What's Happening?
Arteris, a Nasdaq-listed company specializing in network-on-chip (NoC) technology, has announced surpassing 1,000 customer semiconductor design starts using its interconnect intellectual property. A significant finding from their report, 'The State of
Semiconductor Innovation: Insights from 1,000 Design Starts,' reveals that AI-related projects constituted 50% of their design starts between 2021 and 2025. This marks a substantial increase from just 4% in the 2010-2014 period. CEO K. Charles Janac emphasized that AI chips necessitate moving vast amounts of data, leading Arteris to develop wider data pathways and support higher operating frequencies. The company's NoC technology is crucial for connecting components within chips and between chiplets, addressing the growing complexity of modern chip architectures driven by AI, data centers, and automotive systems. Data centers, cloud computing, and other high-performance computing applications also saw a rise, accounting for 43% of design starts in 2021-2025, up from 11% in 2010-2014.
Why It's Important?
This shift in Arteris's design starts underscores a fundamental transformation within the semiconductor industry, driven by the escalating demands of artificial intelligence. The dramatic increase in AI-related projects highlights the critical need for advanced chip architectures capable of handling immense data movement and processing. This trend impacts various U.S. industries, particularly autonomous vehicle manufacturing, edge-based robotics, and advanced medical imaging diagnostics, which require ultra-low latency and massive energy efficiency. Companies that can provide solutions for efficient data movement within chips, like Arteris, are poised to gain significantly. Conversely, traditional chip manufacturers focused solely on general-purpose processing may face challenges if they do not adapt to these specialized requirements. The automotive sector, in particular, is a major driver, as autonomous driving systems demand real-time data processing for safety and functionality, making in-memory computing chips and efficient interconnects indispensable.
What's Next?
The increasing complexity of AI systems suggests a continued demand for advanced interconnect technologies and a move towards more integrated system architectures. Arteris anticipates that 'Physical AI,' involving autonomous machines interacting with their surroundings, will generate the largest number of future design starts. This implies further innovation in chip design to support real-time processing and decision-making in physical environments. The industry may also see a greater emphasis on 'advanced packaging,' which is evolving from a backend assembly step into a critical architectural component influencing memory bandwidth, thermal density, and power integrity. This will likely lead to more vertically integrated infrastructure systems, where silicon, packaging, and infrastructure behavior are continuously simulated from the earliest design stages. Companies that can effectively integrate these elements will be at the forefront of the next generation of computing.
Beyond the Headlines
The evolution of chip design, as evidenced by Arteris's report, signifies a deeper shift in how computing systems are conceptualized and built. The traditional separation of chip design, packaging, board design, and thermal engineering is collapsing, with AI systems requiring a holistic approach. This re-evaluation of system architecture echoes historical computing challenges, such as the SAGE air-defense computer system from the 1950s, which was effectively a large-scale 3D integrated system. The current trend suggests a 'return of physical computing,' where factors like power, thermals, cooling, and communication distance are becoming dominant system problems. This could lead to a new system-level design layer in the semiconductor industry, where continuous simulation across all these elements becomes standard, moving beyond silicon-centric design to optimize for workload movement, memory locality, and infrastructure orchestration. This paradigm shift will necessitate new skill sets and collaborative approaches across engineering disciplines.
