Laravel Cloud's Managed Queues Now Scale to Zero

Laravel Cloud picked up another piece of its scale-to-zero story at Laracon US 2026: managed queues that spin worker capacity down to nothing when there's no work to do, and back up the moment a job lands.

The problem this solves

Queue workers are a strange fit for always-on billing. Most apps have queues that are bursty by nature — a rush of jobs after a bulk import, a quiet stretch overnight, a spike during a marketing send — but a traditional managed queue worker runs continuously whether there's one job a minute or zero jobs an hour. You end up paying for idle capacity the vast majority of the time to be ready for the rare bursts.

Laravel Cloud's Flex compute already brought scale-to-zero to web apps earlier in the year, sleeping and waking application instances in under 500 milliseconds. This extends the same underlying model to queue workers specifically.

How it behaves

When a Laravel Cloud managed queue has been empty for a configured idle period, worker capacity for that queue scales down to zero — you stop paying for compute that isn't processing anything. The moment a new job is pushed, capacity scales back up automatically to pick it up. For most workloads, that wake-up happens fast enough that job latency stays within a reasonable window even after an idle period, rather than requiring you to keep a worker warm just in case.

This is configured per queue connection in your Laravel Cloud environment settings — no change to how you dispatch jobs, no change to config/queue.php, and no code differences between a queue that scales to zero and one that doesn't. From the application's perspective, Order::dispatch($job) behaves identically either way.

Why this matters beyond the bill

Cost is the obvious win, but there's a secondary benefit: it removes the incentive to under-provision workers to save money during quiet periods, which is a common cause of jobs backing up right when a burst actually arrives. With scale-to-zero, you're not choosing between "pay for idle capacity" and "risk being under-provisioned when traffic spikes" — the platform handles both ends of that range for you.

Where to be careful

Scale-to-zero queues are a poor fit for latency-sensitive job types where even a short cold-start delay matters — a job that needs to complete within a second or two of being dispatched to keep a user-facing flow feeling instant, for example. Laravel Cloud lets you configure idle thresholds and, where needed, keep specific queue connections warm rather than scaling them to zero, so you can mix the two strategies: scale-to-zero for your bulk, non-urgent queues, and always-on capacity for the few that need to respond immediately.

Combined with the Flex compute announcement for web apps, it's a clear signal of where Laravel Cloud is headed: treating "nothing is happening right now" as a state that should cost nothing, across every part of the stack, not just the parts that are easy to sleep.