From Material to Application: A Functional Diamond Industrialization Map

Functional diamond is not short of possible directions. Companies are working on thermal management, quantum applications, semiconductor devices and optics, while material suppliers continue to pursue larger sizes, higher thermal conductivity and faster growth.

The harder question is what the industry should solve next. What the material can do, what the product should look like and what customers actually need are still being worked out in parallel. No common answer has yet emerged.

This reminds me of a thought I had while reading Chip War: once semiconductor technology became an industry, how did the industry know what to do next?

One answer is Moore’s Law. It is often treated as a natural rule that moves forward on its own and gives everyone a direction to follow. The causality was actually the other way around.

In Moore’s 1965 paper, the available data covered less than a decade. He initially projected that the number of transistors would double every year, then revised the interval to two years a decade later. What turned the curve into an industry-wide commitment was the National Technology Roadmap for Semiconductors, created by the U.S. Semiconductor Industry Association in 1992 and internationalized as ITRS in 1998.

The official purpose of the roadmap was to pose a challenge: if Moore’s Law was to be maintained, which technical capabilities would have to be developed next? It was not a natural law pushing everyone forward. The industry sat down, turned a plausible curve into a shared document and invested against it. Anyone falling behind could be seen.

That document could not govern the industry forever. As companies moved onto different process routes, the unified international roadmap was broken apart around 2015. Even semiconductors could not use one curve to coordinate the industry for decades. Its effectiveness depended on a relatively clean and extrapolatable physical relationship: continued chip scaling.

That precondition is the scarce resource.

Functional diamond may be missing more than a coordination exercise in which everyone agrees to write a common document. The industry has not yet agreed on the relationship between size, thermal conductivity and cost. That is why this article does not attempt to draw a roadmap saying what wafer size or market value the industry will reach in 2030. It is not that such a chart would be undesirable. It is that a chart drawn now would probably not survive contact with the evidence.

The supply-chain map developed earlier answered the question: who occupies which position in the chain? This article asks the next question: how does a material move from “good performance” to an industrial product that customers purchase repeatedly?

1. Start with the application, then work backward to the material

One recurring detour in semiconductor history is that companies first prove how large or how high-performing they can make something, and only later ask who might buy it. Functional diamond faces the same risk. Announcements such as “we can make six inches” or “we have reached 2,000 W/m·K” attract attention, but they are still some distance from the question of why a customer would have to use diamond.

The first layer of the map is therefore the application itself. Every candidate use should pass at least four tests:

• Is the pain point genuinely severe? The question is not whether diamond performs well, but what the customer cannot solve without it.
• Does diamond offer clear added value compared with copper, SiC, AlN, microchannel cooling or other available solutions?
• What stage is the customer actually at: research interest, sampling, joint development, qualification, small-batch supply, repeat purchase or inclusion in a product specification?
• Can the application generate continuing demand, rather than a story with a large but imaginary market?

Thermal management is a field I have studied in greater depth this year. The usual story is that diamond has extremely high thermal conductivity and is therefore suitable for cooling. Once the problem reaches an application, however, it becomes specific very quickly: which device problem is it solving, do existing solutions genuinely fail, what form should diamond take in the system, how much can system temperature actually fall, and how much will the customer pay for that improvement?

Customer testing can also send unexpected information back to the material supplier. The required size may not be as large as expected, the highest possible thermal conductivity may not be necessary, and the actual bottleneck may lie somewhere else. Quantum, electronic-device and optical applications probably have similar dynamics, but I have not studied them at the same depth and will not make a comparable judgment here.

Materials can sometimes push applications from the other direction. Once a particular size, purity or growth rate is achieved, an engineer may notice a use that had not previously seemed possible. It is not always demand first and material second.

But even when technology opens a new possibility, repeated validation on the application side is what turns that possibility into a product that someone actually buys.

2. Once the application is defined, the material requirements are derived backward

Only after the first layer confirms that the customer has a real unsolved problem and that diamond may create system value does it make sense to work backward to the material metrics.

When evaluating diamond thermal-management products, I do not look only at bulk thermal conductivity. I also look at interfacial thermal resistance, coefficient-of-thermal-expansion matching, total system thermal resistance and long-term reliability.

Together, these determine whether diamond can actually reduce junction temperature in a specific device and by how much. Comparing only which supplier reports the highest thermal-conductivity number can distract from the real bottleneck. The limiting factor may not be the bulk material at all. It may be how diamond bonds to other materials or whether it can survive repeated thermal cycling.

Material specifications are not defined by material companies in isolation. They are pushed backward from a concrete application. A thermal-conductivity result of 2,000 W/m·K may have limited meaning in an application dominated by interface resistance, while it may be decisive in another.

Comparing beautiful performance numbers without naming the application can send the industry’s effort in the wrong direction.

3. Industrial support capability comes after the material–application relationship is clear

This layer is sometimes compressed into the sentence that consistency becomes a commercial issue after applications begin to scale. But the statement hides a very concrete problem.

In the MPCVD industry, I still see companies whose production consistency depends on experienced operators watching the equipment and making visual judgments. The process has not yet reached the stage where data does most of the talking.

There is another difficulty: the same displayed parameters can correspond to very different physical states in tools from different manufacturers or with different structures. The readings may be identical while the material that grows is not.

What turns this from a question of whether an engineer should be more meticulous into a business requirement is the first two layers. Only when the product definition becomes clearer, orders begin to repeat and customers stop accepting “this batch is good, the next one is different” does batch consistency become a direct condition for accepting larger orders.

Why Diamond Cannot Yet Draw a Semiconductor-Style Roadmap

If I were asked to draw a functional-diamond roadmap today, I could not produce a version that confidently states wafer size, thermal conductivity and market value in 2030.

Semiconductors can produce that kind of roadmap because they have accumulated a physical relationship that the whole supply chain recognizes, together with a shared document that industry participants genuinely created and funded against. Diamond does not yet have either foundation.

Perhaps the first questions should be simpler:

1. Which material capabilities has diamond already demonstrated?
2. Which meaningful products and applications are those capabilities beginning to create?
3. What evidence would show that an application has moved from its current stage to the next one?

The resulting map will not look particularly polished. Many entries may have to say “to be validated.” Some technologies have good performance but no settled product form. Some have been sampled but have no repeat purchases. Some can be supplied in small batches but remain too expensive for a commercial market. And some genuinely important applications may not have been noticed yet.

One of the most useful ideas I took from Chip War is that a technology revolution is never completed by one astonishing parameter. Technology becomes an industry through repeated movement among technology, products, markets and manufacturing, with each correcting the others until a system can keep operating and learning. It also requires shared trust in that system.

Functional diamond is probably still at an early stage of that process. The industry has not even reached agreement on where it should invest together. The most useful map today may therefore not be a forecast of who will win, but a continuously updated record of where the industry stands and what must be proven next.