Waymo chose Singapore because Singapore wrote the test
Waymo is skipping New York and Chicago for Singapore in 2028. The reason isn't market size. It's that Singapore published the test you have to pass — and that's a lesson you can copy.
The Waymo Singapore robotaxi launch is set for 2028, and the timeline matters more than the headline. Cars start arriving in "the coming months." Mapping and autonomous testing with human safety drivers happen in 2027.
So a company with roughly 4,000 vehicles in about 15 US cities is committing to two years of unpaid groundwork in a single city-state. The Verge reported the plan on 18 September 2026. Why it chose Singapore is the useful part.
🗺️ Two years of unglamorous work before the first paying ride
Here's the published sequence, laid out plainly:
| Phase | When | What actually happens |
|---|---|---|
| Vehicles land | "coming months" (from Sep 2026) | Cars physically shipped in, no passengers |
| Mapping + supervised testing | 2027 | Autonomous driving with a human safety driver |
| LTA approval | Before any passengers | Safety assessment at CETRAN |
| Public service | 2028 | "Fully autonomous ride-hailing to the public" |
Nothing in that table is a model improvement. It's shipping logistics, HD mapping, a supervised data-collection phase, and a government safety assessment. The hard part of deploying this system into a new market is not the driving policy. It's proving the driving policy to somebody who gets to say no.
Key takeaway: Waymo's Singapore schedule is 100% integration and compliance work. The model is done. The approval isn't. That ratio shows up in almost every serious deployment, and most of us budget for it backwards.
🧪 Singapore didn't win with subsidies. It won with a test rig.
This is the sentence in the reporting I keep coming back to: Singapore requires all autonomous vehicles to pass a safety assessment at the Centre of Excellence for Testing and Research (CETRAN) before they touch a public road. Those tests were developed by the Land Transport Authority (LTA) and CETRAN, with input from the Traffic Police.
Read that as an engineer and it's a conformance suite with a named owner, a physical test facility, and sign-off from the people who deal with the consequences. It's not a committee that reviews your pitch deck. It's a thing you either pass or fail.
And it has clearly been exercised on real workloads before robotaxis showed up. Singapore has already approved:
- Autonomous freight delivery
- Autonomous road-sweepers
- Self-driving buses for airport workers
Motional, the company majority-owned by Hyundai, ran a robotaxi pilot there as far back as 2016. So autonomous driving in Singapore is not new. Somebody was testing robotaxis there a decade before this announcement.
Compare that with where Waymo still can't operate. Per the report, regulatory hurdles and political opposition have kept it out of New York City, Chicago, and Washington, DC — three of the largest, richest, most transit-dependent markets in its own country.
| Market | Status for Waymo | Reason, per the reporting |
|---|---|---|
| New York City | Blocked | Regulatory hurdles, political opposition |
| Chicago | Blocked | Regulatory hurdles, political opposition |
| Washington, DC | Blocked | Regulatory hurdles, political opposition |
| Singapore | Launching 2028 | Published safety assessment; seat on the AV steering committee |
A smaller market with a written test beat three enormous markets without one. Waymo isn't chasing addressable population. It's chasing a decidable process.
📊 The expansion map tells you the same thing twice
Singapore is not a one-off. Look at the announced international queue alongside the domestic one:
| City | Target | Note |
|---|---|---|
| London | end of 2026 | Soonest international launch |
| Tokyo | 2027 | First Asian market |
| Munich | late 2027 | First EU market |
| Singapore | 2028 | First Southeast Asian market |
| Nashville, Denver, San Diego, Las Vegas | announced this year | US markets |
Four international cities on the board while three of America's biggest cities stay closed. Waymo co-CEO Tekedra Mawakana framed Singapore's appeal in her statement as its "reliability, sustainability, and human-centric progress." I'd translate that less diplomatically: the rules are knowable in advance, so the engineering effort has a defined finish line.
There's also a detail that should not be skipped over. Waymo set up a corporate entity in Singapore earlier this year and it participates in Singapore's Steering Committee on Autonomous Vehicles. It was in the room where the standard gets discussed before it announced a launch date. That's not lobbying in the grubby sense. That's finding out what the acceptance criteria are going to be while they're still being written.
🛠️ How to steal this if you're building from Sri Lanka
None of us are shipping robotaxis. The transferable part isn't the cars, it's the posture. Three things I'd actually change in how I work, based on this:
- Write the acceptance test before you need it. If you want a bank, a ministry, a university, or an enterprise customer to adopt what you built, the unblocker is almost never a better demo. It's a test they can run themselves and watch pass. Hand them a script, a fixture set, and a pass threshold. You will be the only vendor who did.
- Treat evaluation as a deliverable, not a chore. If your system classifies, detects, ranks, or predicts anything, publish the numbers with the definitions attached. Say precision and recall on which set, at which threshold. Our confusion matrix calculator and IoU calculator are free and exist for exactly this — you have no excuse for "it works well" as a claim.
- Get into the room early, even when the room is boring. Waymo took a seat on a government steering committee before it needed approval. The local equivalent is an industry working group, a standards consultation, a procurement pre-bid meeting, an RFC thread. Being present while the criteria are drafted is worth more than arguing after they're fixed.
Bottom line: A market with a published, passable test is easier to enter than a bigger market with an unwritten one. That applies to countries, and it applies to your next customer.
💡 What this means for you
If you're a student or an engineer here watching this news and wondering whether it creates jobs nearby — realistically, a 2028 Singapore launch means mapping, fleet operations, safety-driver, and evaluation roles appearing in the region over the next two years, mostly in Singapore itself. Worth tracking, not worth planning your degree around.
The more immediate value is the method. Waymo's Singapore bet says that the cost of entering a market is dominated by how legible its approval process is, not by how big it is. If you're a small team choosing which client, sector, or country to chase, rank your options by how clearly someone has told you what passing looks like. The ones that can't answer will eat years.
And when you're the one asking others to trust your system, be the side that writes the test down. It's the cheapest credibility you will ever buy.
AI-assisted draft, reviewed and approved by Induwara Ashinsana before publishing. Sources are linked inline; if something here is wrong, tell me and it gets corrected.