Depot – Weekly Recap

Depot is a high-performance continuous integration provider, and this weekly summary reviews notable technical and product developments relevant to engineering teams and investors. During the week, Depot emphasized deep infrastructure optimization, enhanced observability in its CI tooling, and improvements to its registry caching and serverless orchestration capabilities.

A key focus was aggressive performance tuning of Depot CI microVMs, where cold-boot times were reduced from roughly seven to nine seconds to sub‑second levels in testing. By refining kernel configuration, trimming systemd services, and tightening cloud‑init behavior, initial boot times were brought down to around three seconds, setting the stage for further gains.

Depot then replaced cloud‑init for configuration delivery using Cloud Hypervisor’s fw_cfg interface and introduced a custom init system tailored for ephemeral VMs. Additional optimizations included reducing console logging, tuning kernel clock and timer parameters for KVM, and backing memory with 1 GB hugepages, helping achieve median boot times near 0.6 seconds and P90 around 1.2 seconds.

These microVM gains point to a strategy centered on lowering CI latency and infrastructure overhead, which could enable faster build cycles and higher throughput per host. If reliably deployed at scale, the optimizations may improve Depot’s unit economics and strengthen its competitive position among cloud‑native CI and DevOps infrastructure providers.

Beyond core performance, Depot expanded its command‑line tooling with real‑time log streaming via a “depot ci logs –follow” option and structured NDJSON exports. A new “depot ci metrics” command and enriched JSON status views are designed to give teams granular visibility into CPU, memory, and run‑level data, supporting better troubleshooting and automation.

The company also upgraded registry pull‑through caching, adding presets for Docker Hub, GHCR, GAR, and ECR, plus reusable upstream credentials to simplify configuration. These enhancements aim to reduce friction for container‑heavy workloads and could increase Depot’s relevance in cloud‑native development pipelines by improving reliability and performance of image pulls.

In serverless orchestration, Depot continued to highlight its “Durable Lambdas” approach for running CI workloads on AWS Lambda beyond the typical 15‑minute limit. By checkpointing state, suspending between events, and resuming via callbacks, the platform seeks to handle complex workloads more efficiently while observing AWS operational constraints.

The architecture splits responsibilities between a Run Lambda and per‑workflow Lambdas, using timeouts to detect stalled jobs and reduce workflow hangs and unnecessary polling. This event‑driven design is intended to improve scalability and cost control, particularly for engineering‑heavy organizations running large or bursty CI workloads.

Taken together, Depot’s recent updates underscore a consistent focus on performance, observability, and operational efficiency across its CI stack. The week’s developments suggest the company is investing in differentiated infrastructure capabilities that, if widely adopted, could support customer retention, higher usage, and a stronger competitive stance in the evolving developer tools market.