An exchange with peers across the strait

Today I had an exchange with peers from across the strait — a life insurance company. A few things stood out:

1. They actually know very little about us, though they are aware that we are working on localization (domestic substitution).
2. Their cloud services rely heavily on GCP, Azure, and AWS; domestic cloud providers have essentially no B2B market there.
3. Their model services rely heavily on ChatGPT, Claude, and Gemini; domestic models also have no B2B market — there even seems to be a relevant restriction, meaning they would rather use American ones than domestic ones.
4. Their regulators require that external service providers keep their infrastructure and data onshore. (These providers have data centers locally on the island.)
5. Moving to the cloud or using external models also requires regulatory applications, communication, and public relations, which is costly in time.
6. They are at an early stage of applying AI within the enterprise. From the conversation, they lag us by roughly three years and are still in the concept-and-planning phase.
7. They give each developer USD 150 per month in token allowances — something we simply cannot do. We are still in the “pay-to-work” stage.
8. They don’t seem eager to apply AI to business processes. Instead, they care more about AI-ops, hoping over time to converge on engineering methods suited to their own company — which seems characteristic of Western methods. This differs from our habits: domestically we seem more anxious, eager to find shallow proof at the value level, such as quantified cost-reduction and efficiency gains and commercial value.
9. Their core business systems are still planning service-oriented refactoring — well behind us.
10. During the exchange, I noticed that almost all their professional terms for AI, and even for IT, are in English. At one point they couldn’t figure out how to translate a term (though the corresponding domestic term is very common), which shows how deep the cultural influence runs.
11. Besides an IT department, they also have a data department, and the internal authority over AI currently rests with the data department; the two are still squabbling. As for the reason, judging from the conversation, it’s probably because the boss sees data, training, and AI as more closely linked — but doesn’t understand that “data” here means data in the broad sense, not just the small amount of data within the company (their business volume isn’t large).
12. A lunch costing 30–40 RMB (about 150–190 NTD) is already considered expensive for them.