Beyond silicon computer chips

While silicon-based chips have driven technological breakthroughs since the 1950s, today’s demands from artificial intelligence (AI) and machine learning (ML) expose their limitations. AI applications often face an “energy and latency crisis,” as vast amounts of time and energy are consumed moving data between computing units and memory—a problem known as the "memory wall." At the same time, the benefits of shrinking silicon transistors have slowed, creating a “miniaturization wall” that necessitates new approaches to chip design and manufacturing.

Tathagata Srimani focuses on augmenting silicon with transformative technologies to create faster, more efficient computing systems, and he specializes in ultra-dense 3D integration of heterogeneous logic and memory.

“My research is about creating transformative NanoSystems by seamlessly integrating diverse nanomaterials and technologies for logic, memory, and sensing,” explains Srimani, an assistant professor of electrical and computer engineering. “My work has demonstrated how such integration can lead to unprecedented gains in energy efficiency and throughput.”

Historically, these NanoSystems were limited to academic prototypes due to challenges in scalability and manufacturability.

By using technologies such as carbon nanotube transistors (CNFETs), resistive RAM (RRAM), and monolithic 3D integration, his designs vertically stack and densely integrate logic and memory in 3D. By integrating logic and memory layers with fine-grained 3D connections, such designs significantly reduce data transfer distances and associated energy costs, offering substantial performance gains over traditional 2D silicon-only chips across a wide range of abundant-data applications, like artificial intelligence.

Teaching and fostering a robust semiconductor ecosystem is as important as developing transformative technologies.

Tathagata Srimani, Assistant Professor, Electrical and Computer Engineering

“My work has focused on transitioning key technologies from ‘lab-to-fab,’” Srimani says. “This effort required addressing issues across the technology stack—from material and device engineering to system design and architecture—resulting in scalable, manufacturable solutions within industrial silicon fabs.”

Srimani’s work has successfully translated CNFET technology and its 3D integration on silicon to U.S. fabs, including SkyWater Foundry and Analog Devices.

“While building 3D NanoSystem chips with carbon nanotube transistors served as a case study,” Srimani notes, “the insights gained are applicable to a wide range of emerging technologies and NanoSystems.”

Building on this work, Srimani’s newly formed NEXUS (Nanoelectronics EXpeditions for Ubiquitous Systems) Research Group, seeks to augment 3D NanoSystems by incorporating a wider gamut of materials and technologies, including magnetics and semiconducting oxides. His team is also focused on tackling critical challenges related to power delivery and thermal management within such 3D NanoSystems. To complement foundational material and technological advances, Srimani plans to create co-design frameworks that derive system architecture and technology targets based on application-specific energy and throughput needs. From an application standpoint, he aims to expand beyond traditional AI applications, exploring paradigms like probabilistic computing for complex optimization, developing hardware solutions powered by heterogeneous integration.

Beyond research, Srimani is committed to shaping the next generation of engineers. He teaches advanced courses in semiconductor devices and computing hardware design, including practical, hands-on experiences like “Hacker Fab,” where students build a semiconductor fabrication facility from scratch.

“Teaching and fostering a robust semiconductor ecosystem is as important as developing transformative technologies,” he notes.

“I’m excited to collaborate within Carnegie Mellon’s diverse and interdisciplinary environment to drive these innovations forward,” says Srimani, reflecting on the university’s unique strengths.