Arteris has surpassed 1,000 customer silicon design starts using its interconnect intellectual property, the company announced on November 5, a milestone that signals a structural shift in chip development: moving data efficiently now carries as much weight as processing it.

The announcement accompanies a company report, The State of Semiconductor Innovation: Insights from 1,000 Design Starts, which analyzes changes across Arteris' customer designs. Its central finding is simple. Designs now contain more specialized processing blocks, and the network on chip — the communication infrastructure linking processors, memory and accelerators — must reconcile competing demands for bandwidth, latency, power efficiency and predictable behavior.

The commercial stakes are measurable. According to Arteris, networks on chip occupy 10% to 13% of chip and chiplet area in sophisticated designs. That share makes interconnect a substantial silicon-cost commitment, not a late-stage plumbing detail. The report argues that data-movement decisions belong at the start of development, alongside choices about compute resources and memory organization, because added processors deliver limited value if data cannot reach them.

How fast has AI transformed the workload mix?

The sharpest change in Arteris' dataset is the rise of artificial intelligence. Designs classified as AI-related grew from 4% of Arteris design starts in 2010–2014 to 50% in 2021–2025 — a twelvefold increase in share. The percentages describe Arteris customer activity, not the whole semiconductor industry, but they chart how dramatically the company's workload mix has turned over in roughly a decade.

The mechanics explain why interconnect vendors care. AI architectures repeatedly move large volumes of data between memory and processing elements. Communication bottlenecks throttle throughput, and unnecessary transfers burn energy. Interconnect design therefore determines whether added computing capability converts into useful performance within practical power budgets.

Enterprise computing shows a parallel shift. Data center, cloud and high-performance computing applications rose from 11% to 43% of Arteris design starts between the two periods. The commercial logic is straightforward: customers building expensive computing infrastructure need architectures that keep those resources productive. The announcement does not quantify resulting customer savings.

What is happening in automotive silicon?

Automotive adds a safety dimension to the intelligence trend. Arteris reports that approximately two-thirds of its automotive design starts carry a functional safety classification related to ASIL, the automotive safety integrity levels. Nearly half combine that safety classification with AI capabilities — evidence of growing overlap between intelligent processing and safety requirements in the company's automotive business.

For vehicles and other systems that interact with the physical world, raw performance is only part of the challenge. Designers must also address reliability and security. Arteris positions its interconnect, integration automation and hardware security assurance portfolio around those intersecting requirements. One caveat applies: a design's classification alone does not establish that a finished product achieved certification.

Which claims deserve scrutiny?

Arteris also reports a ninefold increase in measured design complexity over the last two years. The release does not explain the measurement methodology, so that figure stands as a company-reported indicator rather than a standardized industrywide measure.

The milestone itself requires the same precision. A design start marks development activity. It does not necessarily represent a completed tapeout, a production chip or commercial success, and the announcement does not establish revenue per design. The milestone's value lies in demonstrating accumulated adoption and the breadth of architectures now built on Arteris technology.

For engineering teams, the practical consequence is a reframed question: not which processing blocks to select, but how those blocks will share information. For system companies developing custom silicon, integration expertise gains weight in balancing product differentiation, development schedules and execution risk.

The broader read for executives: semiconductor differentiation increasingly depends on how the whole system works together. More cores and accelerators create opportunity — and more communication and integration challenges. Arteris' customer history shows those challenges moving to the center of chip development. The next test is whether its technology turns architectural ambition into efficient, dependable silicon.