Microsoft is deeply entwined in OpenAI and has obvious reasons to dogfood, yet their people are using Anthropic solutions.

Valuation behemoth OpenAI has been forced by the market to use Anthropic standards a couple times, having no comparable solutions of their own.

… I can see it.

It's Web 2.0 all over again. No moat, winner-take-all (economies-of-scale/network-effect). Just have to out-spend everyone else, and then figure out whether it was worth it all after you win.

Having a cutting edge model that requires tens of billions of dollars to train + a massive concentration of talent and experience + brand + one of, if not the best, coding experiences in Claude Code

These are all moats.

Ain't that for the entire ai field.

FOMO, pretty much

They have a moat on hype.

At least from the software engineer pleb perspective, their moat is that their tools seem to work well more often than not. I wasn't comfortable with the idea of using GitHub CoPilot as our GenAI solution at work, and apparently that was a widespread feeling, because we switched to Claude Code, and it's been a relatively smooth transition from manual coding to GenAI agentic loops.

[dead]

[dead]