Richard Seroter, Google Cloud’s Senior Director and Chief Evangelist, told the Google Cloud Next 2026 conference in Las Vegas that developer loyalty to AI coding tools “is at zero right now.” Google, consequently, doesn’t worry about which tool developers choose. They support all of them, so the choice doesn’t matter.

It’s a strategically convenient statement. And it’s partially true: the AI coding tool market moves fast enough that building loyalty around a specific product can seem irrational. The market shifted dramatically in the past few months alone. Codex crossed 3 million weekly active users in a matter of months. Claude Code added capabilities that were speculative roadmap items three months ago.

But “loyalty is at zero” hides something Google has no incentive to name.

What Google Is Measuring When It Measures Developer Loyalty

Seroter probably means developers don’t have emotional attachment to a specific tool in the traditional sense. The brand loyalty that used to form around Vim vs. Emacs, IntelliJ vs. VS Code, tools that shaped professional identity, isn’t there anymore. Today’s developers try a tool, use it for a few months, move on if something more promising appears elsewhere.

But “loyalty” and “switching cost” aren’t the same thing, and the distinction matters.

A developer who spent three weeks configuring Claude Code with custom project rules, codebase context, style conventions, and automated review workflows doesn’t migrate next Tuesday because Codex shipped an interesting feature. Not out of loyalty to Anthropic as a brand, but because the cost of rebuilding that configuration elsewhere is real and concrete. It’s measured in hours of work, in interrupted habits, in context lost and rebuilt from scratch.

According to The Pragmatic Engineer’s survey of 906 software engineers (February 2026), Claude Code is the most-loved tool with a 46% “most loved” rating. That’s not loyalty at zero. That’s structured preference that consolidated over time.

Loyalty Shifted. It Didn’t Die.

This is what Seroter’s statement misses: developer loyalty to AI tools hasn’t died. It moved from the tool to the workflow that tool made possible.

Someone who has used Claude Code for six months built an invisible layer of dependency. System prompts tuned for their specific codebase. Project rules in CLAUDE.md that reflect the team’s conventions. Operational habits that integrated with the tool until the distinction between “how I work” and “how Claude Code works” became blurred. That configuration isn’t generic with respect to the vendor. It’s specific to Claude Code: its conventions, its APIs, its interaction model.

I covered this in the piece on AI token billing as a lock-in structure: lock-in in AI tools doesn’t show up in signed contracts or exclusivity clauses. It shows up in accumulated configuration, in habits built month after month, in the invisible cost of starting over with a different tool.

Google doesn’t see this loyalty because it measures the surface: which tool you install, which you subscribe to, which you mention in a survey. It doesn’t measure depth: how much of that tool entered your daily process in a way that’s hard to reverse without real cost.

The TechMonk Take: Tacit Loyalty Is More Dangerous Than Declared Loyalty

Google says it doesn’t care which tool you use. Anthropic says it doesn’t want lock-in. OpenAI says Codex is open and interoperable. Everyone agrees: no vendor lock-in here.

Meanwhile, developers keep building increasingly customized workflows around specific tools. Companies pay monthly subscriptions to two or three different vendors, with dependencies nobody formally mapped in an architecture document. The risk is distributed but not zero. It’s just been made invisible.

There’s also a structural incentive problem worth naming. Every vendor in the AI coding space benefits from claiming the market is fluid. If loyalty is zero, then no one is locked in, which means no one is accountable for the costs that accumulate when a tool changes direction, raises prices, or degrades in quality. “Zero loyalty” is also a way of saying “zero responsibility.” The developer who built their entire review workflow around a tool they thought was commodity is left holding the migration cost alone.

Tacit loyalty, the kind that doesn’t show up in any survey but manifests the moment your preferred tool announces a pricing change or downgrades the features on your current plan, is the hardest to evaluate and manage. It doesn’t appear in Google’s market analysis dashboards. It doesn’t appear in Google Cloud Next presentations. It appears when the alternative becomes necessary and you realize how much you built on top of that tool, without having planned to.

The irony of Seroter’s statement is this: Google Cloud Next was a conference designed to convince developers to build more things on Google Cloud. “Loyalty is at zero” is a convenient premise for saying “you can start with us tomorrow, no exit costs.” But the developers who will build on Google Cloud tomorrow will accumulate exactly the same tacit loyalty there that they have elsewhere today.

Next time a vendor announces that developer loyalty is dead, it’s worth asking who benefits from believing it.