The San Francisco Bay Area has long been synonymous with semiconductor innovation, but the current cycle marks a distinct shift. Instead of incremental gains in general-purpose compute, a new generation of founders is building highly specialized chips, from AI inference accelerators to cloud-native CPUs and entirely new fabrication approaches. This shift is being driven by the explosive demand for AI infrastructure, supply chain realignments, and a growing recognition that software gains alone are no longer enough.
What makes this moment particularly compelling is who is leading it. The region’s chip ecosystem now blends seasoned executives from incumbents like Intel with first-time founders tackling foundational constraints in compute, memory, and manufacturing. Together, they are redefining how chips are designed, built, and deployed.
Renée James
Founder & CEO, Ampere Computing
Renée James founded Ampere Computing in 2017 with a clear thesis: cloud infrastructure needed a new class of processors purpose-built for modern workloads. A former president of Intel, James brought deep industry credibility and relationships into the venture, helping Ampere quickly gain traction among hyperscalers.
Today, Ampere’s ARM-based CPUs are positioned as high-performance, energy-efficient alternatives to legacy architectures. With major cloud providers adopting its chips and continued funding support, James has established Ampere as one of the most credible challengers in data center silicon, especially as AI workloads push infrastructure limits.
Gavin Uberti
Co-founder & CEO, Etched.ai
Gavin Uberti is part of a new generation of founders rethinking chip design from first principles. As co-founder of Etched.ai, he is building transformer-specific ASICs optimized exclusively for large language models, eschewing general-purpose flexibility in favor of extreme efficiency.
Etched has quickly captured attention across the AI ecosystem, raising significant funding and positioning itself as a potential disruptor to GPU dominance. Uberti’s approach reflects a broader shift: instead of scaling existing architectures, startups are now designing silicon around the exact structure of modern AI models.
Chris Zhu
Co-founder, Etched.ai
Chris Zhu co-founded Etched.ai alongside Uberti, contributing to the company’s core technical vision around specialized AI inference chips. His work focuses on translating the mathematical structure of transformers into hardware-level optimizations.
As demand for inference efficiency accelerates, Zhu’s role is central to Etched’s differentiation. The company’s approach of building chips for one model class rather than many could significantly alter cost-performance benchmarks across AI deployments.
Robert Wachen
Co-founder, Etched.ai
Robert Wachen rounds out the founding team at Etched.ai, focusing on engineering execution and scaling the company’s ASIC development efforts. His work ensures that the theoretical gains of transformer-specific hardware translate into real-world performance.
As Etched moves from concept to deployment, Wachen’s role becomes increasingly critical. Execution (not just architecture) will determine whether specialized AI chips can meaningfully compete with entrenched players.
Rochan Sankar
Co-founder & CEO, Enfabrica
Rochan Sankar is building Enfabrica at the intersection of networking and AI compute, two domains that are rapidly converging. The company develops high-performance connectivity silicon designed to support the massive data movement requirements of AI workloads.
Backed by strong investor interest, Enfabrica reflects a growing realization: compute is no longer the only bottleneck, data movement is equally critical. Sankar’s focus positions the company as a key enabler of next-generation AI infrastructure.
Jeff Holden
Founder & CEO, Atomic Machines
Jeff Holden, formerly Chief Product Officer at Uber, is now tackling one of the hardest problems in tech: how chips are physically manufactured. Through Atomic Machines, he is developing atomic-scale fabrication technologies aimed at dramatically improving precision and efficiency.
The company’s ambition goes beyond incremental improvements; it aims to redefine how semiconductors are built altogether. If successful, Holden’s work could reshape the economics and scalability of chip production in the years ahead.
James Proud
Co-founder, Substrate
James Proud, known for founding consumer hardware startup Hello, has re-emerged in the chip sector with Substrate. The company is pursuing alternative lithography techniques using particle accelerators, aiming to bypass traditional bottlenecks in chip fabrication.
With substantial funding and backing from prominent investors, Substrate represents one of the more ambitious bets in semiconductor manufacturing. Proud’s transition from consumer tech to deep tech underscores how critical the chip stack has become across industries.
Source: Financial Times reporting
Oliver Proud
Co-founder, Substrate
Oliver Proud co-founded Substrate alongside his brother James, focusing on advancing the company’s fabrication technologies. His work supports the development of next-generation lithography systems designed to scale beyond current industry limits.
As chip demand accelerates globally, Substrate’s approach could offer an alternative path to increasing production capacity, particularly as traditional methods face physical and economic constraints.
Thomas Sohmers
Founder, Rex Computing
Thomas Sohmers founded Rex Computing to rethink CPU architecture at a fundamental level. A Thiel Fellow, Sohmers has focused on memory-centric computing designs that challenge conventional assumptions about performance scaling.
Rex Computing’s work reflects a broader movement toward architectural innovation in chips, rather than reliance on process node improvements. Sohmers is among a small group of founders pushing entirely new paradigms in compute design.
Weili Dai
Co-founder, Marvell Technology
Weili Dai co-founded Marvell Technology and helped build it into a global semiconductor powerhouse focused on storage, networking, and data infrastructure chips. As one of the most prominent female founders in the industry, her influence extends beyond the company itself.
Marvell’s continued relevance in AI and data infrastructure underscores the long-term impact of its founding vision. Dai remains a key figure in the broader semiconductor ecosystem.
Pantas Sutardja
Co-founder, Marvell Technology
Pantas Sutardja, co-founder and CTO of Marvell, played a central role in the company’s engineering direction. His technical leadership contributed to the development of key semiconductor innovations in storage and connectivity.
As the chip industry evolves toward AI-driven workloads, the foundational technologies developed under his leadership remain highly relevant.
A Sector Rebuilding the Foundations of Compute
Across these founders, a clear pattern emerges: the semiconductor ecosystem is no longer focused solely on performance gains; it is rebuilding the entire stack. From specialized AI chips to new fabrication methods and networking architectures, these companies are addressing constraints that were once considered fixed.
This shift is why the Bay Area remains central to chip innovation. It is not just producing faster processors; it is redefining how computing itself is structured.
For more insights into emerging tech leaders, explore our related list on leaders in the semiconductor industry.
