- Light-based interconnects dramatically improve AI computing speed, energy efficiency
- Lam’s precision etch and dep tech make photonics manufacturable at scale
Every breakthrough in computing starts with materials engineering. From carving nanoscale vias that connect high-bandwidth memory to etching optical waveguides that guide light with atomic precision, materials engineering is what makes tomorrow’s technologies possible. In my role at Lam, I see this every day: the reactions inside our chambers determine how reliably, efficiently, and sustainably the world’s most advanced chips can be built.
Why Photonics, Why Now
As AI reshapes data centers and pushes energy demand to unprecedented levels, electrical connections made of copper are reaching their physical limits. Photonics offers a way forward: replacing electrons with photons to transmit data at light speed, using a fraction of the energy.
But making that vision manufacturable requires waveguides, photodetectors, and modulators with sidewalls so smooth that even atomic-scale roughness can scatter light. That’s not simply an engineering problem — it’s a materials engineering problem.
This is where our teams at Lam come in. With advanced etch systems like Kiyo® FX, we can deliver the ultra-smooth profiles photonics demands. Our materials engineering expertise in materials such as silicon nitride enables the fabrication of optical components with unmatched uniformity. And because the silicon photonic integrated circuits themselves enable technologies like co-packaged optics and optical interposer technologies, we’re helping ensure light can be guided and manipulated reliably, even in the future at a wafer-level scale.
By bringing materials engineering and photonics together, we are unlocking the next leap in AI computing: data centers that move data faster, consume less power, and scale sustainably. Just as Lam once enabled the transition to copper interconnects, today we are enabling the shift to light.
Materials Engineering Across the Next Wave of Technologies
Silicon photonics is only one example of how materials engineering underpins the future of computing. The same principles apply wherever the industry is hitting physical limits and demanding new breakthroughs.
Take high-bandwidth memory (HBM). Each stack depends on millions of through-silicon vias, etched with nanometer precision and filled flawlessly with copper. Without advanced materials engineering to control sidewall via smoothness and deliver exceptional conformal deposition of via liners, these dense connections would fail; AI processors would sit idle, waiting for data.
Or consider wide-bandgap materials like gallium nitride (GaN). Their efficiency makes them ideal for power management in data centers, but manufacturing them requires ultra-low damage, atomic scale precision etch and deposition process, wafer bevel engineering solutions, and strict contamination control. Our technology is already enabling GaN on silicon technologies at 200mm. As these migrate to 300mm Lam will have an increasingly important role to play.
Advanced packaging presents yet another frontier. As chips transition from monolithic dies to chiplets and panel-level designs, deposition and clean chemistries must work across diverse substrates, from silicon (Si) to glass, while maintaining adhesion, uniformity, and reliability. The same atomic-level materials engineering control that makes a photonic waveguide possible also ensures chiplets bond seamlessly and function as a single high-performance system. Indeed, the worlds of Si photonics and advanced packaging are increasingly linked with one enabling the future technology roadmap of the other.
Across all these domains — photonics, HBM, GaN, advanced packaging — the common thread is materials engineering. It is the hidden driver that enables new materials to be shaped, connected, and scaled into technologies that once seemed impossible.
Why Photonics Matters Most
Among all these frontiers, silicon photonics captures the stakes most clearly. To keep pace with AI’s explosive growth, data centers must replace electrical interconnects with optical ones — a shift as disruptive as the introduction of copper interconnects decades ago. But this isn’t possible without new materials engineering solutions. Only precise control of etching and deposition can produce waveguides smooth enough, modulators accurate enough, and materials uniform enough to carry light reliably at scale.
By solving these materials engineering challenges, Lam is enabling the world’s transition to a light-speed future. Just as we helped power the rise of copper and 3D memory, today we are pioneering the materials solutions that make photonics real. Together with advances in HBM, GaN power devices, and advanced packaging, photonics defines the next chapter of data center based computing, one where data moves faster, energy is used more responsibly, and innovation continues to scale.
As I look at where the industry is headed, one thing is clear: the future of AI depends on materials engineering. At Lam, we are committed to leading that journey.
David Haynes, Vice President, Specialty Technologies and Marketing, Lam Research