.png){kind=logo}
Yarden Elnir designs the physical infrastructure that determines whether next-generation AI systems can operate at real-world speed and efficiency. Artificial intelligence is often discussed in terms of models and algorithms, but its true limits are set by hardware. Computation only moves as fast as memory allows, and intelligence only scales as far as silicon permits.
As an expert AI ASIC design engineer, Elnir helps lead the shift from general-purpose GPUs to domain-specific ASICs built specifically for generative AI. He has noted that the industry has been attempting to run 21st-century reasoning systems on hardware architectures originally designed for 20th-century graphics. His work replaces that mismatch with purpose-built silicon optimized for dataflow, control flow, and systemwide efficiency.
From an early age, Elnir was drawn to the hard sciences. While many were captivated by software abstraction, he focused on the physics beneath it. “While software feels like magic, hardware is the physics that determines whether that magic can exist at all,” he says. That mindset guided him to Tel Aviv University, Israel’s leading research institution, where he earned a B.Sc. in Electrical and Electronic Engineering.
During his studies, Elnir joined an EU-funded research project on ratchet-based ion pumps. He helped develop custom measurement devices to validate new physical concepts for energy-efficient desalination. The work established a rare balance of theoretical rigor and experimental precision.
That balance led Apple to select Elnir for a highly competitive undergraduate role in its chip design organization. Before graduating, he was offered a full-time position with the Storage Silicon Group.
At Apple, Elnir helped define the flash storage controller architecture embedded across the entire product lineup. His work optimized memory access protocols, improving throughput and responsiveness on iPhones, iPads, and Macs. Because nearly every operation depends on storage performance, his contributions enhanced the daily experience of hundreds of millions of users worldwide.
He also played a key role in next-generation RISC-V computation units by engineering performance improvements by orders of magnitude. His work demonstrated mastery of Apple silicon and semiconductor engineering at both the architectural and system scale. That experience clarified what Elnir saw as the next defining challenge in hardware: not storage, but artificial intelligence.
After Apple, Elnir recognized that AI’s greatest bottleneck was not in algorithms but in infrastructure. Rather than continuing to layer increasingly complex software on legacy foundations, he chose to focus on redesigning the hardware itself.
Today, as a member of technical staff (RTL) at Etched, a unicorn startup valued at over $5 billion, Elnir is a core silicon engineer on the Sohu chip. He designs the central dataflow logic that feeds matrix multiplication engines, the heart of generative AI computation. His focus is on solving the Memory Wall, where powerful compute units sit idle while waiting for data. By building predictive, out-of-order control systems, Elnir hides memory latency and pushes hardware toward its physical efficiency limit. This work defines the next era of AI hardware.
Elnir’s expertise is recognized beyond the industry. He serves as a peer reviewer for Information Fusion, a top 1% computer science journal with an impact factor of 15.5. He has judged elite competitions such as MIT’s HackMIT and Stanford’s FAF Multimodal Hackathon. He was also selected for the Extraordinary Fellowship from more than 1,500 applicants, an acceptance rate of just 1.67%, marking him among the highest tier of emerging technical talent globally.
Elnir’s impact extends into mentorship and organizational leadership. He helps shape technical evaluation frameworks, mentors engineers, and ensures Etched recruits only top-tier talent. His system-level thinking applies as much to teams as it does to silicon. “My goal is to help build the physical infrastructure that makes advanced intelligence accessible, scalable, and sustainable,” he explains.
By redefining how data moves through silicon, Yarden Elnir builds the foundation for the post-GPU era, where hardware no longer constrains intelligence, but enables it by design.
