Google's Credibility With Developers Keeps Declining
Two separate HN threads today about Google's relationship with its developer base, and neither was kind. The 'Google Antigravity bait and switch' thread described what sounds like a product being changed or killed mid-adoption, with a commenter asking the question everyone is thinking: how did Google blow their AI lead? Why is GCP still not a real challenger to AWS? The answer in the thread is blunt: because Google cannot stop shooting its own customers. The 'IBM-ification of Google' post made the same point more formally, comparing Google's current trajectory to IBM's long decline from developer-beloved to enterprise-tolerated.
The pattern here is consistent across months of HN threads. Google kills products, changes APIs, sunsetting things without warning, and has now done it enough times that developers price this risk into their architecture decisions. The Gemini system prompt leak thread ('Gemini randomly dumped its system prompt') added a reliability concern on top of the trust concern.
The counterpoint worth noting: Google still has deep distribution, the best infrastructure at scale, and Gemini is genuinely competitive on benchmarks. But benchmarks do not matter if builders do not trust the platform to exist in two years. That trust is the thing Google is burning.
So what?
If you are evaluating cloud AI providers for a production system, Google's bait-and-switch pattern is a real business risk, not just a developer annoyance. Vendor lock-in to a platform with a history of abandonment is a liability. Build with exit paths in mind, or choose providers with stronger platform stability track records.