Claude just got more generous, and that matters for anyone tracking citations

Anthropic doubled the five-hour Claude Code window for Pro and Max subscribers this week, and the reason it could is a SpaceX deal that, per Ars Technica, gives the company access to "more than 300 megawatts of new compute capacity." Most of the coverage is about developer usage caps. I want to think through what looser compute might mean for the answers Claude returns to everyone else, with the caveat that Anthropic hasn't said anything about retrieval behavior changing.

The compute-to-answer-shape question

But the specific mechanic Ars describes here is different from a partnership press release: Anthropic explicitly removed the peak-hours limit reduction on Claude Code and raised API limits for Opus.

Whether that translates into longer or more thorough answers in Claude.ai is the open question. Anthropic hasn't published anything tying the SpaceX capacity to consumer answer behavior, and I haven't seen a controlled before/after. So treat the rest of this as a hypothesis worth testing, not a finding.

The hypothesis

The interesting case for citations is whether a less-constrained Claude does more sub-query work per user question, and reads further into each result.

Again, I'm not claiming this is happening. I'm claiming it's the thing to watch.

What to do this week

Most teams have great pages for the head query and nothing for the follow-ups, and that gap is what a more generous Claude (if that's what we get) would start to expose.

This is also worth running again in a month to see if the answers themselves got longer or pulled from more sources. That's the actual signal.

The musk thing is a distraction

Ars spends a few paragraphs on Musk's about-face (he tweeted "No one set off my evil detector" after previously claiming Anthropic "hates Western Civilization").

Whether your site has anything worth reading past the first scroll is a separate question, and one the SpaceX deal doesn't answer for you.