AI Pulse · Frontier Labs Watch

Anthropic is reportedly in talks to buy Decart AI for around $6 billion, which would be its biggest acquisition to date. Decart makes software that helps chips run AI workloads more efficiently, plus “world model” simulation tech used for things like autonomous driving and e-commerce. The company was valued near $4 billion just this past May, so this would be a big jump in a short time. Nothing is finalized yet, according to the original report.

Here’s why this matters beyond the price tag. Anthropic is already reported to be building its own in-house chip team. Add a potential Decart deal on top of that, and you’ve got two separate bets pointed at the same problem: the cost of running AI at scale. Compute is the biggest line item for any lab training and serving large models, and whoever solves that bottleneck cheaply gets a real edge. When a company this size makes two moves in the same direction within months, it’s a signal that they see compute economics as the thing that decides who wins the next few years, not just model quality.

For business owners, the direct impact is small today. You’re not buying chips or training foundation models. But it’s worth watching because efficiency gains at the lab level tend to show up later as lower API costs or faster response times for the tools you actually use. It’s the same principle at a smaller scale when we help clients set up their own AI operations. Getting the underlying plumbing right, whether that’s compute, data flow, or workflow automation, matters as much as picking the flashiest model. That’s the kind of thing we build into an AI command centre, and it’s worth thinking about even if you’re not spending billions to do it.