Services
Bring an AI workload, a hardware product, or a team learning AI. We will define one result and the work needed to reach it.
AI and people education work are available by inquiry. Touchdown Inference is in development. Our foundry is in engineering; physical work needs separate approval.
What we can do with you
01 / Hardware & Manufacturing
A semiconductor customer can start with a design, package, test, reliability, or manufacturing question without using our AI service. We can scope a report or qualification path; physical execution and production require separately qualified people, facilities, rights, funding, and release.
Scoped inquiry · no production fab today02 / AI infrastructure
Agree on a successful result, measure your current route, and compare the simplest changes first. Integration or ongoing operation is a separate agreed scope.
Selective intake03 / Education for people
Leaders and teams learn on a real task. They practice using AI, checking its output, finding failures, and improving the work when tools change. This teaches people; it does not train a model.
Available by inquiry04 / Model and compute engineering
If a measured AI task needs a model change, scope evaluation, adaptation or post-training with explicit data rights and a before-and-after test. Dedicated compute and alternative accelerators are separate decisions. Model pretraining is not part of team education.
Separately scopedOne accountable scope at a time. Existing providers, labs, foundries, and specialists can win the work. Your agreement defines Touchdown’s responsibility, outside costs, and who accepts the result.
Method and reporting
Two public examples show our tooling and reporting discipline. They are not proof of a completed customer outcome.
Evidence tooling, documentation, and current source.
See how evidence and limitations are reported.
View the report →Choose your path
Bring a chip or hardware question to the hardware inquiry, or an AI workload to the AI project intake. Ask about team education when people need to learn how to use and check AI. The self-service Inference product has a separate waitlist.