Blaize Holdings, Inc. · NASDAQ:BZAI
Blaize — Q4 2024 Earnings Presentation
28 slides from Blaize's earnings presentation, Mar 27, 2025
Slide 1: Life on the Edge – Investor Presentation Fourth Quarter 2024
Life on the Edge Investor Presentation Fourth Quarter 2024 Copyright © 2024 Blaize. All rights reserved.
Slide 2: Disclaimer
Disclaimer Cautionary Statement Regarding Forward-Looking Statements This presentation (the “Presentation”) contains forward-looking statements within the meaning of Section 27A of the U.S. Securities Act of 1933, as amended (the “Securities Act”), and Section 21E of the U.S. Securities Exchange Act
Slide 3: The Right Platform For The Future Of Edge AI Computing
The Right Platform For The Future Of Edge AI Computing Investment Highlights Full-Stack Hardware and Software Solution Built for AI Inference at the Edge Intuitive Flexibility for AI and Related Workflows Targeting a $71B Global Serviceable Addressable Market by 20281, across Security & Monitoring,
Slide 4: Strong Momentum and Technological Focus to Drive Accelerated AI Adoption
Strong Momentum and Technological Focus to Drive Accelerated AI Adoption Global presence, advanced IP, and market traction to propel growth Select Investors, Customers and Key Metrics and Market Stats Key Relationships 220 Investors ~ US-based with offices in the UK, India & UAE 56 Patents & Applica
Slide 5: Blaize – Life on the Edge
Blaize – Life on the Edge Edge computing brings data processing closer to the devices that generate and use it, reducing latency and improving real-time decision-making Blaize® has created a transformative new compute solution that unites silicon and software to optimize Al from the Edge to the core
Slide 6: Edge Computing is the Future of Real-Time, Secure Data Processing
Edge Computing is the Future of Real-Time, Secure Data Processing Edge computing enables AI adoption across security, retail, autonomous systems, enterprise and more What is Edge Computing? Data is processed close to the source rather than in centralized data centers Select Data Data Public Cloud Ce
Slide 7: Blaize Demonstrates Improved Edge Performance vs. GPU
Blaize Demonstrates Improved Edge Performance vs. GPU Blaize offers better results for Video Analytics use cases vs. GPU solutions 2x 3x 2x 2x Source: Blaize-performed benchmarking tests using YOLOv8 7 Copyright © 2025 Blaize. All rights reserved.
Slide 8: Outperforming Competitors at the Edge
Outperforming Competitors at the Edge Complete solution designed to lead across Edge applications GSP GPU SoCs ASICs FPGA Task Level GSP architecture enables many Need large data sets and Can only handle specific Can only handle specific Requires careful design and Parallelism tasks to be run concur
Slide 9: Blaize's Solutions are Purpose-Built for Edge AI
Blaize’s Solutions are Purpose-Built for Edge AI Focused on what matters at the Edge, complementing the central processing power needed for GenAI Cloud Chip Providers Public Cloud Centralized Data Centers Edge Server Edge Devices Focused on More Intensive Compute Training and High Batch Inference Fo
Slide 10: Solution Overview
Blaize – Complete Solutions to Enable AI Inference at the Edge Full-stack hardware and software platform enabling AI to run at the Edge Built Through over a Decade of R&D with Close Customer Engagement Semiconductor Architecture Designed for Accelerators and Platforms for the Edge Edge Systems Solut
Slide 11: Blaize is Targeting Tremendous Market Growth Across AI Deployments
Blaize is Targeting Tremendous Market Growth Across AI Deployments Focused on high-growth verticals where Edge AI has the largest impact Traction in Large and Growing Verticals The Need for AI Inference Chips is 2028 Global Projected to Far Surpass the Need 2024 Global $ 26.3B Serviceable Addressabl
Slide 12: Blaize Products & Solutions Can Deliver Value Across a Vast Array of Applications
Blaize Products & Solutions Can Deliver Value Across a Vast Array of Applications Expected to address complex and unique use cases with its flexible, efficient architecture and programmability Security & Monitoring Enterprise Edge AI Autonomous Systems Tactical Edge / Defense Smart City Smart Retail
Slide 13: Illustrative Case Study – Security and Monitoring
Illustrative Case Study – Security and Monitoring Turnkey hardware and software solution enabling tactical security and monitoring for in-field defense Full Field Asset Tracking and Monitoring Proposed Solution Blaize® Pathfinder® P1600 Blaize® Xplorer® X600M M.2 SFF Embedded System on Module Accele
Slide 14: Illustrative Case Study – Healthcare
Illustrative Case Study – Healthcare Potential comprehensive solution for X-ray AI analysis for a large healthcare network Tuberculosis X-Ray Analysis with Blaize Solution That Has Been Piloted Blaize Multimodal Dashboards Patient X-Ray Patient Chest X-Image Acquisition for AI Based Analysis Key Ben
Slide 15: Illustrative Case Study – Automotive Design Work
Illustrative Case Study – Automotive Design Work Proof-of-concept of multi-target compiler to enable auto customers to deploy ADAS across any hardware platform Unifying Automotive Hardware: Blaize’s Multi-Target Auto Compiler unifies diverse hardware languages into a unified codebase, enabling seaml
Slide 16: New Breed of Architecture for Performance and Cost Savings
Innovation 1 New Breed of Architecture for Performance and Cost Savings Determines task dependencies to optimize scheduling, minimizing data trips to memory to save power & processing time Traditional CPU/GPU Architecture vs. Blaize Graph Streaming Processor Resulting Differences Tasks completed bef
Slide 17: With Advanced Flexibility, Blaize Does More
Innovation 2 With Advanced Flexibility, Blaize Does More Ability to run multiple workloads for different Edge use cases and reduce strain on CPU/GPU resources Blaize Covers More Workflow Pre-Processing 1 Frame Extraction Motion Detection Background Subtraction AI Model Inference 2 Blaize Can Seamles
Slide 18: Blaize Solutions are Software Programmable
Innovation 3 Blaize Solutions are Software Programmable Competitors sacrifice workload flexibility – Blaize maintains full adaptability AI models are being developed at an unprecedented pace, with new models introduced daily, necessitating adaptable A Programmable, Adaptable Solution technologies to
Slide 19: Blaize Software Swiftly Deploys AI for Real-Time Insights
Innovation 4 Blaize Software Swiftly Deploys AI for Real-Time Insights Supporting the entire development workflow from concept to application deployment at the Edge Problem Targeted Ideal Customer / Decision Maker Development Kit Allows customers and partners to Software scale algorithms on Blaize’s
Slide 20: No-Code AI Studio Accelerates AI Adoption
Innovation 5 No-Code AI Studio Accelerates AI Adoption AI Studio reduces anticipated deployment time and dependency on scarce and expensive engineering resources Penetrates Customers Beyond CTO AI Studio Organization into Traditional IT Departments Import data Accelerating Anticipated Deployment Tim
Slide 21: Architecture in Action – Enabling Multiple Tasks to be Run in Parallel
Architecture in Action – Enabling Multiple Tasks to be Run in Parallel Heterogeneous multiple parallel workloads at the lowest power in a single application Example: 5 Tasks Running in Parallel at the Same Time Task A Input Video Pre-Process Channel Extraction Image Enhancement Pixel Scaling Image S
Slide 22: World Class Suppliers and Engineering Footprint
World Class Suppliers and Engineering Footprint Deep relationships and capabilities across the technical and operational landscape Global Footprint Auto / Systems HQ / Architecture Core Engineering Al Studio Blaize AI Middle East Key Supplier Relationships 1 2 1. USA silicon manufacturing for all cu
Slide 23: Advanced Feature Set Today with a Strong Roadmap for Tomorrow
Advanced Feature Set Today with a Strong Roadmap for Tomorrow Building on a comprehensive technology foundation with advancements to meet customers’ future needs Current Key Features Future Product Enhancements Inference-focused: Tailored for Edge inference Chiplet product to meet customers’ without
Slide 24: Blaize is Strategically Positioned for this Technology Inflection Point
Blaize is Strategically Positioned for this Technology Inflection Point What is the Impact of DeepSeek’s “DeepSeek R1”/Alibaba’s “Qwen 2.5-Max” Contributions? • Combined multiple efficiency mechanisms for a significant reduction in computation and memory bandwidth • Demonstrated state-of-the-art mod
Slide 25: Technology Inflection Point
Technology Inflection Point Focus on efficiency that every industry experiences Qwen 2.5-Max • Mixture of Experts (MoE) approaches reduce • Memory bandwidth requirements • Computational requirements DeepSeek R1 • Mixture of Experts • Reduced memory bandwidth & compute • Multi-Token System • Generate
Slide 26: The Blaize Architecture
The Blaize Architecture The GSP* is purposefully designed for these types of efficiency transitions Architectural Features: • Support for sparse/fine-grained memory accesses needed by MoE’s and other imminent efficiency improvements. • Programmable architectures embrace algorithmic changes that can
Slide 27: Blaize's Experienced Leadership Team
Blaize’s Experienced Leadership Team Founders Senior Management Dinakar Munagala Satyaki Koneru Harminder Sehmi Dmitry Zakharchenko Tiffany Tan Doug Burns CEO CTO CFO Chief Software Officer Chief of Staff Chief People Officer Ke Yin Val Cook Alex Tinsley Santiago Fernandez-Gomez Joseph Sulistyo Chie
Slide 28: www.blaize.com
www.blaize.com 28 Copyright © 2024 Blaize. All rights reserved.