Etched
Custom hardware solutions for AI inference
CORE INFO
Etched is a semiconductor startup that develops custom hardware solutions specifically designed for AI inference, focusing on transformer-based models to enhance performance and efficiency for large language models.
- Etched raised $500 million in a Series B funding round, achieving a $5 billion post-money valuation.techcrunch.com
- The company has secured over $1 billion in customer contracts, highlighting strong market demand for its AI hardware solutions.globenewswire.com
- TechCrunch reported on Etched's emergence from stealth, noting its significant funding and customer contracts.techcrunch.com
WHY WE WOULD WORK AT ETCHED
Innovative Technology
Join a team pioneering transformer-specific ASICs that enhance AI inference efficiency, delivering high throughput and low latency for large language models.
Strong Financial Backing
Benefit from working at a well-funded startup with $800 million raised, including a recent $500 million Series B round, ensuring stability and growth.
Impactful Mission
Contribute to advancing AI infrastructure, making inference faster, cheaper, and more accessible, ultimately benefiting humanity through cutting-edge technology.
Dynamic Team Culture
Collaborate with over 400 talented professionals in a fast-paced environment that values innovation, creativity, and teamwork.
Proven Success
Be part of a company that has secured over $1 billion in customer contracts, showcasing strong market demand and validation for its products.
Career Growth Opportunities
Experience rapid career advancement in a startup that has achieved first-pass silicon success within three years, positioning itself as a leader in the AI hardware industry.
MARKET AND TRACTION
KEY METRICS
✦ KEY METRICGROWTH TACTICS
MARKET POSITION
NOTABLE CUSTOMERS
COMPETITIVE ADVANTAGE
PRODUCT AND TECH
Transformer ASIC
Etched's flagship product, the Sohu Chip, is a transformer-specific application-specific integrated circuit (ASIC) designed to optimize performance for large language models (LLMs), delivering high throughput and low latency for AI inference workloads.
Low Voltage Inference
This architecture operates math blocks at reduced voltage, enhancing the density of floating-point operations per second (FLOPs) and improving power efficiency, which is crucial for scaling AI inference capabilities.
Cluster Scale Memory
A shared memory pool that enhances memory access speeds and reduces latency across the scale-up domain, enabling more efficient processing of AI workloads and improving overall system performance.
AI Inference Solutions
Etched focuses on providing custom hardware solutions specifically tailored for AI inference workloads, making AI processes faster, cheaper, and more accessible for various industries.
Production Capabilities
With a factory established in Taiwan and a data center at their San Jose headquarters, Etched is equipped to support large-scale production of their innovative AI hardware solutions.