APOLLO: Advanced RRAM-Based AI Chip Technology
Opportunity
APOLLO is poised to capitalize on the rapidly growing AI chip market, projected to reach $400 billion by 2030. As AI becomes more integral to various industries, there will be a rising demand for efficient and scalable processors. APOLLO's integration of analog RRAM for computation and storage, alongside its energy efficiency, uniquely positions it to capture a large market share, especially in high-performance, low-power applications.
In the automotive industry, APOLLO's capabilities cater to the $60 billion semiconductor market (by 2028), driven by advanced driver-assistance systems (ADAS) and autonomous vehicles. APOLLO's ability to deliver fast and reliable data processing makes it an ideal solution for automotive advancements.
Additionally, the smart city market, expected to grow to $200 billion by 2025, presents further opportunities for APOLLO. Its low-latency, scalable architecture is suited for real-time data processing needs in traffic management, energy distribution, and public safety, enhancing its market appeal.
Technology
APOLLO advances neuromorphic computing by integrating resistive random-access memory (RRAM) for both computation and storage within a unified architecture. Traditional RRAM systems often face challenges with scalability, high latency, and complexity due to their analog design. APOLLO overcomes these issues by incorporating proprietary peripherals that integrate seamlessly with field-programmable gate arrays (FPGAs), streamlining design and customization. This adaptability makes APOLLO suitable for diverse computational tasks across various environments.
A standout feature of APOLLO is its remarkable energy efficiency, achieving 30 trillion operations per second per watt (TOPS/W), making it an ideal solution for power-conscious, next-generation AI applications. Additionally, its scalable design supports dual-core arrays of up to 128x128, allowing it to handle complex AI algorithms and large-scale data processing effortlessly. APOLLO’s combination of energy efficiency, scalability, and customizable architecture positions it as a game-changer in AI computing and neuromorphic processing systems.
Figure 1. Comparison between conventional crossbar sense architecture and proposed TDC based sensing architecture for compute and storage. a.) Integration of IP in CMOS technology. b.) Conventional ADC architecture is used to indirectly sense current after current to voltage conversion by integrators. c.) CMOS integration of sense architecture in front-end and crossbar in back-end
Download
