Apple Pay Transit in Only 12 US Cities? China Has 336 — Here’s Why That Gap Matters

https://www.macrumors.com/2026/04/28/apple-pay-transit-us-cities
https://www.octopus.com.hk/en/consumer/octopus-cards/products/cross-border/china-t-union.html

That’s actually how I read the MacRumors article yesterday: only 12? What happened?

Apple holds more than half the US smartphone market (and under 20% in China). Yet after all these years, only 12 US transit systems support Apple Pay — Atlanta, the Bay Area, Boston, Chicago, Los Angeles, New York, Orange County, Philadelphia, Portland, San Diego, Seattle, and Washington, DC. The article frames this as respectable by name-dropping a handful of international peers — London, Paris, Hong Kong, Tokyo, Toronto, Beijing, and Shanghai — which makes the US look like it’s keeping pace.

It isn’t. The picture changes fast once you look beyond the headline cities.

In China, Apple Pay works for transit in over 300 cities, because virtually every transit network sits on the unified T-Union card system. Octopus in Hong Kong even sells a single card that works across 336 cities in mainland China and Macao — one tap, one standard, one rail of infrastructure. Beijing and Shanghai aren’t two cities on a list; they’re two of hundreds.

Android isn’t disadvantaged there either. Huawei Pay, Xiaomi Pay, and the other domestic wallets all plug into the same T-Union rails. Google Pay is the one that doesn’t work directly, but the workaround is well-known: load a Hong Kong Smart Octopus onto an Android phone and you’re set across the border.

The wider point is that the US isn’t actually the best place for tech adoption. Americans are great at inventing things; they’re often slower at deploying them at scale. Plenty of markets are hungrier — and faster — when something genuinely useful shows up.

Even AI, where the assumption is that the US is the engine, is more complicated than it looks. It took years — and this week’s renegotiated Microsoft–OpenAI agreement, which loosened the original exclusivity — before frontier US models were officially running multi-cloud. Anthropic got there earlier, with Claude available on AWS Bedrock, Google Cloud Vertex AI, and (now) Microsoft’s platforms. In China, the rhythm is completely different: any new open-source release from DeepSeek, Z.ai, Moonshot, or Qwen tends to land on every major Chinese cloud within days or weeks. Distribution is treated as table stakes, not a negotiation.

Inventing the future and adopting it are two different muscles. The US is unmatched at the first. It’s worth being honest about how often it lags on the second.