The question is whether we’ll design that governance to serve human flourishing, or let it emerge by accident and serve whoever built it first.

Singapore made its choice. The rest of us are still deciding.

Sources

Article generated for the Unscarcity Project — demonstrating real-world AI governance for the Abundant Commons

How Singapore Uses AI to Run a Country of 6 Million (2026)