The Counterargument
Tessica's Take
More Choice in One Place Is How AI Gets Cheaper
Responding to Alexis McCurdy's op-ed: Google Cloud Will Sell OpenAI's Newest Models. Big Customers Go First.

Alexis is right that this starts with the biggest customers. But she's underestimating the problem it chips away at.
Pick a Cloud, No Longer Stuck With Its AI
Until now, picking a cloud company often meant getting stuck with its AI. That's called "lock-in," and it mostly helps the seller. Now Gemini, Claude and OpenAI's models can sit side by side, so the best tool for each job can win.
Matching each job to the right model is also how AI gets cheaper. Astra costs the most to run, so a simple task shouldn't have to pay for it. And it's not only for big companies: Ars reports anyone can request early access, software builders can connect even from OpenAI's own coding tools, and even individual users could see some benefit. Programmers using Antigravity also get another model to choose from, and better tools for them mean better apps for the rest of us.
Tell People Which AI Is Talking
Alexis worries about who's in the room, and it's a fair worry: these models come from a few powerful companies. The test is whether access opens up quickly beyond the biggest customers. And when the system swaps which AI is answering you, you should be told. Being open about that is what makes the savings worth trusting.
A bigger menu isn't power for users yet. But AI models competing for every job is where that power starts.