Your chip passed the evaluation. So why is nobody designing it in?

A founder I know shipped eleven evaluation boards in his first year. Eleven. Each one felt like progress: a real company, a real engineer, real lab time booked. Eighteen months later, zero of those evaluations had turned into a design win. Not because the chip failed. Because an evaluation is not the start of a design win. It is the start of someone else's hobby.

The semiconductor industry has a polite fiction that evaluation leads to design-in leads to revenue. It does, sometimes. But the path is so long and so easily abandoned that treating an eval request as pipeline is how startups starve while looking busy.

An evaluation is an option, not a commitment

Understand what an eval board is from the customer's side. It costs the engineer a few hundred euros and a few afternoons. It buys them knowledge: what your part can do, whether it is worth remembering, whether it might matter for a future architecture. It is a cheap option on a decision they have not scheduled.

The engineer evaluating you is usually not the person who will design you in. They are scouting. The design decision belongs to a product team with a roadmap, a budget, and a launch date, and that team may not know you exist. Until the eval connects to a named product with a date, you have a science project, not a deal.

The design-in window is the only clock that matters

Chips do not get swapped into shipping products. They get designed into new ones, early, while the architecture is still soft. Once the schematic freezes, you are out until the next generation, which might be two or three years away.

So the first question in any semiconductor opportunity is not how is the evaluation going. It is: what product is this for, and when does its architecture freeze. If the answer is vague, the opportunity is vague. If the freeze is in four months and your part is not yet in the reference design conversation, you are already too late for this generation and should plan for the next one.

The evaluation dies without an internal owner

Most evals fail the same quiet way. The engineer gets pulled onto a priority project. The board goes in a drawer. Your follow-up emails get answered slower, then not at all. Nobody says no, because nobody ever said yes.

The defence is to force ownership early. Before you ship the board, ask: who owns this evaluation on your side, what does success look like in numbers, and which product team would use it if it works. An evaluation with a named owner, pass criteria, and a waiting product team is a design-in opportunity. An evaluation without them is shelf decoration. If they cannot name the product team, sell to the product team first.

Your support is part of the silicon

Here is the uncomfortable part. At the evaluation stage, big customers are not only testing your chip. They are testing you. Can a twelve-person company answer a layout question in a day? Will the errata sheet be honest? Will someone pick up the phone in year three?

This is where small vendors win or lose against the established names, because the established name is worse on merit but safer on paper. Answer questions fast, document like a company ten times your size, and be straight about limitations. The engineer championing you is spending internal credibility. Every slow or vague answer costs them some.

Bridge the years between design win and revenue

Even when you win, you wait. Design win this year, production in two, meaningful volume in three. A startup that counts design wins as success will run out of money celebrating. The same trap as the OEM design win, compressed into a smaller package.

So build the bridge deliberately. Charge for NRE and customisation where you can. Sell evaluation platforms and early engineering samples at real prices, because paying customers behave differently from evaluators. Keep a pipeline of next-generation design-ins running so that when one program slips, another arrives. And watch your customer's end market, not just their engineering: a design win into a product that flops is a very expensive trophy.

The honest measure of your pipeline

Strip your semiconductor pipeline down to one question per account: is there a named product, with a design-in window, and a person inside whose job depends on filling it. Yes means a real opportunity, count it. No means an option, value it accordingly.

Eleven eval boards on eleven desks is a marketing channel. One chip designed into one shipping product is a business. The discipline is refusing to confuse the two, and it is the same discipline we apply at SLSCRW to every pipeline we touch: evidence over enthusiasm. If your evals are not converting and you cannot say why, that is exactly what we dig into.