The International Symposium on Quality Electronic Design (ISQED) 2021 started today virtually in the USA. It is the premier electronic design conference. . ISQED bridges the gap between electronic/semiconductor ecosystem members, providing electronic design tools, IC technologies, packaging, assembly and test, semiconductor, etc., to achieve total design quality.
ISQED is the leading conference for design for manufacturability (DFM) and quality (DFQ) issues. ISQED emphasizes a holistic approach toward design quality to highlight and accelerate co-operation among the IC design, EDA, wafer foundry and manufacturing communities.
Arun Venkatachar, VP, AI and Central Engineering, Synopsys, presented the keynote on the confluence of AI/ML with EDA and software engineering. He talked about how AI/ML can help in chip design and product development.
Chip design a tough game to play. There are deluge of challenges. There is debug, DFM, DPT, etc. We can leverage AI and Big Data to design silicon faster, and more cost-effectively. There are connected analytics, insights, time-series, patterns, etc. Algorithms generate a ton of data. Data has become the epicenter. Synopsys has three different vectors of innovation: enabling AI chips, AI-enhanced tools, and AI-driven apps. AI/ML looks at the data.
Synopsys has AI-enhanced tools and apps. These improve the performance, QoR, and productivity, beyond what is possible algorithmically. They are also using RL for design space optimization. ML enables new way of thinking about design.
Another example is VC LP, or faster violation debug with ML. There are manufacturing-related opportunities. AI/ML use cases have finally gone into production at customers. Customers are also more savvier, and understand the importance of good data diligence. Deployment of AI solutions are different than current EDA product deployments. However, not all problems can be solved with AI/ML. Confluence of data, algorithm, etc., is need.
AI/ML can also help build better EDA products, leading to better software engineering. Systemic complexity growth has been happening in product development. Products and engineering complexity is also increasing. We need to improve the release quality and predictability, improve R&D productivity, etc.
Quality can be managed by design, such as preventive measures and built-in quality, validation, such as test and failure analysis, and defect management, such as responsiveness and support. Path to actionable insights need data points, Big Data, and intelligence. AI/ML takes the insights and starts to predict. We can also do quality-by-analytics. You need to know the defects, tests, and code. Insights enable shift-left in quality and improves productivity. Shift-left strategy is enabled by quality-by-analytics and information at the disposal of the developer.
Synopsys has ML-infused apps. There is CodeQuarry, Plan Better, Failure Triage, Bug Triage, Intelligent Test Selection, Release Analytics, and Predictive Score. An example is the code hotspot analysis tool. We need to identify the functions that are hot. This will prioritize the R&D work that yields high RoI.
In bug triaging, new bugs are automatically compared to others, and clustered, based on stack similarity. There is also the check-risk analysis. We need to identify who should review the code change, whether dependent code modules need to be considered, related bugs, etc. Today, you can link and search collaterals across the organization, using NLP.
You need to establish a unified data management strategy. Streaming data access on a unified data platform can enable a true ecosystem via data sharing. Connected analytics can yield key insights and open up new avenues. Tap into the convergence! It all starts with good data-diligent approach and process management. Use AI/ML as a new paradigm shift to improve quality, productivity, and efficiency.
Emmanuel Sabonnadiere, CEO, Leti of CEA Tech, presented on Edge AI and High-Performance Computing, at the ongoing SEMI Technology Unites global summit.
There will be lot of connectivity in the future. Connectivity between the digital world and the real world will accelerate. The vaccine was developed within 12 months, due to biotechnology and bio-electronics. 2020 told us that hardware was back. Big cloud players are investing in hardware. Eg., Tesla. The value is 3-4 times bigger than competitors, as it has demonstrated that it is managing all its electronics and devices.
Mobile phone is, and will remain extremely centric in microelectronics. Over 10 billion units were sold in the market. There could be six devices per user by 2030. We have to continue to invest highly in mobile. AI is also an important part. Energy consumption is now linked with AI and the edge. AI for IoT will be less than 1mW in terms of consumption.
At CEA-Leti, there has been development of cloud. AI will be developed in the cloud. There will be edge AI. The roadmap till 2030 is based on six technologies. In the memory part, CEA Leti has done advanced innovation. Most of the data has to be treated locally. In future, computation can be increased by adding chiplets. To develop edge AI, we also need new platforms. With imec in Belgium, and Fraunhofer in Germany, together, we have created More than Moore technology. We have one of the most advanced SoCs for next generation of edge AI chips.
We also have to be on par with the green equation. We are generating a portion of CO2 worldwide. We are looking at 60 billion connected devices by 2030. The next big thing is to turn our minds to how we develop future technology. Performance means better productivity. Now, there is performance and consumption. For the future, we are looking at these areas. In future, we hope to create something useful with the world, and improve the green equation.
What’s next in AI?
Mukesh Khare, IBM Research, presented on What’s Next in AI: Our Vision for the Future of AI Hardware, at the ongoing SEMI Technology Unites global summit.
The next generation of advanced computing will demonstrate our power. We will communicate results that build on the existing knowledge. We also need to invent a new computing paradigm. We can solve this using AI hardware. We need to tap quantum computing through the cloud. The new hybrid cloud environment will bring virtual computing power.
IBM Research is building AI with fluid intelligence. We want memory and bandwidth to further mature. AI is stretching today’s computing hardware. The IBM AI center’s goal is to improve computing performance by 1,000 times by 2029. It will enable more holistic computing. Red Hat is working to build a software stack. AI center is also launching the third generation of chip. We are building a purpose-built AI hardware. The new chip is a marvel. It comes with in-memory computing hardware.
IBM Research also launched a new packaging center in Albany. We are also working to bring AI for business and industry automation. We have enabled models and methods that will help solve future problems.
Ms. Maria Marced, President, TSMC Europe, presented on Shaping the Digital Transformation, at the ongoing Technology Unites global summit, organized by SEMI. 2020 was a difficult and challenging year. We have learnt new ways to live and work, and also learn.
There has been big momentum in 5G and IoT. 5G and HPC are now driving digital transformation. Innovation is also driving growth in the automotive industry. ADAS level 3, 4, and 5 connectivity and electrification is expected by 2030.
The 2020 global semiconductor market growth was asynchronous to the world real GDP growth. This growth was raised from 6 percent to 10 percent by late Dec. 2020. There is now a surge in demand for digital transformation.
TSMC has 6- and 8-inch, and 12-inch giga fabs. It also has advanced packaging. It will install a 12-inch fab in Phoenix, Arizona, USA, in 2021. We are also going to raise the capex. TSMC has capacity leadership. It has also invested in green manufacturing. TSMC has also been trying to achieve manufacturing excellence.
TSMC also has technology leadership.
Today, TSMC’s innovations in technology are driving digital transformation. It has achieved N5 in production, N3 in development, and N2 in path finding.
Next, Simon Segars, CEO, ARM Holding Plc, presented on Rebuilding Better for the Digital Future. We have learnt a lot from 2020. We have shifted to doing so many things online. The networks have survived the massive surge in activity. People talk about a new normal, and we need to think what it will be.
The private sector and governments have to engage in global climate projects. Air pollution was somewhat down in 2020. ARM has promised to be net zero carbon by 2030. We have taken the same ethos to processes and data centers. We need everyone involved in production of electronics and data centers to think about how we can further reduce energy. We need to think about how we can do all this in a more efficient manner.
Healthcare and telehealth have made some strides in 2020. We can already start using wearables and sensors. Home monitoring is however, not yet developed. We need to take all the information from the electronics and move to healthcare.
Education has undergone similar shift in 2020. Educationists have been challenged to change their approach. You also have to maintain the content on the screen. For children, it is also challenging. AR/VR has been around for some time. An example is the Tower of London. We need to improve the accessibilities. We also need to develop trust. Digital healthcare especially, needs trust.
Semiconductor sector has had a great 2020. We need not run the risk of missing out on big opportunities. Now is the time to engage. There is momentum in how people are now thinking about technology. We need to think through the hard challenges.
Edge AI + Vision Alliance and SiliconExpert held a conference today on: Adding Visual AI/Computer Vision to Embedded Systems. There was special focus on the key things that every engineering manager should know.
Phil Lapsley, Co-founder, BDTI, and VP of Special Projects, Edge AI and Vision Alliance, explained the typical embedded visual AI system. He touched upon the typical visual AI tasks and apps. Data is key for AI projects. Understanding exactly what you are trying to do will drive your data collection efforts.
Jeff Bier, co-founder, president and engineering manager at BDTI, and Founder, Edge AI and Vision Alliance, added that most ‘visual AI’ is performed by neural networks (aka ‘deep learning’ or ‘deep neural networks’). Neural networks require a very different way of thinking about the software development process. Classification and detection of the images is very important. Example apps are detecting package on your front porch, detect person in your backyard, advanced driver assistance, detect defects in the production line, etc.
Lapsley said that AI is not magic! It can do some things better than humans, but it has limitations. The laws of physics still apply.
Bier noted that data sets are required for testing and training. At least three large, independent data sets are required to train and test a neural network:
- Training set: data shown to the neural network for its learning process (i.e., training).
- Test set: data used during development to evaluate the health of the training process.
- Product-level test set: Data used to evaluate the performance of the product as it might behave in the real world.
Collection, screening, labeling and maintaining relevant data is the most time consuming (and critical) part of deep-learning-based product development. It usually requires numerous iterations and continual expansion. The nature of the data required is usually product-specific. A large ‘lake’ of high-quality data is extremely valuable.
Focus on precision
Precision is a metric that answers the question: “If I show the system an unknown image, and it says, it’s a cat, how sure am I that it’s a cat?” Lapsley said that training is a lengthy process that needs to be done with the data.
ML accuracy metrics are often misleading and lack critical context. It is important to understand what is presented and ask the right questions. Most commonly used performance metrics are:
- Accuracy—most often the average accuracy is given. This is the most misleading and vague metric of them all.
- Per-class precision—if the system tells me an image is a cat, what’s the chance it’s correct?
- Per-class recall—if I show the system a cat, what’s the chance it correctly recognizes it as a cat?
- Per-class false negative rate—the proportion of cases or events of a given class that were incorrectly excluded.
- Per-class false positive rate—the proportion of cases or events of a given class that were incorrectly included.
Maintaining control of data and experiments is critical. You need to know what data the model was trained on. Most applications start with an existing neural network. You can also develop a neural network from scratch. Popular neural networks for embedded applications are:
- Classification: Inception v3, ResNet-50, ResNet-34, and MobileNet v2.
- Object detection: MobileNet v2-SSD, YoloV3, V4, V5 (different variants).
You can do visual AI on a wide range of processors. The right one depends on your needs and constraints. Do remember that results fluctuate. Your model’s accuracy doesn’t just get better and better. As you run the experiments, you will have wins and losses. You will get different performance in the real-world. Different skill sets are usually needed. For instance, any C or C++ embedded programmer isn’t necessarily who you want training your neural network. It is all about the data. Test and training data costs money and takes time to acquire.
Celebrating 10 years this May, the 2021 Embedded Visual Summit will be held virtually from May 25-27, 2021. There will be over 60+ talks on computer vision and AI, along with demos and technologies.
Semiconductor Industry Association (SIA) held a conference to make sense of the trends that shaped the global semiconductor market in 2020, and look ahead to what is in store for 2021.
The participants were Andrea Lati, VP, Market Research, VLSI Research, Dale Ford, Chief Analyst, Electronic Components Industry Association (ECIA), and CJ Muse, Senior MD, Head of Global Semiconductor Research, Evercore ISI. Falan Yinug, Director, Industry Statistics and Economic Policy, Semiconductor Industry Association, was the moderator.
Andrea Lati, VLSI Research, said that for the IC trend, the growth has accelerated since November 2020 as component shortages had strengthened prices. DRAM and NAND started 2020 very strong. They dropped during the middle of the year. The rebound happened during H2-2020.
IC recovery has since been sustainable, including for analog, power, etc. IC inventories have also been improving. They were running 8 percent above a year ago, in December 2020. The chip price performance index (CPPI) was relatively flat in 2020 despite high inventories in memory. Steady increase in Q4-2020 bodes well for 2021 prospects.
Dale Ford, ECIA, said there was a whipsaw disaster in 2020 that required a nimble response. There was a supply chain impact in 2020 due to the government quarantine orders and directives on company’s workforce and operations. Things calmed down after a time. The index of concern was quite high by August 2020, but, it is now coming down. Q4-2020 data will see numbers improve.
CJ Muse, Evercore ISI, said the semiconductor industry had gone through a correction in 2019. We were set up very well during 2020. Things were not as bad as feared in 2020. TSMC revenues were flat in April. There was an over reaction in automotive production. By September, TSMC-Huawei embargo had happened.
What’s ahead in 2021?
Andrea Lati noted that there will be continued worldwide GDP growth in CY21 from 2H20. Cloud and hyperscale datacenter will be a key drivers for the semiconductor industry. Hyperscale capex is at an all-time high. Cloud investments are supported by strong financial performance. 5G proliferation will be another big driver. 5G smartphone shipments will double in 2021. There will be increasing deployments for 5G base stations.
VLSI Research has forecasted 12 percent growth in semiconductors for 2021. Memory will lead the way. There will be continued recovery in auto, industrial, etc. Capex remained top-heavy despite increased spending by Chinese manufacturers. TSMC will definitely increase the capex, along with Intel, among the top 10 spenders. Semiconductor and equipment recovery is on track. There is buildout of IT infrastructure, 5nm demand ramp, 5G growth, memory capacity buildout, etc.
Trillion-dollar industry by 2036?
Dale Ford felt that the annual revenue cycle trends are up, starting from Sept. 2019. Annual revenue growth profile continued steady through 2020. It broke positive in August 2020. There are now strong demand and technology drivers. Semiconductors sit right at the top of the profile. Average lead times have also improved, especially, for controllers and processors. There was an upward trend in analog and logic components. The demand for discrete components and automotive components are also in the news.
There has been solid start to the current cycle. Most cycles last about four years. The technology/market forces are aligning to support growth in 2020+. Semiconductor industry has become much more responsive to the market indicators. We have an opportunity to see the ‘swoosh’ scenario. There are concerns about the global economy. However, electronics and semiconductors have been the biggest beneficiaries of the free trade.
The long-term semiconductor growth trends are moving toward $439-$472 billion by January 2021. It can easily move to $750 billion by 2030, and perhaps, a trillion dollars by 2036. Some positives include medical equipment, data centers, telecom infrastructure and 5G, solid-state drives, touchless solutions, memory, and sensors. The triumvirate of cloud, 5G and IoT, will make the long-term future looks very bright.
CJ Muse said there was higher OSAT pricing throughout the year. PCs grew 13 percent in 2020. It will probably be flat in 2021. Semiconductors are benefitting from being the component for the new economy. There will be 30 percent growth in DRAM and 12 percent growth in NAND. Industrial is just beginning to recover. IoT and smartphone are going to see huge growth. The party is just getting started. 2021 will be a great year, followed by 2022. There will be more supply, leading to some buffer stocks.
The world economy is depending on semiconductors, as the last year has shown. The impact going forward, will be on the supply chain. There are applications of AI for supply chain management, and key performance indicators and predictors. We need to deal better with the Black Swan events in the future. The SIA is also looking at a study on the supply chain, which will be coming out soon.
Ford added that the Covid-19 crisis needed agile and nimble response. We are dealing with an industry that deals with how long it takes to produce a chip. Automotive lines were being shut down. Demand came back stronger, than expected. They need products built on 200mm. Investments have been more on the leading-edge technologies.
For lack of a low-cost component, other things can get held up. TSMC, UMC, etc., are taking some steps, but that won’t solve matters that easily. There are challenging questions regarding chips that automotives need. Muse remarked that semiconductor contracts with automotive manufacturers are long lasting. These chips are built for the different vehicles. Chip makers may want to cut supply.
Semicon spends healthy
Andrea Lati noted that spending levels have been healthy. China has also come up strong. TSMC, Samsung, etc., are seeing sectoral trends. They anticipate greater demand ahead. We are looking at tight market conditions. We may end up looking good in 2021. That should drive more capacity increases.
Long-term growth factors for the global semiconductor industry are there. Dale Ford felt that the future is bright. There have been many products that have shaped the world. The markets won’t fail in the future for the lack of innovation and technology.
Lati felt there could be some geopolitical risks in some parts of the world. Muse added that rising prices will occur in 2021. Ford added that there can be a policy of bifurcation with China. There could be changes in supply chain, and that could be a concern.
Precision Medicine World Conference (PMWC) 2021 began in Silicon Valley, USA today. Keith Yamamoto, UCSF and Session Chair, welcomed the audience. The panel discussed how Covid-19 led to disruption of biomedical research and healthcare.
Healthcare practice, data sharing, telemedicine, clinical trial design, and enrollment adapted in real time, and opportunities emerged to establish value-based strategies that could transform 21st century healthcare through collaboration around a big-data ecosystem.
Dr. Jeffrey R. Balser, Vanderbilt University Medical Center (VUMC), said there is a need to collect patient information. REDCap (Research Electronic Data Capture) facilitates co-operative research.
REDCap is a web-based software solution and tool set that allows biomedical researchers to create secure online forms for data capture, management and analysis with minimal effort and training. The Shared Data Instrument Library (SDIL) is a relatively new component of REDCap that allows sharing of commonly used data collection instruments for immediate study use by research teams. There is also a cancer patients database. These kinds of tools are critical.
We are trying to prioritise the vaccine, which is right now in limited quantity. We can develop automated ways to pull out patients. That capability is not yet there at scale in the USA. There is also lot of mechanical stuff around telehealth. At Pfizer, people pay by the month. We are not there yet. We need pre-authorization for everyone to do telehealth business. We need to schedule people for vaccination. We need an infrastructure around telehealth that scales for the country.
Dr. Yvonne Maldonado, Stanford University School of Medicine, added that this has been a challenging time. Besides being an academic researcher, part of her role is to work on clinical response. They were able to build a proprietary FTA PPE for Covid-19. They rapidly developed clinical trials for the outpatients. We took care not to risk the exposure to the other patients. We studied, patients, trends, and risk factors. We are tracking several thousands of people around the Bay Area. We studied the population impact of this disease. We also built more community engagement.
One other aspect that needs to be conquered is: how do we find people? They have access to different modalities. We need to approach them at the community level. There are mobile phones that can be tapped into, if required. Where is also the national framework for healthcare? We need to deal with that. We have the opportunity now.
Dr. Peter Walter, UCSF, said that their labs were initially shut down. We used technology for a different purpose. The nanobody was developed. It is a simple version of the antibody. We accomplished our tasks within five months. The research needs to be continued, and carried on to the next steps. We have some information, so far. We need more clinical testing to be done. We also need to take the ball from one player to another. The distribution of nanobodies would become easier, over time. There is need for a more creative approach for the future.
Dr. Ralph Snyderman, Duke University, noted that poor people have less access to telehealth. We need to extend those. There is a tremendous need for interconnectedness. There has been a failure, and there is need for an infrastructure for the continuity for care. Developing solutions were a series of one-off. We need to bridge the last mile. They had immunized 14,000 people, but that is a small number. We also need to have implementation science. Also, to participate at a minimum, you need a smartphone. If people can come to the health center, we can look after them. Who can give every participant a smart device? We need to have the capability to get distance technology to the people. Basic science alone will not be sufficient.
Edge AI and Vision Alliance (formerly the Embedded Vision Alliance) hosted a webinar today on advancing AI processing architecture for intelligent vehicles
Horizon Robotics is a leading edge AI computing company focused on smart mobility. Its next generation platform is called Journey 3 that is shipping today.
Deon Spicer, Horizon Robotics’ Director of Sales, said the company has a global footprint, with headquarters in China. It has markets in China, Europe and United States. There were six model launches in CY2020. It has 40 design-in contracts and has 60+ ongoing projects. The AI algorithm team competed in the Waymo Open Dataset challenges and performed admirably.
The Horizon dual-core BPU (brain processing unit), the Bernoulli v2 architecture, is natively designed for deep learning. It uses MIMD and maximizes memory read/write in BPU. There is parallel computation with multiple ALUs. Horizon has been able to deliver high performance with low latency.
There is the Matrix 2 perception compute system. Each system has 4x1080P camera inputs at 30fps. It comes with a library of AI algorithms, and is used for 3D detection of vehicles and pedestrians. Functional safety is very important for Horizon and its customers. It has IS) 26262 ASIL D process certification.
Horizon offers the Journey edge AI processor, Open AI toolkit, Matrix perception computer, and co-designed and co-optimized AI algorithms for camera and LiDAR perception.
Horizon AI technology enables more features, more flexibility, and better overall performance. It uses less than 3W amd less than 60ms latency. There is L4 surround vision. With the LiDAR, and camera perception and fusion, there are 6x cameras doing semantic parsing. There is fused output, and improved perception.
The detector trained on one LiDAR/dataset can generalize to a different LiDAR/dataset. Horizon also develops multimodal solutions for smart cockpit in China. Driver monitoring, distraction alert, smoking detection and mitigations are some of the features available.
Journey 3 on-chip ISP delivers outstanding performance. Horizon is now working in the Journey 5 AI processor and move on to Journey 6 in 2023. Journey 5 will start sampling in the later half of 2021, and ship in 2022. It plans on new architecture for Journey 6, with new BPU cluster, new CPU cluster, new peripherals, etc.
There is an easy-to-use AI toolkit available for developers from Horizon. The performance compiler optimizes the performance. Horizon also has a quick start guide for customers. The key performance indicators for Horizon are accuracy, speed, and efficiency. They have MAPS or mean accuracy-guaranteed processing speed, as a new metric for fair performance evaluation.
Jean-Christophe Eloy, CEO and President, Yole Développement, presented on Innovation in Advanced Packaging is Coming from AI Processors, at the ongoing ISS 2021.
There is a classification of processor chips, such as vision processors, AI accelerators, and GPUs. There are programmable logic processors, such as configurable SoCs and FPGAs. Server market is the most growing logic market. Datacenter is also around. There are edge datacenters that do AI training, and apps are pushing the demand.
The road to augmented intelligence is here. Led by PCs, and later, smartphones, it has moved on to robotics, holographic interaction, etc. AI inference and training have different requirements. High computing is required for these. More and more models are trained with the growing number of data. Computing requirements rely on the app. You need different devices for learning and inference.
Impact of AI
The impact of AI is huge. AI in imaging is also possible. There are AI acceleration chips. The revenues for logic for datacenters is also increasing, from $18 billion in 2019 to $33 billion in 2025e.
High-end performance packaging is growing. It is defined as the front-end packaging technology. Hybrid bonding is the new wave of integration. 3D SoCs are also emerging. There are advantages of die partitioning. It consists of splitting a dye’s functions and redistributing them.
3D SoC is likely to be adopted in telecom and infrastructure servers, and mobile and consumer, for game stations and laptop PCs. Large OSATs are separated from the rest. The top 8 OSATs have continued with heavy investment in capex and R&D. In the packaging supply chain, there is need to go head-to-head or along with the big guys.
The overall advanced packaging revenue forecast for mobile and consumer is high, at CAGR of 5.5 percent. Megatrends are pushing devices, such as AI, Big Data, 5G, IoT, smart, etc. Packaging is adapting to these trends.
The IoT TechEx North America 2020 began in the USA. The event co-locates IoT TechExpo, AI & Big Data Expo, Blockchain Expo, and Cyber Security and Cloud Expo, respectively.
The opening panel discussion of the Blockchain Expo was on: The Future of enterprise technology: Predictions of 2021 and Beyond.
The participants were Duane Jacobsen, CEO, Quipu, Gari Singh, CTO, IBM Blockchain, Ms. Lisa Butters, GM, GoDirect Trade, Honeywell, and Rahul Vijay, Global Connectivity Supply Chain & Operations, Uber. Ms. Daniela Barbosa, VP World Wide Alliances at Hyperledger, was the moderator.
Ms. Daniela Barbosa said that we have the technologies. How do you build the community and get the network requirements?
Rahul Vijay, Uber, said that blockchain is changing how do we view businesses. At the end of the day, it is the trusted middleman. Crypto currency is threatening the concept of banking. There are teams at Uber thinking about how to re-invent the model going forward. This is more of a 5-year strategy.
Ms. Lisa Butters, Honeywell, noted that the benefit of being more transparent is a big benefit to the consumer. We look deeper into the ecosystem, eg., aviation. We get the airlines, component manufacturers, etc., into the ecosystem. There are many benefits for consumers. You kind of become a repair shop that is the gold standard. Its a lot to ask for being transparent with your data. You need to have trust in the system.
Duane Jacobsen, Quipu added that they are doing projects. It got into crypto currency and blockchain. We are now getting opportunities to build real blockchain products. You need to focus on your core competency. Blockchain-as-a-service is being made available at the level of cost for startups.
Gari Singh, IBM, said people are having conversations around the problems we are trying to solve using blockchain. How do we apply the technology to that? How do we drive those down to the right protocol stacks? You should have context that makes sense for the people. We need to have more use cases.
Ms. Barbosa said it is important to have technology and regulatory requirements. How do you move it to the next stage? How do you educate people?
Ms. Butters said there are big and small players. No. 1, it is important to make it easier for companies to participate. We need to build on our relationships. Nobody likes giving up their data. Honeywell executives are doing lot of education in the market. It is a lot of education about how the technology works. There is need to have distributed ledger in the network.
Gari Singh added that people want to participate in the data. There are community nodes. People should not be as worried. Lot of people think that all data goes everywhere. That’s not actually correct. There can be multiple sub-groups.
Jacobsen noted that we are seeing the need for a platform for start-ups. One client has built an IoT lock system for others to utilize. They would utilize and transact on that chain. It is a private chain. It is an exchange opportunity within a network.
Vijay said there are various things Uber is engaged in. We also do food delivery within a certain radius. If we can come up with a ledger of food not going stale, that can improve our delivery area. Blockchain can probably help. Asset tracking is also a part of blockchain. There are autonomous cars, as well. Are they going to develop a private network for such vehicles? There is some talk going on.