CI economics should include the machines you already own.

Abstract

Hosted CI bundles execution capacity, isolation, scheduling, storage and operations. Customer-owned machines reduce purchased compute while adding power, hardware, maintenance, queueing and developer time to the cost model.

This engineering note studies when local capacity and separately selected hosted capacity can reduce total cost per successful pipeline. The model defines the accounting assumptions; the evaluation protocol measures correctness, time-to-result and operating effort. Tbench Actions supplies the local execution layer, while hosted overflow and managed persistent caches remain outside the current plan.

Put your capacity to the test.

For repetitive workloads on suitable company hardware, compare a local execution policy with the best eligible hosted baseline. Agree on reliability and high-percentile completion-time targets before measuring.

Account for operational overhead, local contention, retries and peak capacity alongside avoided cloud spend. The result tells your team which jobs belong on each execution path.

Account for the costs that move.

Let N be jobs per month, L the jobs completed locally, t the hosted billable minutes per job, r the hosted rate, F the platform fee, and K the all-in local operating cost. For a homogeneous workload with the same residual-cloud performance:

K includes power, hardware depreciation, storage, maintenance, and developer disruption. Real comparisons also include provider allowances, cache and artifact costs, network transfer, and retries. Evaluate heterogeneous workloads as a sum of per-job costs.

The break-even condition.

Subtract the hybrid cost from the hosted baseline:

Under the stated assumptions, hybrid execution is cheaper if and only if F + K < L × t × r. The right side is the cloud cost avoided. Apply your measured runtimes and operating costs to evaluate the condition for your fleet.

Apply the model to your environment.

• Choose isolation appropriate to the workload’s trust class.
• Measure peak-time queueing as well as monthly worker-hours.
• Use independent release checks and branch protection.
• Represent relevant inputs and trust boundaries before reusing a cache.
• Include external model/API charges and related service costs.
• Use current provider pricing, allowances and measured hardware performance.

Separate orchestration from execution.

The GitHub integration preserves GitHub’s workflow engine. The independent Tbench protocol runs immutable approved job policies from a local terminal, an existing CI agent or a signed queue. Both use organization-owned capacity; neither changes the cost assumptions above into a guaranteed savings claim.

Use the 15-day approved organization trial to measure a workload your team actually runs. Local application verification has its own check policy and evidence report, separate from task benchmarking and official audit certification.