Building the materials foundation for AI: Syensqo on advanced materials and AI as a two-way driver
Synopsis
This MIT Technology Review Insights conversation produced in partnership with Syensqo records the views of Mike Finelli, Syensqo's chief technology and innovation officer and chief North America officer: AI is pushing semiconductors and data centers toward physical limits, which piles up more simultaneous requirements on advanced materials, and Syensqo is responding by developing materials for high-voltage data center architectures, advanced sealing materials for semiconductor manufacturing, and thermal-management solutions including direct immersion cooling fluids, while working with Microsoft on AI agents that digitally synthesize millions of candidate molecules, predict their performance through physics-based simulation, and rank them down to roughly a hundred candidates for laboratory
Interpretation
The article frames material demand with a "performance pyramid" and an "and, and, and principle": as requirements such as high temperature, purity, electrical performance, chemical resistance, plasma resistance, and long-term stability stack up, materials move toward the top of the pyramid of high-performance specialty materials. Against the common narrative that credits algorithms, compute, or data centers for AI progress, it positions advanced materials as an equally critical enabling layer that is "actually increasingly defining what's going to be possible." This is an executive's industry framing and conceptual vocabulary drawn from the conversation; the text offers no quantitative comparison or experimental data.
Syensqo is developing three material directions for next-generation AI infrastructure: materials for high-voltage data center architectures, high-performance sealing materials used inside semiconductor fabs and wafer tools, and thermal-management solutions including fluids for direct immersion cooling. The article ties each direction to a concrete setting: high-voltage architectures can enable greater computing power while improving energy efficiency, seals must withstand "aggressive plasmas, reactive chemicals" with lower outgassing and higher purity demands, and air cooling is described as inefficient and energy intensive. This is a company's own account of its R&D directions; the text provides no performance metrics, test conditions, or third-party validation.
The article highlights cross-industry transfer: insulating polymers developed for electric vehicle bus bars, dielectric-fluid direct immersion cooling know-how, and a lithium-ion battery cathode binder can be carried over to higher-voltage, higher-energy-density data centers and to energy storage systems. It explicitly maps materials and thermal-management knowledge accumulated in automotive electrification onto AI data centers and the storage needed alongside renewables. This is the interviewee's experience-based account; the text gives no measured results from the transferred applications.
On research method, Syensqo partnered with Microsoft so that AI agents digitally synthesize "millions and millions" of candidate molecules, use physics-based simulation to predict performance plus toxicity and sustainability, and rank them, leaving roughly a hundred candidates to be synthesized in the lab, which Finelli says lets the team go "broader, deeper, and faster." Rather than the traditional path of picking a small candidate area from expertise, literature, and patents and iterating through trial and error, the described workflow covers the whole candidate space first and then converges on a small experimental set. This is a company's own description of its workflow; the text does not disclose model details, hit rates, or comparisons against the conventional process.
Perspective
The article is aimed at business and technology readers interested in the intersection of AI infrastructure and advanced materials, and its scope is limited to Syensqo's own product directions and R&D practice, including high-voltage data center materials, semiconductor sealing materials, direct immersion cooling fluids, a sustainable portfolio management tool, and an AI-assisted molecule screening workflow. It is useful for understanding how a company treats materials as an enabling layer for AI, how it puts sustainability at the start of research, and how automotive experience can be carried into data center settings.
A careful reader would still want to know the actual laboratory hit rate among the roughly hundred AI-ranked candidates, how well physics-based predictions match measured performance, lifetime and reliability data for high-voltage architecture materials and sealing materials under real operating conditions, and the specific improvement of next-generation low-global-warming heat transfer fluids over today's fluids. The text mentions that "88% of our portfolio now is a sustainable product" without describing how that share is defined or bounded. The article is also partnership-produced content, so its statements come from the company side and can be cross-checked against independent technical sources.