AI for Materials: Why We Back Software in the US and Products in China
CMVC invested in Citrine Informatics in the US and InnoAero in China. Both apply AI to materials development, but they capture value differently: Citrine sells a software platform; InnoAero develops, manufactures and sells alloys. This perspective explains how customer readiness, data access and manufacturing economics shape our investment decisions. This expanded article incorporates Min Zhou’s presentation at the AICM breakfast meeting on 2 April 2025.
Start with what the customer will buy
AI can help researchers navigate the many possible compositions and processes involved in developing a material. But a promising model is only part of an investment case. We also need to understand who will pay, what they are buying and how the company can grow beyond its first customer.
An established chemical company may want software that helps its own scientists develop better formulations. Another customer may simply want a wire that is thinner, stronger or more reliable. The first can support a software business. The second can support a materials business that uses AI internally.
Our investments in Citrine and InnoAero reflect these different routes to commercial value.
The US platform opportunity: make industrial knowledge more useful
In our assessment of the US opportunity, established materials and chemicals companies offered a promising customer base for specialist software. Some have substantial experimental knowledge, research teams and processes for buying and deploying enterprise technology.
Citrine’s platform combines materials data infrastructure with AI to help customers develop and optimize products. It also helps preserve research knowledge so that it can be reused. Citrine’s company overview
The customer continues to make and sell the material. Citrine supplies a tool that can improve the customer’s development work. If that tool becomes part of a recurring workflow and proves useful across further projects, it can support an enduring software relationship.
That opportunity comes with a demanding commercial test. Industrial sales cycles can be long, implementation takes work, and AI for materials must compete for budgets. A compelling demonstration is not enough. We look for evidence that scientists use the platform, customers see value and adoption extends beyond an initial project.
China’s product opportunity: sell the result of better discovery
Our China assessment led us toward a different emphasis.
The industrial ecosystem offers access to manufacturing capabilities, suppliers and opportunities to make and test materials. These can help a startup connect computational discovery with a physical product and a customer application.
Selling a standalone software platform can be harder when customer data is fragmented, trust around data access is still developing, or purchasing favors individually negotiated projects. A business can win research assignments without building a repeatable software model.
A product company takes another route. It uses AI to develop a material, then earns revenue by supplying that material. Customers can assess the product against their requirements without first committing to a new research platform.
The investment question becomes whether better discovery can translate into a product advantage that survives qualification and manufacturing at scale.
InnoAero: the alloy is the business
InnoAero illustrates this approach. The company uses AI to discover new metal alloys, then makes and sells the alloys.
Our investment case focused on high-performance copper alloys and a demanding application: wiring for satellites. Conductivity alone is insufficient. The material must also meet mechanical requirements while supporting thinner, lighter wires.
The development process connects customer requirements with models, alloy recommendations, experimental preparation and property testing. Results feed back into the next iteration. AI helps guide the search; making and testing the material establishes whether a candidate works.
For us, the commercial appeal lies in connecting that capability to a product customers need. The company must still demonstrate consistent manufacturing, successful qualification and repeat demand. These are central to the investment thesis, alongside the quality of its discovery technology.
Different business models, the same investment discipline
This is a preference informed by the opportunities we see, rather than a fixed rule about geography. A Chinese software company with strong customer adoption can be attractive. A US materials company with a credible manufacturing route can be attractive too.
For a platform, we examine data access, integration into the customer’s work and the potential for recurring use. For a product company, we examine performance, qualification, production economics and the ability to supply reliably.
In both cases, we ask where the company captures the value it creates. Citrine captures value through software used by materials producers. InnoAero captures it through materials developed with AI.
That is why we can invest in the same broad technology through different business models. Our task is to identify the form of company best equipped to turn scientific progress into a business customers will sustain.