Toshiba Corporation

Toshiba's Technology Strategy

Dr. Shiro Saito CTO Corporate Executive Vice President Toshiba Corporation

CyberStrengths in digital PhysicalStrengths nourished in transformation for industry × manufacturing

We turn on the promise of a new day

Toshiba Group’s Vision

1 Technology Development to 2 Become a CPS Technology Company Activities for new business creation 3 In Closing 4 © 2019 Toshiba Corporation 3 1Toshiba Group’s Vision  Process towards Growth  CPS Technology Company Image Realized by Technology  R＆D Structure  R&D Investment

FY16 17 18 19 20 21 22 23

Toshiba Next Plan (Phase 2) Transform into a CPS Technology Company

Toshiba Next Plan (Phase 1) Strengthen Core Earning Power

Secure a sustainable financial base

Completed recovery of financial base;

shifting towards growth phase 2 © 2019 Toshiba Corporation 5 CPS Technology Company Image Realized by Technology Through R&D and open innovation, bring Edge capabilities to key components for existing businesses; use Toshiba’s IoT architecture to create new enterprise services

*1 *2 Enterprise Service DE DX

Social Manufacturing IoT Infrastructure Energy Logistics *3 IoT AI＋Genome IoT IoT IoT VPP Cyber Platform Highly intelligent technologies Data Analytics Operations AI Prediction, Diagnosis, Monitoring Technologies to enhance Sensing Communication Security Actuation Edge capabilities Use IoT and AI to provide functions that connect PhysicalComponents, Systems, People Cyber and Physical Existing New Outside company Power HDD/ New energy Electronics Industrial computers External component systems Digital Evolution (Tandem solar cells, P2C) *1 *TIRA compliant Digital Transformation *2 Robots Superconductivity Virtual Power Plant (Toshiba IoT Reference *3 Precision medicine Architecture) SCiB™ Building-related/POS (Micro RNA, Liposome) © 2019© 2019 Toshiba Toshiba Corporation Corporation 6 R＆D Structure Promotion of digital services, new businesses and IoT architecture deployment; three new organizations established* to strengthen security-related technologies *Cyber-Physical System Promotion Division, Digital Innovation Technology Center and Cyber Security Technology Center

Research & Corporate Research & Toshiba Corporation Development Division Development Center Cyber-Physical System Promotion Division Cyber Security Technology Center New Business Development Office Corporate Software Engineering & Technology Center Digital Innovation Technology Center Corporate Manufacturing Engineering Center

Key group companies R&D organization Overseas R&D Organization Energy Systems Research & Development Toshiba Research Europe Limited Toshiba Energy Systems & Solutions Corporation Center - Cambridge Research Laboratory - Bristol Research and Innovation Laboratory Toshiba Infrastructure Systems & Solutions Infrastructure Systems Research & Development Corporation Center Toshiba America Research, Inc. Electronic Devices & Storage Research & Toshiba (China) Co., Ltd. Toshiba Electronic Devices & Storage Corporation Development Center - R&D Center

Toshiba Digital Solutions Corporation Software & AI Technology Center Toshiba Software (India) Private Limited

Toshiba Tec Corporation Research & Development Center Toshiba Software Development (Vietnam) Co., Ltd. © 2019 Toshiba Corporation 7 R&D Investment Accelerate investments to support profit and medium- to long-term profit

R&D investment*1 Major R&D projects Total investment plan: JPY 930B (FY19-23) Energy systems Focus on growth fields such as clean energy (renewable energy, *2 Energy system Infrastructure system VPP , supercritical CO2 turbine, etc.), and asset management. Retail & Printing Solutions Storage & Device Infrastructure systems (B JPY) Digital Solutions Others Focus on fields that support growth, including SCiB™ application, robotics, highly efficient motors, power supply 200 systems, etc.

150 Electronic devices and storage Focus investment on the development of power devices (Si, SiC), automotive semiconductors, and HDD for data centers. 100 Digital solutions Focus on development for business model transformation 50 such as RECAIUS™ , and SATLYS™

0 Building solutions Continue investments to develop products that create new FY17 FY18 FY19 FY20 FY21 FY22 FY23 value and that align with target customers. *1 Before segment reclassification in April 2019 © 2019 Toshiba Corporation 8 *2 Virtual Power Plant 2Technology Development to Become a CPS Technology Company

Technology Development Policy Differentiate competitive components and enhance their capabilities for the Edge Reinforce AI-based digital technologies Deploy IoT reference architecture and services

© 2019 Toshiba Corporation 9 Technology Development Policy Create original CPS with competitive components and AI & IoT technologies Resolve social issues and maximize enterprise value through technology In FY2019, focus on realizing DE and DX deployment

Differentiate competitive components and enhance their capabilities for the Edge

Reinforce AI-based digital technologies

© 2019© 2019 Toshiba Toshiba Corporation Corporation 11 Increase and reinforce Edge capabilities of components Realize CPS for social infrastructure and energy by making full use of Toshiba’s customer base, technology and products Toshiba's Edge Capability Enhancement Components x AI & IoT technology create competitive edge components AI that support CPS IoT ・ Safe and highly reliable Battery ・ Power saving, low power consumption ・ High-speed response and autonomous control ・ Sensing and status monitoring ・ Automation and labor saving

© 2019© Toshiba 2019 Toshiba Corporation Corporation 1212 Edge Capability Enhancement Strategy: Power Electronics Pursuit of device performance x Intelligent control and circuits Promote development of edge devices that support social infrastructure and energy CPS

Highly reliable , New Development of next generation Si power devices materials high performance Source Gate

SiC Low voltage power MOSFETs for automobiles + power devices n P GaN ・Toshiba’s original trench structure and process for UMOS-X n- ・Industry’s top class device performance Next Low on-resistance 25% reduction FP (Compared with the 100V-class U-MOS VIII process, Drain generation Si according to our research , as of March 2019) MOSFET Intelligent IGBT control and circuits Enhancement of 8-inch process capability at Kaga Toshiba Electronics Corporation Built-in Built-in High Failure Network plug Protection functional FY17→FY20 driver IC circuit gate control prediction connection & play x1.5

① ② Expanded application: SiC power devices Enhanced ・ Supplied propulsion systems using All-SiC inverter to new railway vehicles*4 collaboration Intelligent control and circuit technology ・ Introduced the world’s first propulsion system using All-SiC inverter, with Cyber *5 *6 Totally Enclosed PMSM , and SCiB ™ World’s first ①Gate control: High-efficiency gate drive IC with high-speed switching*1 20-25% power reduction in normal operation and noise suppression West Japan Railway Company Tokyo Metro Co., Ltd. Marunouchi Line ②Device monitoring and protection: High-speed, multiple transmission isolator IC*2 volume 38%reduction World’s highest level multiplexing (3ch) & compared to current product highest-speed*3 (35MHz) transmission of isolated signals Series 227-1000 Series 2000 ② ① In operation from February 2019 *1 http://www.toshiba.co.jp/rdc/detail/1902_01.htm *4 https://www.toshiba.co.jp/infrastructure/en/news/20190314.htm © 2019 Toshiba Corporation 13 *2 IEEE international conference “ISSCC 2020”(San Francisco, Feb., 2020) *5 Permanent Magnet Synchronous Motor *3 As an isolation amplifier *6 https://www.toshiba.co.jp/cs/topics/back-number/20181011.htm Edge Capability Enhancement Strategy : Image Recognition Processor Visconti™5 Application of DNN*1-hardware IP realizes high speed, low-power and high accuracy recognition Contributes to driving that secures safety of vehicle surroundings, driver and passenger In mass-production Sample Concept TM *2 Visconti™4 Visconti 5 Shipment Future Strong recognition technology, Algorithm that Expand application of even at night ・Recognition of drivable area supports DNN DNN and AI application ・Simultaneous recognition of Enhanced driving support Limited autonomous driving multiple objects Front monitoring, automatic parking

Processing speed*3 Approx. Power efficiencyx10*3 DNN image recognition + 3D TM Prediction + Decisions using AI 「Visconti 4」 technology Toyota Motor Corporation Approx. vehicles equipped with Visconti™ x4 「ViscontiTM5」 DNN accelerater *4 Winner of Grand Prix, “2018 Japan ① Lower power consumption during DRAM access New Car Assessment Program Deployment/application Automobiles, surveillance Place temporal data retention SRAM close to executing unit (JNCAP)” Award camera, digital signage, ②Parallelization of multiply-accumulate units operation process 4 processors, each with 256 multiply-accumulate units AGV/drones, etc. ③ Lower power consumption during SRAM access *1 DNN is a machine model using artificial deep neural networks inspired by human brain function. Allowing a series of DNN calculations to be executed by one SRAM access *2 shipping of sample product from September, 2019 TM © 2019 Toshiba Corporation 14 *3 Compared with previous Toshiba’s SoC (presented at International conference “ISSCC 2015”). Visconti is a trademark of Toshiba Electronic Devices & Storage Corporation. *4 https://toshiba.semicon-storage.com/ap-en/product/automotive/image-recognition.html All other company names, product names and service names may be trademarks of their respective companies. History of Rechargeable Battery Development - Toshiba's Contributions 1976 1980 1985 1991 2004 2008 Present day

S. Whittingham Akira Yoshino Lithium-ion battery Start of SCiB™ mass prod. 2019 Nobel Prize 2019 Nobel Prize for Chemistry for Chemistry Commercialization Li/TiS2 battery proposal Established basic concepts (Sony Corporation) Widely used in automotive (electrode reaction basics) of lithium-ion battery Kashiwazaki Works and industrial fields Manufacturing Base

Discovery of lithium cobalt oxide cathode material Development of rechargeable battery with lithium titanium oxide negative electrode Widely used as the positive electrode material Realization of a highly-safe, long-life, rapid charging rechargeable In lithium-ion batteries used in mobile devices battery that contributes to solving environmental and energy problems

Main uses Main uses

Cell phones Laptop PCs Power storage systems Li EV・PHEV for stationary use

Smart phones Toshiba Executive Fellow Norio Takami, Hiroki Inagaki "Prime Minister Award" Koichi Mizushima 2016 NIMS Award National Commendation for J.B. Goodenough, 2 others Invention 2019

K. Mizushima, P.C. Jones, P.J. Wiseman, J.B. Goodenough: Mat. Res. Bull. 15, 783 (1980) “Large long-life rechargeable battery with excellent charge-discharge performance" University of Tokyo Toshiba Special Award President’s Special Award © 2019 Toshiba Corporation 15 Edge Capability Enhancement Strategy : SCiB™ Higher capacity & higher output x Battery service Respond to potential markets and needs by leveraging SCiB™ strengths

New material Automotive (HEV, commercial vehicles) Cycle characteristics Shipped 39M cells Zero accidents ・ industrial (railways, AGV, etc.) more than 20,000 times New structure Higher capacity and higher output -30℃ NTO Car sharing, EV buses, anode railways, power system

Battery as a Service LTO anode Battery diagnosis and controls Battery system O&M Charging curve analysis (CCA) monitoring 6 minute

SCiB™ high capacity technology Battery life diagnostic technology • Joint development of Accurate battery life estimation by diagnosis of degradation at material level Next generation anode material • *1 battery grade niobium Applicable to a Li-ion battery using a variety of cathode and anode material NTO anode achieves capacity Power Cathode material material with CBMM in Internal 1.5 times during ultra-rapid charging *2 Storage Module Cell Electrode Material Brazil System State Shift of operation Cathode SOC*3 range in x1.5 Battery capacity cathode and anode Battery Anode Potential Anode material rapid charging -

Drive rangeDrive Capacity Toshiba Method CCA estimates the internal state of battery, based on the curves of the cathode and the anode material, by

at ultra Conventional approach LTO anode cell NTO anode cell performing fitting to the charging data. © 2019 Toshiba Corporation 16 *1 Titanium Niobium oxide *2 https://www.toshiba.co.jp/cs/en/topics/back-number/20180619.htm *3 SOC：State of Charge Edge Capability Enhancement Strategy : SCiB™

As advances in mobility realize the start of MaaS and other changes, Technologies that enhance edge capabilities will promote SCiB™ service businesses providing safe, secure battery systems

TM Technology and product capabilities Switch from sell out to service business SCiB• Long life Customer data/Service fee • Rapid charging Toshiba

Battery Information Service Customer Battery residual values company Battery life diagnosis and AI technology diagnostic technology Battery Service • Digitization of value Selling information Highly accurate diagnosis System • Battery Long life and safe operation with charge control Customer base built Toshiba+ Partner through co-creation Customers• Diverse operating achievements in Provide Service automotive, railways, power system and AGV subscription contracts Value supply as a Realize market expansion total Watt hour per price(ΣWh) and sustained, stable profit © 2019 Toshiba Corporation 17 Edge Capability Enhancement Strategy : Automation and Labor Savings with Intelligent Robots Bring intelligence to clumsy robots capable only of pre-set tasks Achieve unsupervised automation of non-routine work in logistics, manufacturing, and retail

Strengths of Toshiba Robots Rapid Automation system integration technology Advanced AI technology Toshiba Intelligent Robots Recognition and sensing technology Providing solutions for labor shortages Gentle Precise and increased workloads, etc. 1967 ~ 1970s ~ Over 50 years Postal sorting Various AI technology machines manufacturing research Holding heavy Skilfully gripping and High-speed automatic facilities loads carefully packing without gaps adjustment of probe angle De-palletizer Piece-picking Robot Weld Inspection robot World first handwritten Electrolyte injection Accumulated data, TM post code reader sealing device for SCiB domain knowledge

Core technologies for intelligent robots Technology for situation appropriate autonomous movement 19/3 Delivered 19/12 Scheduled to deliver 20/1 Scheduled to start the first machine the exhibition machine PoC at customer Recognition Planning Mechanism control Efficient transfers by Unsupervised Automatic in-store technology technology technology AGV with arm autonomous transportation check of price tags Mobile palletizer Cart-transporting robot Shelf-monitoring robot

・High precision ・Motion planning to ・Firm grip on heavy object region segmentation avoid collisions objects 20/1 Scheduled to deliver 19/8 Implemented 18/5 Implemented ・High-speed ・Optimum packing and ・Exerted force feedback the first machine PoC at customer PoC at customer object posture estimation loading control © 2019 Toshiba Corporation 18 Reinforce AI-based digital technologies

© 2019© 2019 Toshiba Toshiba Corporation Corporation 19 The Strengths of Toshiba AI Technology AI technologies and data Toshiba has accumulated enable deeper analysis of customer data, and realize solutions for customer issues in the CPS domain, such as manufacturing and maintenance

Point 1 8,290 Many years of AI technology 5,930 development 5,223 5,102 1956 1970 1980 1995 2010 2015 2019 4,406 1st AI boom 2nd AI boom 3rd AI boom Search and Knowledge Machine learning/ inference expression deep learning

Over 50 years of history in spoken language, image processing and analytics ⇒ Cumulative total of AI-related patents: No. 3 in the world, No. 1 in Japan* IBM Microsoft SAMSUNG NEC Modified from Technical Trend Report published by the World Intellectual Property Organization (WIPO)

Analysis Prediction Optimal planning ・ Equipment failure ・ Plant operation plan A “site” ・ Product defect analysis ・ Infrastructure prediction ・ Railway operation plan AI tech. Point 2 full of data degradation analysis ・ Power demand ・ Infrastructure 180 DB Toshiba’s manufacturing and maintenance prediction maintenance plan items has accumulated master data and provided Recognition Judgment and control Interaction AI technologies A “site” ・ In-vehicle image ・ Robot motion planning ・ Spoken dialogue system are ready to use where recognition ・ Path planning for within Toshiba a place to put it into practice ・ Object recognition for autonomous driving people are logistics group involved ・ People tracking for © 2019 Toshiba Corporation 20 crime prevention etc. Enhancement of AI Technology From AI with manual intervention to AI that learns from limited data Ultimately advance to self-learning AI

Field Accumulated data, AI DeepCo cultivation-creation of cutting- domain knowledge edge AI technology “Enhancement at business fields “Enhancement at cutting-edge research laboratories” with large amount of data” Technology development close to on-site issues Deep cultivation through open innovation Apply deep-learning to automated HMLasso*1 identifies failure factors from classification of defects in semiconductors Combining limited supervised data with many missing values learning with transfer learning *2 achieves highly accurate Joint research with the Institute of Statistical• Mathematics High-speed, high-accuracy identification automatic defect of causes of quality and yield classification with less deterioration teaching • Reduction in estimation error of Defective part approximately 41% against the cutting- Inspection data Used in-house semiconductor factory edge algorithm

Presentation at International Joint Conference on Artificial Intelligence (IJCAI), one of Winner of ISSM2018 Best Paper Award（FY2018） the most important international conferences in the AI field (FY2019)

Structured weakly supervised learning Unsupervised learning AI with Supervised learning AI that learning Semi-supervised Transfer learning learns from Automatic manual learning Self-learning intervention limited data knowledge AI Full information learning Incomplete information learning acquisition Explainable AI *1 HMLasso (Least Absolute Shrinkage and Selection Operation with High Missing rate) © 2019 Toshiba Corporation 21 *2 http://www.toshiba.co.jp/rdc/rd/detail_e/e1908_01.html *3 CoCoLasso (Convex Conditional Lasso) Ref. ：Datta, A., & Zou, H. (2017). CoCoLasso for high-dimensional error-in-variables Examples of AI Technology Development Toshiba AI technology generated from co-creation and field capabilities

Strengthen Deep Learning tech. through open innovation World-class performance that achieves Singular Value with joint research by RIKEN AIP Weight Matrix Decomposition ・Scalable DNN that adjusts computation level after learning ・Parallel distributed learning that reduces learning time Joint research by RIKEN AIP*1 and Toshiba Collaboration Center*1 Input Layer Hiden Layer Output Layer https://www.toshiba.co.jp/about/press/2018_12/pr1702.htm

グランプリ

PV Power Generation Prediction Grand Prix winner in “PV in HOKKAIDO” PV Agreement on train timetable planning *2 Adoption of coestation as Power Generation Prediction Contest *3 “virtual operator” project with Greater Anglia, a train TEPCO Holdings, Inc. operator in the U.K.*4 Hokkaido Electric Power Co., Inc. *3 https://www.toshiba-sol.co.jp/news/detail/20190327.htm © 2019 Toshiba Corporation 22 *2 https://www.toshiba.co.jp/rdc/detail/1907_02.htm *4 https://www.toshiba-sol.co.jp/en/news/detail/20190910.htm AI Human Resources Development Reinforce AI human resources that support CPS by introducing practical education using actual data

AI engineer training program developed with the University of Tokyo*1 2,000People Train AI engineers at a scale of 100 per year About 3x 10 sessions in total (a 1.5-hour lecture people increase 350 and 2 hours of exercises for each) Lectures and A wide range of study, from classic machine New employees  learning to the latest deep learning Focus on exercises for practical use （200 people） exercises

In-house training, etc. Final assignment problem period of about 1 month *2 （700 people） Examine and implement assignments using real big data 750 people owned by the company, etc. Presentations on final assignments in poster format Utilize knowledge and programming Presentation experience gained in lectures and exercises  to present results of final assignment Develop problem-solving skills through implementation Now FY2022 *1: Graduate School of Information Science and Technology - University of Tokyo © 2019 Toshiba Corporation 23 *2: Workplace challenges can be used DE DX

Social AI Infrastructure Energy Manufacturing IoT IoT Logistics VPP ＋Genome IoT IoT IoT

Deploy IoT reference architecture and services

Remote Monitoring Service for Railway Vehicles Dashboards Social Remote management and maintenance service for chillers Real-time plant efficiency monitoring Infrastructure Building wellness service

Energy Failure prediction Meister Cloud™ Series for manufacturing IoT Manufacturing Distributed & coupled simulation platform for in-vehicle control model Optimal power generation planning service AI image inspection service Design and maintenance data linkage between documents Logistics Logistics IoT cloud service

© 2019 Toshiba Corporation 25 IoT Service Development Radio Railways Building Water Roads in Social Infrastructure Business waves Accelerate the development of attractive systems and their application to IoT services, by taking advantage of domain knowledge and components in social infrastructure

Expand “As a Service” (Railways) Platform business Customer asset DE/DX Transit Scheduling ICT Solution management service Step2： Maintenance service Develop platform business Business service for Business services making things visible utilizing collected data Step1： 12 Develop individually Remote monitoring Service Failure analysis Life cycle assessment customized digital service support service service service

Railway vehicle Information Power supply Fare collection Physical system system system system ・Components ・Systems ・Domain Propulsion, Power supply, Train traffic control, SCADA for power supply Automated ticket gates, knowledge Air-conditioning, Transportation planning system, Fare collection management Train information Transmission, Power feeding © 2019 Toshiba Corporation 26 IoT Service Development Radio Railways Building Water Roads in Social Infrastructure Business waves Contribute to our customers’ business by CPS-loop connecting cyber ＆ physical world

Toshiba Customer (Manufacturing ) (O&M) Adoption of remote monitoring service for railway vehicles Cyber 12 Produ- Opera- Maint- Service Design citon tion enance Data Data Image provided by Japan Freight Railway Company Asset Management Manufacturing Quality and Improvement of stable transportation, Data Integration Improvement Failure prediction, Productivity Improvement for social infrastructure Improvement of operational efficiency for our products (H/W, S/W, Service) owned by customers

 Risk reduction for public AI  Prediction Sensing  Optimization sewerage system CPS-Loop Recognition Physical

Better Efficiency Optimization of pump control with Customers’ operation and prediction of rainwater inflow  maintenance fields Toshiba’s manufacturing field sites © 2019 Toshiba Corporation 27 IoT Service Development in Energy Business With Toshiba’s deep expertise in power plants, Provide IoT solutions that meet customer needs

Expand “As a Service” Use prediction and optimization technologies to realize and provide Total value chain optimization service DE/DX Collaboration with other companies through open API technology Step2： Develop platform business Extend proven in-house plant technology to external power plants

Dashboard service Design and maintenance data linkage Step1： between documents Develop individually Performance monitoring to provide performance customized digital evaluation and abnormality prediction service Failure prediction using operation data service

Develop businesses in response to diverse potential needs Physical AND ・Components Steam turbine Central load- Adjustable-speed ・Systems dispatching office pumped-storage ・Domain No.1 share system hydropower system Turbine and combustor Tandem solar cell Autonomous hydrogen in North America share for supercritical CO2 cycle development energy supply system knowledge ※On an installed capacity basis. Source: No.1 No.1 TM” Toshiba (2004~2018 Cumulative Actual)) in Japan Global share power plant “H2One ※Plant number base

Customer Operation data / work data Data access using information models Abundant ＋ By using engineering data and know-how, experience in Toshiba Engineering data data can be handled in a structured form existing business / Know-how ＋ Proactively expand into focus business Detecting signs of equipment anomalies contributes to improving thermal efficiency and areas increased power plant availability

・De-carbonization Start of demonstration experiment at ・Diversification of electric power supply CPS Saltillo Thermal Power Station, ・Effective use of energy Physical Cyber Mexico Loop owned by Mitsui & Co. and NMF Consulting Holdings , etc (invested in by Mitsui)

Component

Feedback to Physical ＋ Press announcement released on Improve operability of November 1, 2019 customer energy assets © 2019 Toshiba Corporation 29 Use of Manufacturing IoT & Service Development, with examples Using technologies and know-how in manufacturing X cutting-edge digital technology, Implementing innovation in manufacturing and construction of digital base inside and outside of company

Expand “As a Service” External Productivity application improvement by DE/DX Manufacturing IoT solutions standard IoT base Unified data base for Denso Corp. manufacturing Toshiba’s AI Utilize Data utilization, technology Facility management Advanced analysis Nachi Fujikoshi Corp. Step2： using IoT/big data Develop platform Manufacturing Storage Integrated (under verification) business and O&M data data base modelling Collect KIOXIA Corp. IoT data collection Aggregation,visualiza- Internal Unified data base for tion and traceability manufacturing Step1： Precisely reproduce sites of manufacturing deployment of manufacturing data Develop individually and usage of products in digital space Himeji Operations - Kashiwazaki customized digital Semiconductor Operations service Digital twin

Function, data model, Innovative consulting for utilization know-how Corporate Manufacturing manufacturing based on Physical enshrined at site Engineering Center abundant experience ・Master data Production system Information Material and device Technology and engineering tech. system tech. process tech. Contribute to enhancement of ・Domain knowledge Mechanical structure Manufacturing know-how in business strength using knowledge Packaging tech. design and manufacturing ・Field capabilities Technology Manufacturing site and experience nourished at tech. manufacturing sites Control Optical Mechatronics Electronic device that Toshiba Group have tech. tech. tech. packaging tech. cultivated © 2019 Toshiba Corporation 30 Activities3 for new business creation

© 2019 Toshiba Corporation 31 Toshiba’s Efforts to Resolve Social Issues – New Business Creation New Business Development Office established in January 2019 Accelerate business incubation and open innovation

*1 CVC: Corporate Venture Capital

*1 Business CVC function incubation (fund of approx. function 10 billion yen)

  Build a platform for business incubation based Invest in three venture capital companies in the digital, life science and energy fields  on the lean startup methodology Working toward commercialization of quantum key distribution system with two companies*2, one in the US and one in Europe  Build a platform to create synergies between *2 Quantum Xchange, Inc., British Telecommunications plc VB networks and existing business units © 2019 Toshiba Corporation 32 Cutting-edge Technologies to Resolve Social Issues *4 Disease prediction / prevention AI Japonica array*2 Micro RNA*3 Biodegradable liposome Precision medicine Random survival forest Learning model optimization Diabetes severity analysis *1 *3 Collaboration with Medical knowledge *2 Collaboration with National Cancer Center Japan/ *1 Collaboration with Kanazawa University Tohoku University Tokyo medical University *4 Collaboration with Shinsyu University

Tandem solar cells Power to Chemicals*5 Turbine and Combustor for Supercritical CO2 cycle power plant Energy Cell

Cell 1.9m

*5 Verified in the national project by Ministry of the Environment (FY18-22)

Quantum cryptographic Simulated bifurcation Quantum bifurcation *6 *7

communication machine ポテンシャルエネルギー Machine Optimal Quantum applications energy Potential solution 001 Excitation励起強度 field 量子分岐現象Quantum 2 strength X bifurcation phenomena

KPOネットワーク Network *6 Cross-ministerial Strategic *7 The Exploratory Research for Advanced© 2019 Toshiba Technology Corporation (ERATO33), Innovation Promotion Program (SIP) X1 MEXT - Quantum Leap Flagship Program(MEXT Q-LEAP) Cutting-edge Technology - Precision Medicine

Diseases Prediction AI ＋ Prevention of Diabetic Nephropathy Cancer detection technology using Micro RNA

【Disease Prevention】 High sensitivity Ultra-early detection of *1 Achieved world’s top level prediction accuracy of 96% Micro RNA chip cancer using health check in prediction of risk of diabetes. blood tests *1: Risk prediction accuracy for diabetes in three years Can detect cancer in 13 organs ＜Highly accurate AI utilizing high quality data of several years＞ Space-saving Learning model 1 drop of blood, Random survival forest inspection device optimization technology in only 2 hours Identify about 99% of cancer Average results of patients several decision trees Collaborating on implementation with the National Medical verification is Cancer Center and Tokyo Medical University 1 possible 2 High robustness Possible to extract the impact Realized high accuracy using ＜Cancer detection technology using Micro RNA＞ of individual input item against data set provided by Tokyo 1 Cancers in 13 prediction Midtown Clinic organs Prevent worsening of diabetic Identify cancers 【Prevent worsening】 in 13 organs Started joint research nephropathy in patients (dialysis) (e.g. breast, stomach) Kanazawa University Long-term clinical Stratify patients 2 og (Concentration of Micro of Micro RNA) (Concentration og

Toshiba l information/ Data to suggest At stage 0 Identify cancers at Healthy cancer identification Pathological information analysis AI optimized treatment Medical knowledge stage 0 individuals for 13 organs © 2019 Toshiba Corporation 34 Cutting-edge Technology - Energy

Tandem solar cells Power to Chemicals (P2C) Semi-permanent drive for electronic mobility without charging Achieved high ton-per-year throughput and saved space High-power supply with a small footprint at low cost Available to adjacent to the CO2 emission facility Renewable energy P2C device Transparent Cu2O Solar Cell Cu2O+Si Tandem solar cell (image of Carbon Dioxide verification device) Word’s first Transparent Long-wavelength Cell

2 CO2 electrolysis cells 1.9m Cu O light Cell Short-wavelength light CO, H2 CO2 throughput 10.6 ton/y（Max） Si Valuable Chemicals Verified in the national project by (FY18-22) ＜Low cost tandem solar cell stacking Cu2O and Si＞ Image of P2C plant Ministry of the Environment ＜CO2 electrolysis that realizes high ton per year throughput＞ 2 1 Transparent Cu O solar cell Low cost Cu2O+Si Succeeded in creating transparent Cu2O solar cell ❶Transparency Tandem by controlling the presence of oxygen ❸High efficiency ・EV ・Robot World’s fastest CO2 conversion rate Residence Si 1 2 ・Convenience Mega solar CO2 conversion rate of 645mA/cm Stores etc CO2 electrolysis cell 2 Low cost ※target value Less than 1/100 in comparison Efficiency Cell structure that directly uses CO2 with one of GaAs tandem 15% 20% 25% 30% 35% 2 ❷ low cost Direct CO2 gas reaction by three-phase （less than 1/100） Three-phase boundary control technology 3 Higher efficiency than Si GaAs boundary control Achieved higher efficiency than Si High cost Tandem 3 Porous catalyst electrode by combining Cu2O and Si in tandem with superior CO2 diffusion Catalyst electrode © 2019 Toshiba Corporation 35 Cutting-edge Technology - Quantum Applications

Simulated bifurcation machine Quantum cryptographic communication Resolving combinational optimization problems on the world’s fastest Development and standardization of platform technology for service business and largest scale

Taking full advantage of a high-speed algorithm Secure communication platform safe from risk Simulated of decryption attack by quantum computers bifurcation in the world’s fastest automated algorithm financial transaction system dx ∂H i = = Dy ∂ i Coexistence with existing dt yi dy ∂H i = − = −(D A arbitrage trading machine with optical fiber networks; promote dt ∂x i N 2 as a platform through interface −(D − p + xi )xi − cαhi + c∑ Ji, j x j this algorithm can issue trade j=1 and network configuration orders in (microsecond) standardization 30 µs Collaborate with partners to solve problems in areas other than finance, ＜Platformization of Toshiba’s original such as genetic analysis quantum cryptographic communication technology＞＜FPGA implementation of Arbitrage trading machine＞ 1 Unique wavelength application technology Arbitrage trading loop ・Coexistence with data multiplexing transmission and quantum 1 High-speed response cryptography communication at O-band (1.3μｍ light and solution accuracy wavelength band) ・Even with a 30μs response, best solution ・Achieved the world’s fastest quantum key distribution speed accuracy is 91%. Captures 1ms arbitrage and proposed design for world’s longest distance trading opportunities. communication on C-band (1.55μm light wavelength band) 2 Easy to install 2 A platform that provides cryptographic keys ・Can be installed on a commercially ・A platform that provides encryption keys for various applications available FPGA. No need for with standardized encryption key provision interface installation on large-scale equipment. © 2019 Toshiba Corporation 36 4In Closing

CyberStrengths in digital PhysicalStrengths nourished in transformation for industry × manufacturing

We turn on the promise of a new day