Rate Limits, Side Quests, and the Myth of a Clear Frontrunner

By Bryce Stuckenschneider | Published 2026-04-01 | Industry | 7 min read

It's April 1st, and I promise there's no prank here. Just an honest look at the AI landscape as Q1 2026 wraps up — rate limits, side quests shelved, and why the moment that makes your platform decision easy isn't coming.

If you've been waiting for one lab to pull away from the pack and make your platform decision easy, the last two weeks should be a wake-up call: that moment isn't coming. And it's not just about which model scores highest on benchmarks. It's about whether the infrastructure behind that model can keep up when the world shows up to use it.

Ask anyone who built workflows, agents, or daily operations on Claude how the last two weeks went. Max subscribers watching five-hour session windows evaporate in 90 minutes. Prompt caching bugs draining token budgets overnight. Peak-hour throttling that turned a $200/month plan into a "maybe try again after lunch" plan. Anthropic's own engineer acknowledged the changes publicly, noting that roughly 7% of users would hit limits they'd never seen before.

This isn't a hit piece on Anthropic. What happened makes sense if you zoom out. The Pentagon standoff drove a wave of new sign-ups. The Super Bowl campaign put Claude in front of millions. Claude Code went from developer darling to mainstream tool seemingly overnight. Stack all of that on top of each other and you get a hockey stick Q1 that no infrastructure team on earth could have perfectly anticipated.

Meanwhile, over at OpenAI, Fidji Simo held an all-hands meeting that basically said "we've been chasing too many side quests." In short order, they shelved the adult mode feature, shut down the Sora standalone app (along with a billion-dollar Disney deal), and declared an enterprise pivot.

So where does that leave operators like us? Paying attention. Staying flexible. Not going all-in on any single provider without a backup plan.

The model wars are one thing. The infrastructure wars are something else entirely. Both matter when you're building real systems for real businesses.

And if you're tracking what's next, keep your eyes open. Both labs have new models in the chamber. This race is far from settled.