The Budapest‑based AI chipmaker aiMotive announced today that its next‑generation automotive neural‑processing unit, aiWare5, is now available for licensing, marking the first ISO 26262 ASIL‑B certified NPU IP on the market.
aiWare5 is a purpose‑built automotive inference accelerator that pushes the limits of on‑vehicle AI while meeting the safety standards required for production‑grade systems. The new IP follows a soft launch at CES and a partnership with Socionext to broaden its ecosystem for Tier 1 and OEM customers. It is positioned as a drop‑in upgrade to the company’s previous aiWare4, promising higher throughput, dynamic FP8 scaling, and a suite of development tools that aim to shrink time‑to‑silicon for autonomous‑driving workloads.
What aiWare5 Brings to Automotive AI
aiWare5 is not a generic data‑center accelerator repurposed for cars. It was engineered from the ground up for the unique constraints of automotive compute—low power, deterministic latency, and rigorous safety compliance. The IP delivers up to 2.5 TOPS per watt, a figure that Gartner estimates will become the benchmark for edge AI chips in vehicles by 2027. Its architecture supports emerging model families such as large language models (LLMs), vision transformers, and state‑space models (SSMs), which are increasingly being explored for in‑car perception and decision‑making.
The platform also includes a GPU‑optimized emulator that can simulate silicon performance faster than real time, allowing developers to validate large‑scale software‑in‑the‑loop (SiL) scenarios before any hardware is fabricated. Combined with aiWare Studio, an offline optimizer that predicts on‑chip performance within 5 % of silicon, the toolchain promises to cut development cycles by an estimated 30 % according to internal benchmarks.
Technical Highlights
- Dynamic FP8 scaling – Enables higher precision when needed while preserving the ultra‑low power envelope of FP8 for the bulk of inference.
- Safety‑first design – First NPU IP to achieve ISO 26262 ASIL‑B certification, delivered as a Safety Element out of Context (SEooC) with full compliance documentation.
- Future‑proof workload support – Native kernels for LLMs, vision transformers, and SSMs, preparing automotive platforms for the next wave of generative AI services inside the vehicle.
- Emulation‑first development – The GPU‑accelerated emulator provides bit‑accurate modeling, reducing reliance on physical prototypes and accelerating silicon validation.
Safety and Certification Edge
Safety certification has long been a bottleneck for automotive AI silicon. By securing ASIL‑B status, aiWare5 gives OEMs a pre‑validated safety foundation, shaving months off the functional safety verification process. The SEooC approach means the IP can be integrated into higher‑grade ASIL‑C or ASIL‑D systems with minimal additional effort, a claim corroborated by a recent Forrester study that found safety‑certified IP can reduce overall system validation time by up to 25 %.
Implications for Enterprise AI and Marketing Teams
While aiWare5 is a hardware IP, its ripple effects extend to enterprise AI teams tasked with deploying in‑car services. The deterministic performance and safety envelope simplify the business case for integrating advanced analytics, personalized infotainment, and even on‑device generative assistants. marketing teams can now promise truly on‑board AI experiences without relying on constant cloud connectivity, a differentiator that aligns with the growing consumer demand for privacy‑first solutions.
Moreover, the licensing model allows semiconductor partners to embed aiWare5 into a variety of form factors—from dedicated AI coprocessors to heterogeneous chiplets—opening new revenue streams for OEMs and Tier 1 suppliers. The reduced development timeline also means faster time‑to‑market for next‑generation ADAS and autonomous driving features, a competitive advantage in an industry where IDC predicts AI chip spend will grow 30 % YoY through 2028.
Competitive Landscape
aiWare5 enters a crowded field that includes Nvidia’s Drive AGX, Qualcomm’s Snapdragon Ride, and Intel’s Mobileye EyeQ. Unlike its rivals, which often rely on general‑purpose GPU cores or adapted data‑center designs, aiWare5’s automotive‑first architecture and ASIL‑B certification give it a distinct safety and efficiency edge. Analysts at McKinsey note that safety‑certified AI chips could capture up to 15 % of the automotive AI market by 2030, suggesting aiMotive’s niche positioning may translate into tangible market share if adoption accelerates.
Market Landscape
The automotive AI market is on the cusp of a paradigm shift. Gartner projects that AI‑enabled vehicle functions will generate $120 billion in revenue by 2027, driven by autonomous driving, predictive maintenance, and personalized in‑car experiences. However, the industry faces a talent and tool shortage; developers need hardware that can be modeled accurately without extensive silicon access. aiWare5’s emulator‑first workflow directly addresses this gap, aligning with the broader trend of “software‑defined silicon” that IDC highlights as a key growth driver.
Safety standards remain a decisive factor. With ISO 26262 ASIL‑B certification now a reality for an NPU IP, aiMotive sets a new baseline that competitors will need to match to stay relevant in the safety‑critical segment. This move could also accelerate the adoption of generative AI inside vehicles, a capability that analysts at Forrester predict will become a mainstream feature in premium models by 2029.
Top Insights
- Safety‑first IP – aiWare5’s ISO 26262 ASIL‑B certification gives OEMs a pre‑validated safety layer, cutting functional safety verification time by up to 25 %.
- Emulation‑driven development – The GPU‑optimized emulator enables bit‑accurate SiL testing faster than real time, potentially reducing development cycles by 30 %.
- Future‑ready workloads – Native support for LLMs, vision transformers, and SSMs positions aiWare5 as a platform for the next generation of in‑car generative AI services.
- Competitive differentiation – Unlike generic GPUs, aiWare5’s automotive‑first design offers higher efficiency (≈2.5 TOPS/W) and safety compliance, a combination few rivals can match.
- Enterprise impact – Faster time‑to‑market and on‑device AI capabilities empower marketing departments to promise privacy‑centric, personalized experiences without heavy cloud reliance.
