Blaize

What Blaize does

Blaize is a programmable AI inference platform company that designs purpose-built silicon and software for edge and hybrid edge-to-core AI deployments. Founded in 2011 and headquartered in El Dorado Hills, California, Blaize went public on NASDAQ (ticker BZAI) in January 2025 via SPAC merger with BurTech Acquisition Corp. at an approximately $1.2 billion valuation. The company employs roughly 300 people globally, with about half based at its Hyderabad R&D center in India, and additional offices in the US (San Jose), UK (Leeds, Kings Langley), UAE (Abu Dhabi), and South Korea (Chungbuk).

The company's core technology is the Graph Streaming Processor (GSP), a proprietary, programmable AI inference architecture that executes neural networks as directed graphs using an on-chip hardware scheduler, delivering task-level parallelism with claimed 16 TOPS at 7W on current Samsung 14nm silicon and 50x less memory bandwidth versus GPU/CPU alternatives. Blaize sells hardware products (Pathfinder embedded SoMs at $299-$399 in volume, Xplorer accelerator cards at $999) and software platforms (Picasso SDK, AI Studio), but is increasingly shifting toward the Blaize AI Services Platform — application-level AI delivered as modular APIs with usage-based and outcome-based pricing, supported by Forward Deployed Engineering under a revenue-share model. The Hybrid AI architecture combines GSP silicon with GPU infrastructure under an intelligent workload scheduler that routes queries to the optimal processor, delivering 2.4x more video streams per rack and ~60% less power per rack than GPU-only approaches.

Blaize targets four primary segments: Automotive & Mobility (OEMs and Tier 1 suppliers for ADAS, in-cabin sensing, and sensor fusion), Smart Vision (industrial, retail, smart city computer vision), Enterprise Computing edge-to-core, and Infrastructure Providers (CSPs, telcos, system integrators). The company serves customers including NeoTensr (up to $70M contracted for APAC edge data centers), Winmate (~$15M first-year rugged computing target), TCC (Saudi Public Investment Fund subsidiary for highway monitoring), the Government of Telangana (AI Innovation Hub MoU), and SPHERE AX (Korea Physical AI partnership). Samsung is both foundry partner (14nm current, 5nm planned for GSP2) and investor via Samsung Catalyst Fund. FY2025 revenue was $38.6M with a $206.9M net loss; management has guided to $130M for 2026, though Q1 2026 revenue came in at $2.7M amid global memory shortages, the 2025 10-K filed going-concern doubt, and 35 law firms are investigating potential securities fraud following short-seller reports on the NeoTensr agreement.

Name

Blaize

Legal name

Blaize Holdings, Inc.

Website

https://blaize.com

Company type

Public

Founded year

2011

Operating status

Operating

Headcount range

101–250 employees

Short description

Blaize designs programmable AI inference silicon (Graph Streaming Processor) and software platforms for edge and hybrid edge-to-core AI deployments, serving automotive, smart vision, enterprise, and infrastructure-provider customers globally. Listed on NASDAQ as BZAI.

Ownership category

akta.pro rank

Product category

Edge AI Inference Hardware and Software Platforms

NAICS

Computer and Peripheral Equipment Manufacturing (33411), Software Publishers (5132), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518210), Computer Systems Design Services (541512)

SIC

Services-Computer Programming, Data Processing, Etc. (7370)

akta.pro industry

PRIMARYHDAAAAAC

AI Inference Hardware & Edge Accelerators

Hardware, Technology & Telecom/Artificial Intelligence & Machine Learning/AI Infrastructure & Compute (GPUs/ASICs, accelerators)

SECONDARYHDAAAAAA

AI Accelerators (GPUs/TPUs/NPUs/ASICs)

Hardware, Technology & Telecom/Artificial Intelligence & Machine Learning/AI Infrastructure & Compute (GPUs/ASICs, accelerators)

SECONDARYHDAAAAAB

AI Server Systems & HGX/Accelerator Platforms

Hardware, Technology & Telecom/Artificial Intelligence & Machine Learning/AI Infrastructure & Compute (GPUs/ASICs, accelerators)

SECONDARYHDAAAJAB

On-Device Inference Runtimes & SDKs (mobile/embedded)

Hardware, Technology & Telecom/Artificial Intelligence & Machine Learning/Edge AI & On-Device Machine Learning

SECONDARYHDAEANAC

Model Serving, Inference & Deployment Platforms (APIs, Edge/On-Prem)

Hardware, Technology & Telecom/Enterprise Platforms & Databases/Artificial Intelligence Platforms for Enterprise

SECONDARYHDAEANAJ

AI Application Enablement Platforms (Copilot/Agent Frameworks, SDKs)

Hardware, Technology & Telecom/Enterprise Platforms & Databases/Artificial Intelligence Platforms for Enterprise

SECONDARYHDAAAAAI

AI Compiler, Runtime & Kernel Optimization Software (CUDA/ROCm/XLA, graph compilers)

Hardware, Technology & Telecom/Artificial Intelligence & Machine Learning/AI Infrastructure & Compute (GPUs/ASICs, accelerators)

Keywords

• Edge AI inference
• AI accelerator hardware
• Programmable AI processors
• Edge computing platforms
• AI inference silicon

Headquarters

HQ city

San Jose

HQ country

United States

HQ region

North America

Offices9 records

Markets served

GTM type

B2B

Offering type

Hardware or Manufacturing

Cost components

Technology or R&D, Personnel, Supply Chain, Operations, Marketing or Sales, Infrastructure

Revenue model

1. AI Hardware Sales (Edge Computing Platforms): Sale of AI inference accelerator products including the Blaize Pathfinder and Xplorer product lines (SoMs, accelerator cards, embedded kits). Hardware is sold directly to enterprises, embedded into partner platforms (Winmate rugged systems), or through channel partners. Products range from embedded SoMs ($299–$999 in volume) to full accelerator systems.
2. Blaize AI Services (Application-Level APIs): Recurring revenue from production-ready, application-level AI services delivered as modular APIs. Supports flexible commercial models including usage-based pricing and outcome-based services, creating recurring AI service revenue beyond traditional hardware sales. Forward Deployed Engineering is delivered through a revenue-share model where Blaize earns when customers earn. Targets cloud providers, data center operators, and system integrators seeking to monetize AI infrastructure.
3. Software Licenses and SDK Subscriptions: Licensing of the Blaize Picasso SDK and AI Studio software platform for edge AI application development, optimization, and deployment. May include support subscriptions and professional services for platform onboarding.
4. Professional Services and Forward Deployed Engineering: Forward Deployed Engineering places Blaize engineers directly in customer and partner environments for application integration, workflow configuration, and ongoing optimization. Delivered as a revenue-share model for infrastructure partners, and as professional services engagements for enterprise deployments.
5. NeoTensr APAC Infrastructure Contract: Co-branded AI edge data center infrastructure contracts across the Asia-Pacific region. Building on a $20M+ order recognized in Q4 2025, the expanded contract with NeoTensr is expected to generate up to $50M in revenue within the first year, bringing total potential contracted value to $70M. Covers hybrid AI systems across edge and enterprise data center environments in the Asia Pacific region, with fulfillment beginning Q2 2026.

Pricing tiers

Go-to-market motion3 records

Distribution channels6 records

Marketing channels6 records

Core offering

Blaize designs, manufactures, and sells programmable AI inference processors based on its proprietary Graph Streaming Processor (GSP) architecture, edge AI computing hardware platforms (Pathfinder embedded SoMs and Xplorer accelerators in EDSFF, M.2, and PCIe form factors), and AI software platforms (Picasso SDK, AI Studio code-free platform, AI Services Platform). Products are sold to enterprise and government customers in automotive, smart vision, enterprise computing, defense, and infrastructure provider markets for energy-efficient AI inference at the edge and across hybrid environments.

Product overview

Blaize offers a programmable AI platform purpose-built for inference in real-world environments, comprising the Graph Streaming Processor (GSP) architecture as the foundational core technology and two main product lines: AI edge computing hardware (Pathfinder and Xplorer platforms) and AI software (AI Studio and Picasso SDK). The platform is complemented by the Blaize AI Services platform which delivers production-ready, application-level AI services as modular APIs. The GSP architecture enables energy-efficient AI inference with up to 60x greater systems efficiency versus GPU/CPU alternatives, supporting computer vision, multimodal AI, and sensor-driven applications across edge, cloud, and data center deployments. The Hybrid AI architecture combines the proprietary GSP with GPU-based infrastructure for flexible workload distribution.

Differentiator

Problem solved

Functional benefit

Brands

• Blaize AI Services Platform: Production-ready, application-level AI services platform that enables cloud providers and enterprises to deploy AI applications without building from scratch
• Blaize AI Studio
• Blaize Pathfinder
• Blaize Xplorer
• Graph Streaming Processor (GSP)
• Picasso SDK

Products and services

• Blaize AI Services Platform Production-ready, application-level AI services delivered as modular APIs. Packages multimodal inference, business logic, management, and orchestration into composable services for cloud providers, system integrators, and enterprises to deploy AI without building from scratch. Includes Forward Deployed Engineering support and supports vision, video, document processing, speech, and multimodal workflows.
• Blaize AI Studio The first open and code-free AI software platform spanning the complete edge AI operational workflow from idea to deployed app. Includes DataOps, DevOps (low-code/no-code drag-and-drop visual interface), and MLOps capabilities with AI-assisted data preparation and automated model deployment, plus a Marketplace for sharing AI/ML artifacts and trained models.
• Blaize Picasso SDK True graph-native and open software development environment that accelerates the entire AI application development cycle. NetDeploy automatically converts and optimizes AI frameworks to Blaize hardware, improving accuracy and performance while shortening manual deployment time from months to weeks or days.
• Blaize Pathfinder P1600 Embedded System on Module Embedded System on Module with dual ARM Cortex A53 processors, MIPI CSI/DSI interfaces, H.264/H.265 encoder/decoder, and 16 GSP cores delivering 16 TOPS AI inference performance at 7W power. Winner of the 2022 Edge AI and Vision Product of the Year award for Best Edge AI Processor.
• Blaize Xplorer X1600E EDSFF Accelerator Enterprise-grade small form factor accelerator with 1 Blaize 1600 SoC (16 GSP cores, 16 TOPS), soft ISP, 4 GB LPDDR4, PCIe Gen 3.0. Designed for power-constrained environments and available at $299 per unit in volume quantity.
• Blaize Xplorer X600M M.2 Accelerator Commercial grade M.2 small form factor accelerator with Blaize 1600 SoC (16 GSP cores), H.265/H.264 video encode/decode up to 4K@30FPS, and 2 GB LPDDR4. Designed for commercial-grade edge AI deployments.
• Blaize Xplorer X1600P PCIe Accelerator Standard PCIe-based accelerator in half-height, half-width form factor providing 16-64 TOPS of AI inference performance within a low power envelope as a replacement for power-hungry desktop GPUs. Available at $999 per unit in volume quantity.
• Blaize Xplorer X1600P-Q Quad PCIe Accelerator Quad PCIe Accelerator with four Blaize 1600 SoCs (64-80 TOPS total), 16 GB LPDDR4, and PCIe Gen 3.0 16 lanes. High-performance configuration for demanding edge server applications, forming the basis of co-branded AI edge data center infrastructure deployments.

Quantifiable outcome

• 60x greater system-level efficiency vs. GPU/CPU solutions
• +8 more outcomes

Named customers6 records

Segments5 records

Ideal customer profiles4 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

AI capability9 records

Feature5 records

Partnerships

Nine partnerships are on record, tiered core and minor.

• SPHERE AXcoreStrategic or Co-development Partner · 24 June 2026SPHERE AX, a Korean Vision AI and Edge AI Computing company, signed a strategic MOU with Blaize at the National Assembly in Seoul to pursue joint development and commercialization of AI semiconductor-based products for the global Physical AI market. The partnership combines Blaize's AI semiconductor technology optimized for physical AI with SPHERE AX's vision AI software expertise, targeting applications across smart cities, smart factories, industrial safety, security, robotics, and mobility. The two companies intend to establish a Korea-based AI product development and production system linking overseas semiconductor technology with domestic productization for global supply. The CEOs of both companies indicated the collaboration represents the first step toward developing and producing commercial AI products in Korea.
• Winmate Inc.coreStrategic or Co-development Partner · 4 May 2026Blaize and Winmate signed a Strategic Partnership Agreement to integrate Blaize's energy-efficient, industrial-grade AI chips into Winmate's rugged computing systems including drones, handhelds, vehicle-mounted units, and embedded edge devices for defense, border security, maritime, and healthcare operators. The partnership targets approximately $15 million in first-year business with a three-year initial term. Winmate selected Blaize following technical evaluation for its power efficiency, thermal performance, and compact form factor. The companies showcased integrated solutions at COMPUTEX 2026 in Taipei following a US debut at the Michigan Defense Expo.
• PT Datacomm DiangrahacoreChannel Partner/ Reseller/ Distributor · 21 April 2026Blaize and PT Datacomm Diangraha, an Indonesian IT service provider with 30 years of regional expertise, signed a non-binding MOU at Gitex Asia 2026 in Singapore to establish a technology alliance exploring AI inference solutions across Indonesia. The alliance focuses on AI Inference as a Service on Datacomm's DCloud platform, physical AI, public safety, surveillance, industrial AI, and logistics applications. Indonesia's AI sector is expanding at 31% CAGR — the fastest in Southeast Asia. The MoU was subsequently expanded into a three-way partnership with Nokia to deploy hybrid AI inference infrastructure across Indonesia and the broader APAC region.
• NeoTensrcoreStrategic or Co-development Partner · 16 April 2026Blaize and NeoTensr entered into a contract worth up to $50 million to jointly develop and deploy co-branded AI edge data center infrastructure across the Asia Pacific region. The partnership builds on a $20 million+ order recognized in Q4 2025, bringing total potential contracted value to $70 million. The co-branded AI servers, based on Blaize's quad card configuration, can handle 200+ simultaneous camera streams with advanced AI analytics including LLM and VLM inference for smart city, industrial automation, logistics, retail intelligence, and security applications. Note: This partnership was subsequently the subject of short-seller allegations of fraud in April 2026.
• GSiL (Korea's Leading Smart Safety Company)minorStrategic or Co-development Partner · 5 March 2026GSiL, a Korean smart safety company, signed a non-binding MOU with Blaize to co-develop next-generation 'Physical AI' safety technology combining Vision AI, Generative AI, and real-time data to detect, analyze, and predict physical safety risks in industrial and public environments. The companies plan to launch an initial pilot program at Samsung affiliates' industrial sites in Korea before expanding into global EPC markets in the Middle East, Southeast Asia, and Europe.
• Government of Telangana (India)coreStrategic or Co-development Partner · 28 January 2026The Government of Telangana signed a non-binding MoU with Blaize at the World Economic Forum in Davos to launch the Telangana AI Innovation Hub (TAIH). The agreement outlines Blaize's intention to establish a Research and Development Centre in Telangana to advance AI computing for applied AI initiatives and government-led projects. The collaboration aims to facilitate applied research, pilot programs, and deployment-oriented AI initiatives across critical sectors including public sector governance and infrastructure modernisation. India represents a major growth market for deploying AI in sectors such as agriculture, public safety, and traffic management.
• Nokia Solutions and Networks SingaporecoreStrategic or Co-development Partner · 27 January 2026Nokia and Blaize signed a strategic MOU to jointly explore, develop, and deploy edge and hybrid AI inference solutions across the Asia Pacific region. The collaboration combines Nokia's networking, automation, and cloud infrastructure capabilities with Blaize's programmable AI inference platform to enable Practical AI and Physical AI systems in telecom, industrial, and smart infrastructure environments. The non-binding MOU establishes a framework for joint use case development, reference architectures, validation of production-ready deployments, and go-to-market initiatives including customer workshops and pilot programs. The partnership was validated at Nokia's Network Innovation Lab in Singapore. The collaboration was later expanded into a three-way partnership with PT Datacomm Diangraha to deploy hybrid AI infrastructure across Indonesia and the broader APAC region.
• YottaminorChannel Partner/ Reseller/ DistributorBlaize partners with Yotta for data center services in India, supporting the company's expansion of AI infrastructure deployment in the Indian market as part of its broader Asia-Pacific strategy.
• Samsung (Foundry Partner)coreTechnology or IntegrationSamsung is Blaize's industry-leading foundry partner providing proven semiconductor manufacturing solutions. Blaize's GSP architecture is currently manufactured on Samsung 14nm process technology, with the next-generation GSP2 targeting Samsung 5nm for automotive ADAS applications. Samsung Catalyst Fund is also an investor in Blaize.

Scale indicators10 records

Recent moves7 records

Expansion highlights8 records

Broad incumbents

• NVIDIA: NVIDIA is the dominant AI accelerator company with the Jetson edge inference family and Triton/TensorRT serving platforms. While much larger and broader, it competes directly with Blaize in automotive (Drive), edge AI (Jetson), and data-center inference — the same three primary segments Blaize targets.
• Qualcomm: Qualcomm offers the Cloud AI 100 inference accelerator and Snapdragon Ride platforms for automotive ADAS, both of which overlap directly with Blaize's data-center inference and automotive segments. As a broad incumbent with deep automotive Tier-1 relationships, it is a key competitor for design wins.
• Intel (Movidius / OpenVINO / Habana): Intel's Movidius VPU family, OpenVINO toolkit, and Habana Labs Gaudi inference accelerators compete with Blaize in both edge AI inference and data-center inference. Intel is a broad incumbent with established enterprise and developer ecosystems.

Direct peers

• SiMa.ai: SiMa.ai builds machine-learning SoCs for embedded edge inference with an emphasis on multimodal AI at low power. It is a direct GSP competitor in robotics, industrial, and automotive edge inference.
• Ambarella: Ambarella produces AI vision processors for edge applications including automotive ADAS, security/surveillance cameras, and industrial robotics. Its CV series SoCs compete with Blaize Xplorer and Pathfinder in smart vision and edge inference.
• Hailo: Hailo develops purpose-built AI inference processors for edge devices (Hailo-8, Hailo-15), targeting automotive, smart cities, retail, and industrial vision. It directly mirrors Blaize's GSP-based value proposition with a similar TOPS/Watt pitch and overlapping customer segments.

Emerging players

• Mythic AI: Mythic develops analog compute-in-memory AI inference processors for edge devices, targeting power-constrained vision and sensor applications. It overlaps with Blaize's value proposition of high TOPS at low watts but remains pre-scale.
• Tenstorrent: Tenstorrent develops programmable AI accelerators based on RISC-V and custom silicon for both training and inference, targeting data-center and edge workloads. It is comparable to Blaize in competing for CSP and enterprise inference sockets with a novel architecture.
• Syntiant: Syntiant makes ultra-low-power neural decision processors for always-on edge AI in audio, vision, and sensor applications. It is comparable to Blaize on the edge-inference-for-constrained-environments axis, though at much lower compute tiers.
• Groq: Groq builds LPU-based inference accelerators for data-center and edge inference, targeting CSP and enterprise customers with a similar low-latency, programmable pitch to Blaize's GSP. Different architecture but directly comparable inference economics narrative.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights7 records

Customer concentration

• LinkedInlinkedin.com/company/blaize-ai
• Xx.com/blaizeinc

Compliance1 record

Revenue estimate

Valuation estimate

Number of profiles

Profiles12 records

Subsidiaries3 records

Funding overview

Funding rounds7 records

Investors18 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Blaize