SEMI's mid-year forecast in July laid out a clear growth curve for the entire semiconductor equipment industry: test equipment, packaging equipment, wafer fab equipment, each with its own figures and growth rate. Read through the whole report and one term never shows up: IC programming equipment. That's not a gap in the reporting. It reflects something more basic — this category doesn't have an independent slot in the industry's standard measurement framework at all. That absence is worth more thought than any single growth figure in the report.
Start with what SEMI actually reported. The mid-year forecast, published July 14, 2026, put global semiconductor equipment sales at $165.9 billion for 2026, up 23.2% year-over-year — the industry's fifth consecutive year of growth, running from 2022 through 2026. SEMI expects that momentum to hold through 2028, when total sales are projected to reach $229.5 billion, marking seven straight years of growth by that point.
Break it down by back-end category and the numbers get more specific. Test equipment, after surging 55.3% in 2025, is projected to grow another 31.0% in 2026 to $15.3 billion. Assembly and packaging equipment is projected to grow 9.6% to $6.7 billion. SEMI attributes this growth to rising device complexity tied to AI, particularly demand for advanced packaging linked to HBM. By 2028, test equipment is projected to reach $20.8 billion, assembly and packaging $8.6 billion.
Nothing wrong with these numbers. What's worth noticing is how they're categorized. SEMI's back-end equipment framework centers on two categories: test equipment and assembly and packaging equipment. IC programmers and automated programming systems don't fall into either bucket, and they don't get broken out as their own line item. In other words, in a report that represents the industry's authoritative benchmark, IC programming equipment simply doesn't exist as an independent category in the mainstream market-sizing framework.
That doesn't mean programming equipment doesn't matter, or that the industry isn't growing. Growth in wafer fab equipment and packaging/test equipment reflects expanding wafer starts and a growing variety of ICs upstream — and every one of those ICs has to go through programming before it reaches a finished product. Programming volume should, in theory, track the broader semiconductor industry's expansion in lockstep. But "should be growing" and "can prove it's growing" are two different things. When a category never gets folded into an authoritative measurement framework, its value and scale become hard for anyone outside the industry to assess or compare objectively.
Worth pushing this one level deeper: why is IC programming equipment invisible in the first place?
One likely reason is industry structure itself. Test equipment and packaging equipment have long had a set of large, publicly traded companies with substantial market caps behind them — Teradyne and Advantest, for instance, in test equipment. Those companies' earnings reports and market positions give statistical bodies a reason to carve out an independent category in the first place. An agency typically tracks a category because it already contains companies large and transparent enough to make the tracking meaningful. In IC programming, a large share of participants are small and mid-sized specialized firms, and the business model itself spans everything from equipment sales to contract programming services. That business boundary is fuzzier than the cleaner lines around "test equipment" or "packaging equipment" — and that fuzziness is likely part of why the category never got tracked on its own.
A second angle is how the programming step itself gets positioned. In a lot of customer and media framing, "programming" gets folded into the broader category of "test" — after all, on many production lines, programming and functional test genuinely sit next to each other, sometimes even integrated into one station. But that framing conflates two fundamentally different things. Test answers the question "is this IC good or bad?" Programming answers a different question entirely: "did the data get written into this IC?" One is a judgment call. The other is the production act itself. That distinction carries real industry weight. A judgment call can be sampled — you can measure yield through spot-checking, and you can ease a bottleneck by adding more test stations. A production act, once it goes wrong, has already consumed the full material cost of that IC. A firmware error can't be fixed by running more tests afterward — the unit just gets scrapped. The two differ in cost structure, equipment architecture, and how they affect yield. Lumping them together lets test's louder voice in the industry drown out programming's independent value.
Worth being upfront here: the analysis above is drawn from industry observation and structural reasoning, not from a study some authoritative body has specifically run on "why doesn't IC programming equipment have its own measurement category." If this read is off, industry voices are welcome to push back — but the question itself is worth putting on the table seriously.
The two reasons laid out above — historical statistical inertia from being folded into "test," and the industry's own lack of a language for communicating externally — sit at two different levels. They run in parallel, and they don't contradict each other. The first explains why programming equipment can't find a place in the existing measurement framework; that's a historical cause. The second is a challenge the industry still has to face on its own, even if the measurement framework eventually gets updated: even if someone agreed to track IC programming equipment as its own category tomorrow, what metric would they actually use? That question doesn't have an answer yet either.
If IC programming wants a seat at the table alongside test equipment and packaging equipment in industry narratives, it may need to answer a more basic question first: what metric would let an outside observer actually understand this industry's growth and value?
Borrowing "equipment sales revenue" — the metric test equipment uses — probably isn't the right path. In IC programming, equipment sales and contract programming services often run as parallel business models side by side. Looking at equipment sales revenue alone would understate the actual output value the industry creates. A few more specific capability metrics might get closer to what this industry actually is: how many ICs can be programmed per unit of time (UPH — a metric already used widely inside the industry, though almost nobody outside it knows what an industry-level benchmark actually looks like); how many IC types and protocols a platform supports; how long it takes to go from sample IC to production-ready adaptation once a new protocol launches. Right now, these metrics function mainly as internal engineering language — used by companies internally or by customers evaluating equipment — and haven't been assembled into a narrative framework that can tell outsiders clearly which direction this industry is actually progressing in.
Look at it from another angle: wafer fab equipment built its authoritative statistical standing largely because the whole industry converged on a shared technology roadmap — process-node iteration driven by Moore's Law, for instance — giving outside institutions a clear coordinate system to measure growth against. IC programming equipment still lacks a widely recognized "technology roadmap" narrative of its own. Protocol iteration (UFS moving from 4.0 to 5.0, for instance), the breadth of IC types and packages a platform can adapt to, gains in yield and efficiency — all of this progress is genuinely happening, but it hasn't been organized into a statistical language that outside institutions are willing to adopt. JEDEC's UFS standard iterations define the IC's interface specification. But "IC interface standard" and "programming equipment capability benchmark" operate at two different levels — one defines what the IC itself should look like, the other defines what level the equipment should perform at. Standardizing the latter is still a blank space in this industry.
That might be the real point of this observation: IC programming equipment isn't short on growth. It's short on a coordinate system that can actually explain that growth to people outside the industry. Whoever moves first on this — whether by pushing an industry association to establish a unified capability benchmark, or by getting out ahead with verifiable, comparable efficiency data communicated proactively to the outside world — gets the chance to make this long-invisible category visible for the first time to people outside it.
The semiconductor equipment industry just entered its fifth straight year of growth — a well-established backdrop, verified repeatedly by authoritative institutions. But when the official narrative behind that growth doesn't include the words "IC programming equipment" anywhere, industry practitioners might want to sit with a different question: it's not that this industry doesn't matter. It's that this industry hasn't yet found the language to prove to the outside world that it does.
For anyone thinking through where this industry stands right now, here's a concrete question worth asking: if someone asked you how much the IC programming equipment industry grew this year, do you have a defensible set of numbers you could actually give them as an answer?
