> ## Content Index
> Fetch the complete content index at: https://top-margin.ghost.io/llms.txt
> Use this file to discover other available public pages before exploring further.

# In Defense Chips, Qualification Is the Product
- URL: https://top-margin.ghost.io/in-defense-chips-qualification-is-the-product/
- Published: 2026-10-05T10:59:21.000Z
- Updated: 2026-10-05T10:59:21.000Z
- Author: Marcus Cole

Four semiconductor awards landed in September and the largest of them was sixteen million dollars. A defense electronics house took that sum to re-qualify a radiation-hardened process, with the wafers run at a commercial foundry in upstate New York and the packaging and test done in Virginia. A research agency put seven million into a second phase of work on cooling gallium nitride chips by growing diamond into channels behind the device. A competitor on the same programme reported more than a fourfold gain in radio frequency power density and moved to its own second phase. The Army picked a power semiconductor company to develop ten kilovolt silicon carbide switches.  
  
Set those numbers against the fabs. A leading edge plant costs tens of billions. These four awards together would not buy a month of construction.

### From the Battlefield to the Balance Sheet

The mismatch is the point, and it resolves as soon as you see what is actually being purchased. None of this money buys capacity. It buys qualification, which is the documented demonstration that a specific part made by a specific process on a specific line behaves the way a system designer needs it to, under radiation, under heat, over years. Capacity without qualification sells nothing into these systems. Qualification without much capacity sells a great deal.  
  
That inverts the usual semiconductor economics. In the commercial world scale is the moat, and the company with the bigger fab wins on cost. In this corner of the market the moat is a qualified part number with flight heritage or radar heritage, and it is defended by the switching cost on the customer's side. Requalifying a system around a different component is measured in years and in test campaigns, not in purchase orders, which is why a small supplier with an approved part can hold pricing that its volumes would never justify.  
  
It also explains the structure of the radiation-hardened award. The wafers come from a commercial foundry that has no interest in this volume, the value-adding steps sit with the defense house, and what is being restored is the ability to offer customisable parts on a mature process rather than a single bespoke chip. The asset is the process qualification and the design library on top of it. The fab is a service.

### **The Dual-Use Reality Check**

Two of these four technologies are civilian businesses with a defense first customer. Gallium nitride at high power density is the amplifier in a radar and also the amplifier in a base station, and the thermal problem being solved with diamond is the same thermal problem limiting power electronics everywhere. Ten kilovolt silicon carbide is not really a weapons component at all. It is the switch in grid scale converters, in traction and in the medium voltage equipment that heavy industrial loads and charging infrastructure need.  
  
That is the familiar pattern and the money is flowing the familiar direction, with the defense customer paying for the hard qualification step because it is the customer that cannot wait. The commercial market inherits a proven device it did not fund.  
  
The radiation-hardened line is the exception that is quietly becoming a rule. Hardened parts used to be a small captive market for national security spacecraft. The commercial constellations now going up in large numbers need the same tolerance in their higher orbit and higher reliability variants, and that demand did not exist when this capability was allowed to lapse.

### **The Capital Signal**

The capital signal is that in specialty semiconductors the scarce asset is an approved part, not a factory, and the government has worked out that it can buy approval for single digit millions when buying plants costs billions. Expect more awards of this size and shape, aimed at the qualification step rather than at tonnage of silicon.  
  
The structural read is to look through the headline numbers in this sector. A company announcing a large capacity investment in a commodity process is making a commodity bet. A company announcing a small award to qualify a device on an existing line is buying an annuity, and the second is worth more per dollar committed. Watch for the second phase awards and the completed qualifications rather than the groundbreakings, because a programme that moves to phase two has cleared a technical gate that most do not, and a qualified part has cleared the only gate the customer cares about. The unglamorous news is where this money actually turns into revenue.  
  
*Signal: Defense chip money is buying qualification rather than capacity, and a qualified part number is the most defensible asset in the sector.*

![](https://storage.ghost.io/c/e3/0e/e30efc82-f19f-427f-b131-8005b5e7416a/content/images/2026/10/in-defense-chips-qualification-is-the-product-photorealistic-1.jpg)

Marcus Cole, Top Margin