Kinara
Kinara is a San Jose-based fabless semiconductor company that designs discrete RISC-V neural processing units (Ara-1 and Ara-2) for on-device edge AI inference, serving OEMs in industrial, automotive, IoT and computer vision markets. Acquired by NXP Semiconductors for $307M in October 2025.
- Company typePrivate
- Founded2013
- HeadquartersSanta Clara, United States
- Headcount51–100
- GTM typeB2B
- OfferingHardware or Manufacturing
What Kinara does
Kinara is a US fabless semiconductor company founded in 2013 and headquartered in San Jose, California, that designs discrete neural processing units (DNPUs) for edge AI inference. The company sells two product generations: the Ara-1, delivering 6 eTOPS, and the higher-end Ara-2 (Ara240), delivering 40 eTOPS, both built on a proprietary programmable RISC-V dataflow architecture with integrated MAC units rather than a repurposed DSP or GPU. The Ara-2 generation explicitly targets modern AI workloads including large language models up to 7 billion parameters at approximately 12 tokens per second on-device, as well as vision-language models, vision-language-action models, agentic AI and generative AI applications.
Kinara's business model is a fabless semiconductor model: it designs silicon IP, outsources fabrication, and sells discrete DNPUs plus a proprietary SDK to OEMs and system designers in edge-AI markets such as industrial automation, automotive, IoT and computer vision. Customers integrate Ara DNPUs alongside host processors (including NXP i.MX), and qualify the hardware through standard design-in cycles. The company was backed by Exfinity Venture Partners and Tiger Global Management across funding rounds of $2.1M (2016), $12.4M (2018) and a Series B of approximately $34.7M in 2021, led by Ravi Annavajhhala as CEO with co-founders Rehan Hameed (CTO) and Wajahat Qadeer (Chief Architect).
In October 2025, NXP Semiconductors acquired Kinara for $307 million, absorbing the company's edge-AI NPU roadmap and engineering team into NXP's broader automotive and industrial semiconductor platform. Pre-acquisition, Kinara operated with a lean organization of 51-100 employees and supplemented its direct OEM sales motion with ecosystem partnerships, including a SiFive collaboration on RISC-V vector processors (May 2025) and membership in the AMD Xilinx Edge AI Alliance (April 2025).
Kinara firmographics
Firmographics- Name
- Kinara
- Legal name
- Kinara, Inc.
- Website
- https://kinara.ai
- Company type
- Private
- Founded year
- 2013
- Operating status
- Acquired
- Headcount range
- 51–100 employees
- Short description
- Kinara is a San Jose-based fabless semiconductor company that designs discrete RISC-V neural processing units (Ara-1 and Ara-2) for on-device edge AI inference, serving OEMs in industrial, automotive, IoT and computer vision markets. Acquired by NXP Semiconductors for $307M in October 2025.
- Ownership category
- akta.pro rank
Kinara industry classification
Industry- Product category
- Edge AI Semiconductor Chips
- NAICS
- Computer and Peripheral Equipment Manufacturing (33411)
- SIC
- Electronic Computers (3571)
- akta.pro primary industry
- Edge AI Hardware & NPUs (SoCs, modules, accelerators) (HDAAAJAC)
- akta.pro secondary industries
- AI Compiler, Runtime & Kernel Optimization Software (CUDA/ROCm/XLA, graph compilers) (HDAAAAAI), Compute & Acceleration ICs (CPU/GPU/AI/FPGA/ASIC) (HDAHABAC), On-Device Inference Runtimes & SDKs (mobile/embedded) (HDAAAJAB)
Keywords
Where Kinara is headquartered
LocationHeadquarters
- HQ city
- Santa Clara
- HQ country
- United States
- HQ region
- North America
Offices1 record
Markets served
Kinara business model
Business model- GTM type
- B2B
- Offering type
- Hardware or Manufacturing
- Cost components
- Technology or R&D, Personnel, Operations, Supply Chain, Marketing or Sales
Revenue model
- Discrete NPU Hardware Sales: Sale of AI accelerator chips (Ara-1 and Ara240 discrete NPUs) to OEM customers for integration into edge AI systems in industrial, automotive, and IoT applications.
Go-to-market motion1 record
Distribution channels2 records
Marketing channels3 records
Kinara product offering
Product offeringCore offering
Kinara designs, develops, and sells discrete Neural Processing Units (DNPUs) — specifically the Ara-1 and Ara240 chips — built on a programmable RISC-V-based dataflow architecture for edge AI inference workloads. The chips deliver up to 40 eTOPS of performance and target vision-centric edge use cases, generative AI, LLMs, VLMs, VLAs, and agentic AI in industrial, automotive, and IoT applications. The company also provides a comprehensive AI software development kit with model optimization tools and a preoptimized model library.
Product overview
Kinara was an edge AI chip company specializing in high-performance, power-efficient discrete neural processing units (DNPUs) for edge AI workloads. The company offered two generations of NPUs: Ara-1 (first-generation, up to 6 eTOPS for vision-centric applications) and Ara240/Ara-2 (second-generation, up to 40 eTOPS optimized for generative AI, LLMs, VLMs, VLAs, and agentic AI). Kinara also provided a comprehensive AI software development kit including model optimization tools and preoptimized AI models. In October 2025, NXP Semiconductors acquired Kinara for $307 million to integrate these discrete NPUs and AI software into NXP's portfolio of processors, connectivity, security, and analog solutions for industrial and automotive edge markets, enabling complete and scalable AI platforms from TinyML to generative AI.
Differentiator
Problem solved
Functional benefit
Brands
- Ara-1: First-generation discrete neural processing unit delivering up to 6 eTOPS for vision-centric edge AI use cases.
- Ara240
- Ara-2
Products and services
- Ara-1 Discrete Neural Processing Unit Ara-1 is Kinara's first-generation discrete Neural Processing Unit (NPU) delivering up to 6 eTOPS (equivalent TOPS) of AI performance for vision-centric edge AI use cases. It features a programmable RISC-V-based dataflow architecture with proprietary MAC units, supporting CNN-based computer vision workloads. The chip is designed for OEM customers integrating edge AI inference into industrial, automotive, and IoT applications.
- Ara240 (Ara-2) Discrete Neural Processing Unit Ara240 (also known as Ara-2) is Kinara's second-generation discrete Neural Processing Unit delivering up to 40 eTOPS of AI performance. It is optimized for generative AI, large language models (LLMs), vision language models (VLMs), vision language action models (VLAs), agentic AI, and system-level acceleration. The chip supports both CNNs and transformer-based architectures, enabling real-time generative AI and LLM execution on AI-enabled compute and embedded systems with low latency, lower operational costs, and enhanced data privacy.
- Kinara AI Software Development Kit (SDK) The Kinara AI Software Development Kit is a comprehensive software stack providing SDK tools, model optimization utilities, and an extensive library of preoptimized AI models for building, optimizing, and deploying AI applications on Ara-1 and Ara240 NPUs. The SDK has been integrated into NXP's eIQ software development environment, enabling developers to deploy AI applications across the combined NXP and Kinara portfolio.
Quantifiable outcome
- Up to 40 eTOPS AI performance for edge inference workloads
- +1 more outcomes
Companies that use Kinara
Customer profileSegments3 records
Ideal customer profiles3 records
Kinara technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- No
- API docs
- API detail
Core technology
AI maturity
App detail
AI capability10 records
Feature5 records
Kinara partnerships and signals
Strategic signalPartnerships
Two partnerships are on record, tiered core and flagship.
- SiFivecorePartnership to offer bare metal access to RISC-V vector processors. Combined SiFive RISC-V cores with Kinara's Ara-2 NPU (40 TOPS) for edge AI applications. Demonstrated on USB form factor enabling rapid prototyping and development of AI-enabled systems.
- NXP SemiconductorsflagshipKinara was a long-standing member of NXP Partner Program prior to acquisition. Discrete NPUs (Ara-1 and Ara240) integrated with NXP i.MX application processors (i.MX 8M Plus, i.MX 95) for edge AI use cases. Acquired by NXP for $307 million in October 2025.
Scale indicators1 record
Recent moves5 records
Expansion highlights5 records
Kinara competitors and assessment
Company assessmentBroad incumbents
- Ambarella: Ambarella designs vision AI SoCs for edge cameras, automotive, and IoT, with integrated CVflow accelerators. It competes with Kinara indirectly by embedding AI compute into application processors for the same vision-centric end markets.
- NVIDIA (Jetson): NVIDIA's Jetson platform delivers edge AI compute (Nano, Orin, Thor) for robotics, industrial, and automotive markets. It is a broad incumbent that competes with Kinara Ara NPUs on high-performance edge AI, with a much larger software ecosystem.
- Qualcomm: Qualcomm integrates Hexagon NPUs into Snapdragon and Snapdragon Ride platforms for mobile, IoT, and automotive edge AI. It competes with Kinara by offering integrated NPU-in-SoC rather than discrete accelerators, pressuring the standalone NPU form factor.
- Intel (Movidius / OpenVINO): Intel's Movidius VPU line and OpenVINO toolkit target edge AI inference for vision and industrial workloads. It overlaps with Kinara in industrial and IoT edge deployments, with deeper distribution but older discrete-accelerator silicon.
- MediaTek: MediaTek integrates APUs (NPUs) into its Dimensity, Kompanio, and Genio SoCs for mobile, Chromebook, and IoT edge AI. It competes with Kinara by embedding NPU compute into application processors, eroding the case for discrete accelerators.
Direct peers
- Hailo: Hailo builds discrete edge AI accelerators (Hailo-8, Hailo-15) for vision and GenAI inference across automotive and industrial. It is the most direct competitor to Kinara's Ara discrete NPU line, with overlapping customers, similar TOPS-per-watt positioning, and a comparable software stack.
- Syntiant: Syntiant develops ultra-low-power neural decision processors for edge audio/vision/IoT. It overlaps with Kinara on edge AI inference silicon for always-on and battery-powered devices, though focused on lower TOPS tiers than Ara240.
- BrainChip: BrainChip offers neuromorphic Akida processors for low-power edge AI. Like Kinara, it targets on-device inference with an event-driven architecture, serving industrial, automotive, and IoT use cases, but uses spiking neural networks rather than CNN/transformer dataflow.
- Mythic: Mythic builds analog in-memory compute AI accelerators for edge inference. It directly competes with Kinara's Ara NPUs on power-efficient edge AI silicon, though pursuing a fundamentally different compute-in-memory architecture.
Emerging players
- Sima.ai: Sima.ai develops MLSoC edge AI chips for vision applications. It overlaps with Kinara's Ara-1 in vision-centric industrial and IoT endpoints, targeting similar OEMs but with a more application-tuned SoC rather than a discrete accelerator.
Market position
Weaknesses4 records
Competitive moat5 records
Key risks5 records
Key highlights7 records
Customer concentration
Kinara social profiles
Digital presenceKinara financial estimates
Financial estimateRevenue estimate
Valuation estimate
Kinara leadership team
Management profileNumber of profiles
Profiles3 records
Kinara funding detail
Funding detailFunding overview
Funding rounds3 records
Investors4 records
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Kinara M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Kinara
What does Kinara do?
Kinara designs, develops, and sells discrete Neural Processing Units (DNPUs) — specifically the Ara-1 and Ara240 chips — built on a programmable RISC-V-based dataflow architecture for edge AI inference workloads. The chips deliver up to 40 eTOPS of performance and target vision-centric edge use cases, generative AI, LLMs, VLMs, VLAs, and agentic AI in industrial, automotive, and IoT applications. The company also provides a comprehensive AI software development kit with model optimization tools and a preoptimized model library.
Is Kinara a public or private company?
Kinara is a private company. It is classified as corporate owned and is currently acquired.
When was Kinara founded?
Kinara was founded in 2013. It employs 51 to 100 people.
Where is Kinara based?
Kinara is headquartered in Santa Clara, United States, in the North America region.
How does Kinara make money?
One revenue line is on record: discrete NPU Hardware Sales.
Who are Kinara's main competitors?
Broad incumbents on record are Ambarella, NVIDIA (Jetson), Qualcomm, Intel (Movidius / OpenVINO) and MediaTek. Direct peers are Hailo, Syntiant, BrainChip and Mythic. Sima.ai is listed as an emerging player.
Does Kinara have an API?
No public API is recorded for Kinara.
What industry is Kinara in?
Kinara's product category is Edge AI Semiconductor Chips. Its primary akta.pro industry code is HDAAAJAC, Edge AI Hardware & NPUs (SoCs, modules, accelerators), with a secondary code of HDAAAAAI, AI Compiler, Runtime & Kernel Optimization Software (CUDA/ROCm/XLA, graph compilers). Its NAICS code is 33411 and its SIC code is 3571.