Thematic ● Global November 2020 5 years ROI on digital factories Smart farms: Even though farm picking can be well paid in some parts of the world, the conditions of the labour do not sit well with the work force of today and farmers often have difficulty in finding workers. Due to this many companies and start-ups are busy developing low-cost cutting edge robots to do their bidding, using soft-grippers and machine vision. We place this smart farming technology between the “backlash” and “real applications” window, as even though these robots are being made, they have slightly missed the boat so to speak to help farmers during the social distancing era during the pandemic. But we expect the pandemic to accelerate the technology and adoption over the coming few years. Companies to watch include: EEMEA: Safaricom LatAM: SLC Agricola, Be Green, Promit, InCeres, Solinftech 10s For a farm bot to pick a strawberry Moving over distance: Drones: When one thinks of drones, probably consumer drones come to mind – like the drone which caused havoc over Gatwick airport in 2018. However, enterprise drones are being used today in a number of industries including construction, mining and agriculture. They are easier to operate as they are in a controlled environment. We place these types of drones into the “real applications” phase of the disruption framework. The pandemic has made governments loosen regulations to permit drones to be used to alleviate distribution issues for medical purposes. This may lead the path towards drones for commercial use going forward. We place these type of drones in between in the early part of the “real applications” part of our framework. Companies working on drones include: Asia: Terra Drone US: Workhorse (deploying already for last-mile on trucks), Honeywell (expects drone business to be USD120bn by 203032 and recently bought UAV business from Ballard Power), Zipline (medical deliveries, next commercial), Amazon (Prime Air trials), Walmart drones, Ambarella (chip designs for drones), AeroVironment, Northrop Grunman, Lockheed Martin, Kratos ______________________________________ 32 "Honeywell launches new business unit to capture drone market", Reuters, June 2020. 44 Thematic ● Global November 2020 and oil, but passenger vehicles, trucks and datacentres of the future will decarbonise and use technologies like hydrogen and batteries. Logistics: Warehouse automation: We believe that the growth of ecommerce in the DM over the coming decade, going from 10-30% of physical commerce to up to 50% by 2030, will inspire companies to automate their warehouses further. At the moment 80% of warehouses are still manually operated, with only 5% being fully operated. Today, autonomous guided/mobile vehicles, semi-automated picking systems and automated storage/retrieval systems are the tools which are used to automate factories. There are companies which generate a total of USD15bn of revenues in global warehouse automation today. So we place warehouse automation into the “real applications” to “new normal” part of our framework. Looking forwards, we believe more manual tasks like picking will be automated with machine vision and soft grippers. For example, in the US alone, this would put at risk 2m jobs, as this task in the warehouse needs humans – 90% of this side of the process is done manually. This kind of technology is still in between the “backlash” and “real applications”. When it moves to “new normal”, there will be a lot of disruption. Companies involved in warehouse automation include: Europe: Ocado, Deutsche Post (as well as warehouse automation, they user AR to assist picking and use exoskeletons too), Magazino/FEIGEG (picking) Asia: KION (owned by Weichai Power), Alibaba, Yaskawa US: Locus Robotics, GreyOrange (picking), Amazon, Walmart 5% Of warehouses are deemed automated today Digital factories: Similar to warehouses, only 6% factories are digitised so far. So there is a lot of upside. This is less a matter of the technology or platforms not existing but more about rolling out when it makes economic sense to upgrade factories. Companies have to make the judgement of whether the ROI makes sense. We place digital factories in between the “real applications” to “new normal”. Companies to follow include: Asia: Shenzhen Innovance, Fanuc (Japan), Cognex (US), Airtec (Taiwan), Hiwin (Taiwan), EEKA Fashion Europe: ABB, Seimens, VW (announced 2,200 robots for a digital EV factory recently. 1.400 from Fanuc and 800 from ABB)31 LatAm: WEG US: Honeywell, Rockwell ______________________________________ 31 "2,200 robots for automation at VW plants", Electrive, October 2020. 43 Thematic ● Global November 2020 63bn By 2025 for drone services worldwide AVs: A lot of investment and talent has gone into the promise of passenger vehicles but we believe that it makes more sense for commercial applications like trucks to become AVs faster than consumer vehicles. This is due to AVs in commercial applications making meaningful economic savings in lowering cost of operation from using more automation and the fact that they don’t need to operate in urban areas but instead can operate over more long, uncomplicated commercial routes. So we would place passenger AVs on the “hype mania” to “backlash” window and commercial AV trucks in the very early part “real applications” phase. Companies developing AVs road transport include: Asia: Weichai power, Zoomlion EEMEA: Yandex, AVL Research & Engineering (AVs), Trukker (AVs/AV trucks) LatAM: HiTech, iara (AVs) US: Otto, Nikola (AV trucks), Waymo, Uber, Cruise, (ride-hail), Amazon’s Zoox (lastmile). Tesla We have seen investments in mining AV trucks by: Newmont, BHP, Fortescue Metals Group, Rio Tinto, and Hancock Prospecting in Australia; Suncor Energy and Canadian Natural Resources in Canada. 80% of surface mining autonomous haul fleet belong to Rio Tinto, BHP and Fortescue. Suppliers include: Komatsu and Caterpillar. 91% Driver cost reduced from AV trucks 45 Thematic ● Global November 2020 Automation infrastructure Automation infrastructure will enable efficiencies within a variety of industries over the coming few years and beyond We look at key trends enabling growth in automation technologies through logistics, moving things over distance and 3D printing subthemes within our disruption framework and list over 60 companies exposed to the themes We outline 16 1st tier, 2nd tier, and ESG implications from our themes Spinning Jenny to 21st century AI bots powered by hydrogen Automation technologies are set to accelerate to take physical human labour fully out of the economy As a species, humans have been using technology to improve our productivity for eons. For a long time, most productivity tools were powered physically by people, for example using a lever to amplify power or striking flint with hard minerals to create fire. However, over the last two centuries we developed technologies like the Spinning Jenny (1764) for production of wool or cotton, separate condenser for James Watt’s steam engine (1781) and steam locomotive (early 1800s) powering our railways during the industrial revolution. It can be argued that WW2 then instigated computer technology to increase the scale of automation further. Gradient of estimated expectation vs. reality Chart 1B. HSBC Disruption Framework: Automation infrastructure Moving over distance Logistics autonomous vehicles, Picking, packing, factory, smart-farm, drones, hydrogen, remote sensing, artificial intelligence, battery electric machine learning 3D printing Production supply-chain, food, clothes, printed circuit boards Early disruption Hype mania Backlash window Real application New normal Source: HSBC Today, modern automation is driven by a combination of robotics, AI and sensors for our warehouses and digital factories. Looking forwards, to where the next investments are taking place, if they pan out, we will likely see growth in the use of drones, autonomous vehicles and 3D printing within industry. Automation so far has been powered by fossil fuels, like coal, gas 42 Thematic ● Global November 2020 Story and key trends for automation infrastructure 1. The logistics story: rise of automating distribution robots The efficiencies we gain from our associates and robots working together harmoniously – allows us to pass along lower cost to our customer Tye Brady, Amazon Robotics’ chief technologist Logistics automation needs disruption to cope with demand Growth of remote commerce set to grow logistics automation Before the pandemic, DM physical commerce was between 10-30% online, estimates James Pomeroy, our Global Economist. However, we believe that the pandemic is accelerating consumer behaviour to rely more on online consumption. Pomeroy now estimates that up to 50% of goods consumed could originate online by 2030, in the DM. To achieve this number, we must have sufficient automated logistics infrastructure to cope with this new demand. Chart 2B shows that global warehouse automation is worth USD15bn today and is set to grow to USD27 by 2025, according to IFR. Chart 3B. Most robotics is in logistics Chart 2B. Global warehouse automation and robot installation growth 20 24 600 20 500 400 10 300 200 5 100 0 2012 2013 2014 2015 2016 2017 2018 2019e 2020e 2021e 0 Global Warehouse Axis Title Automation market Annual Installation of industrial robots Source: HSBC, IFR, RoboBusiness 000 units USDbn 15 700 16 12 8 4 0 2017 '18 '19 '20 Logistics Medical '21e '22e Field Defence Source: HSBC, IFR Although Amazon’s fulfilment centres have over 200,000 robots working its warehouses today (up from 30,000 in 2015)33, it’s an outlier with 80% of warehouses globally still operated manually, with no supporting automation. 15% of warehouses are so called “mechanized”, meaning they use conveyers, sorters, goods-to-picker solutions but are not necessarily considered automated. Only 5% of warehouses today can be deemed automated. So there is significant room for growth and disruption, and given the prevailing winds of online consumption, automation is especially important for such high volume and low margin businesses to make economic sense. What robots do automated warehousing need? The online shopping model requires 300% more warehousing space than store-based fulfilment.34 Euromonitor say that e-commerce implies there will be a need of 2.3bn square feet ______________________________________ 33 "As robots take over warehousing, workers pushed to adapt", AP News, December 2020. 34 ROBO Business 47 Thematic ● Global November 2020 Chart 5B. Automated robots in warehouses Chart 6B. Agri-bot system environments today 12 10 1% ($ bn) 8 6.8 Polytunnels 5.2 6 12% 4 2.4 2 0 0.3 0.7 0.6 0.8 1 0.9 1.6 1.3 1.8 2.6 3.2 3.8 Open field 4.6 Greenhouse 49% 2017 2018 2019 2020 2021 2022 2023 2024 2025 Autonomous Mobile Robots (AMRs) Automated Guided Vehicles (AGVs) Source: HSBC, Logistics IQ Vineyard Orchard 3.1 4 11% 27% Source: Sensors Smart farming We have to address population growth, climate change and labour issues, and that has brought a lot of interest to technology Prof. Slaughter, (Biological and environmental engineering at UC Davis) Picking automation How accurate are picking robots and where are they used today? UK farmers reported shortages of fruit and vegetable pickers during beginning of the pandemic. At that point they needed to fill 90,000 jobs. Even before COVID-19, pickers were earning USD50 per hour in California for example, but farmers still struggled to recruit. One option UK farmers were considering was to ship workers from other nations but this can be a political hot potato in today’s environment. A tempting alternative that’s become more of a reality today, could be automation using robots with 3D vision and soft grippers. A number of companies like Octinion (Belgium start-up), Dogtooth (UK), Agrobot (Spanish) and Harvest CROO (US) are creating robots to pick fields. Chart 7B shows us the accuracy of smart farming with robotics today; for example, note that citrus fruit can be picked in about 2.5 seconds but the accuracy of the pick is only 50%. On the other hand, some types of strawberries take 10 seconds to pick and the accuracy rate is 100%. So the hit rate varies from food item to item in smart farming technologies today, but with hardware like Lidar, ultrasonics, hyperspectral/thermal cameras and other 3D sensing devices getting cheaper, and AI/ML (artificial intelligence/machine learning) improving, the hit rate will improve over time and may compete against human labour. In addition to picking on farms, smart farming technology is also automating other activities: disease detection, seeding, spraying, crop scouting, weeding, harvesting (picking) and general plant management. Chart 6B shows us the various agricultural environments in which robotic systems are deployed; 49% is within open fields (unstructured areas) and 27% in greenhouses. 49 Thematic ● Global November 2020 3D printing: Additive manufacturing (AM): Mass printing at the consumer level didn’t take off in the end and we place that firmly in the “backlash” part of our framework. Within industry, AM has been used for rapid prototyping for years now. But for a number of reasons, AM may be moving within industry to shorten the supply chain, re-shore to potentially make it more resilient. We place this into the “real applications” phase even though the industry is worth USD10bn today. Furthermore, 3D printing also exists for things like making organs, food, homes, PCBs and more but this is in the “early disruption” space, meaning it’s still pretty much in the venture capital, angel investing and R&D phase of our disruption framework. 3D printing companies include: Companies involved with 3D printing: 3D Systems, Proto Labs, Faro Technologies, Materialise, ExOne 10bn/12trn The size of the 3D printing market vs total manufacturing sector 46 Thematic ● Global November 2020 of additional warehouse space needed by 2035. Chart 3B shows that the logistics robotics market is larger than the robotics for medical, field and defence spaces combined today. Chart 4B shows selected companies making logistics automation robots and below are the different types of logistic robotics and automation which will fill those warehouses going forward35: Autonomous guided/mobile vehicles (AGV, AMV) is set to be 14.7% of global warehouse automation by 2025. AGV/AMV has doubled in last 3 years and will have 35% CAGR till 2025. See chart 5B to see that AGV/AMV market is just over USD1bn in 2019 and set to balloon to over USD11bn by 2025, a tenfold growth. Picking systems is set to be 12.3% of the market by 2025, and is the toughest technologically speaking, to automate fully. But if retail wants to become more cost efficient in the longer term, reducing or eliminating labour is key in this process. Today, it’s mostly about bringing the item to the picker, to minimise picker travel time, rather than the robot doing the picking. Automated storage/retrieval systems (AS/RS) will be 12.3% of the market by 2025. It includes automated racking, shelving and shuttling systems, permitting denser warehouse storage. Table 4B. Warehouse Automation Robot Players Company Country Solution Function Amazon U.S Kiva Automated guided vehicle (AGV) Fetch Robotics U.S HMIShelf AGV GreyOrange U.S Butler Picking HAHN Group/Rethink Robotics Germany Baxter/Sawyer Packaging/assembling Hitachi Japan Racrew Picking IAM Robotics U.S Swift Picking/AGV Kollmorgen Germany Pick-n-go Picking Locus Robotics US LocusBots Picking/AGV Magazino/FIEGE Logistik Germany TORU Picking Scallog System France - Picking Swisslog Switzerland CarryPick, AVGPick, Autostore Picking and automated storage/retrieval (AS/AR) system 6 River Systems US AGV Chuck Sources: HSBC, LogisticsIQ What about automated digital factories? Just to contrast logistic robotics to industrial companies with factories for production, only 6% of factories are fully digitised as of yet according to a PwC survey, comparable to fully automated warehouses, which is 5% of the market so far. They also contrast this in the survey where 91% of industrial companies are investing into implementing digital factories. The key reason for digitising factories is that they bring efficiency in production (on average 12% efficiencies), with most expecting ROI over 5 years. In addition, digital factories are likely to use the following technologies in the next 5 years, in the process of deploying automation technologies: connected sensors (39% in use today, 64% of respondents to use in 5 years’ time, according to a PwC survey), 3D printing (18% today, 37% in 5 years), AR/AR (13% today, 33% in 5 years), humanoid robots (12% today, 22% in 5 years), AI (9% today, 20% in 5 years) and drones (2% today, 4% in 5 years). ______________________________________ 35 LogisticsIQ 48 Thematic ● Global November 2020 Chart 7B. How good is robot farming today? Picking rate (%) 100% 80% 60% 40% 20% 0% 0 10 Apple Mushroom 20 Strawberry 30 (sec) Harvesting time Tomato Cherry Tomato 40 50 Cucumber Melon 60 Citrus Source: Sensors 2. The moving over distance story: autonomous and fuels We wanted flying cars, instead we got 140 characters Peter Thiel, Venture Capitalist Pandemic accelerating drones for medical purposes Is it a bird, a plane… no it’s a drone! For some years now, drones have been much vaulted as having disruptive transport potential for small items, through the air, for B2C. However, legacy regulations in the developed world have limited the scope and drones have seen more success in the EM, due to some of these nations accepting the leapfrog potential of the technology. Nevertheless, the pandemic has enabled drone companies to get waivers from governments all over the world, to deploy them. Perhaps this might also increase social acceptance of the use of drones, too, going forward. Below are some examples of what drone companies had been up to during the pandemic36: United States: Zipline got permission from the FAA to deliver medical supplies and PPE to medical centres in North Carolina during the pandemic. Zipline had been using drones in Africa for a number of years, delivering to hospital settings. In Virginia, Alphabet’s Wing drone project delivered pharmacy and take-out orders within 10mins, during the crisis, alleviating last-mile delivery infrastructure for businesses. China: Antwork (part of Japanese Terra Drone) had a licence from Civil Aviation Administration of China (CAAC) and used drones to deliver medical supplies and test samples in Zhejiang province (February 2020). They halved transport time compared to road transport. ______________________________________ 36 "COVID-19 transport brief", International Transport Forum OECD, June 2020 50 Thematic ● Global November 2020 Chart 8B. Drone market North America Europe MEA Total 50 40 Chart 9B. Drone industry growth (2019) South America Oceania Asia 43 300% 200% 22 20 28 10 33 37 100% 40 2022e 2023e 2024e Source: HSBC, Droneii estimates 2025e Real Estate 2021e Surveying 2020e Agriculture 0 Mining 0% Construction 30 Source:DroneDeploy Ireland: Manna Aera got clearance from the aviation authority to deliver medication and supplies to households in the rural town of Moneygall, with a payload of 4kg, after VC between patient and doctors (April 2020). They did 100 flights a day and look to expand to more towns in Ireland and the UK. Could expand into non-medical items like groceries. Ghana: Zipline, the US company, used drones from April 2020 to deliver contactless test samples from 1,000 rural health facilities, to labs in Accra and Kumasi. Drone transport avoided difficult routes by road, and time was reduced from several hours to one hour in some instances. Switzerland: Matternet had permission to fly beyond LOS (line of sight) since 2017, transporting blood samples in hospitals in Lugano, for 20km with 2kg payload. But Switzerland did not use drones during the pandemic. In addition to using drones during the pandemic for delivering medical related products, a number of countries have been using drones for surveillance and enforcement. China, France India, Italy Oman, Colombia and the US are some of these countries – using drones to enforce lockdowns and identify citizens with high body temperature (though, in May 2020, French courts banned drones with cameras due to privacy concerns). Furthermore, some nations used drones to sanitise spaces too: China (900km^2 in 20 provinces, 2,600 drones), Korea (10,000m^2 in 10 mins), India (Delhi, Indore City, Madhya Pradesh) and US (sports arenas and concert venues). B2B drone use is growing… Whilst drone use for commercial last mile deliveries (like Amazon) is what bubbles up when one thinks of these flying contraptions, and they may be easier to deploy in the future for consumers due to these pandemic induced accelerated trials, various industries are already using drones today for B2B. For example, Terra Drone Corporation worked with Plimsoll UAV Brazil to service oil & gas customers, Sky-Futures worked with Bureau Veritas for doing industrial inspection services in Europe and Identified Technologies used drones for haul road and slope analysis in the mining and infrastructure sector. Chart 9B shows the growth rates of various industries utilising drones in 2019. Chart 9A shows that the drone market was USD22bn in 2020 and is expected to grow to USD43bn by 2025, with the biggest growth regions to be Asia and North America, according to Droneii. On the other hand, the drone services market was USD4.4bn in 2019 and is expected to grow to USD63.6bn by 2025 (55% CAGR)37, becoming larger than the physical drone market itself, again similar to IoT discussions earlier when we highlighted platform services can generate more sales than hardware. ______________________________________ 37 Droneii 51 Thematic ● Global November 2020 The AV ready countries… Autonomous road transportation Aside from having the right technology developed for autonomous vehicles (which we will look at shortly), the key for deploying autonomous vehicles for trials (like drones we spoke of earlier) is to have the right regulatory environment. Chart 10B shows us a list of 30 countries ranked by how ready they are for deploying autonomous vehicles, the autonomous vehicles readiness index, taking account of various factors including: quality of roads, technology infrastructure, cybersecurity, population living near test areas, AV related patents, industry partnerships, government funded pilots and more. The top 10 include: Singapore (#1) and South Korea (#7) from Asia, UAE (#8) from the Middle East, United States (#4) from the Americas, Netherlands (#2), Norway (#3), Finland (#5), Sweden (#6), UK (#9) and Denmark (#10) from Western Europe. Chart 11B also shows us the list of countries that have supportive policies and legislation for AVs. Note that top is Singapore, with UK in second place. Table 10B. Overall AV readiness by country #1 Singapore #2 The Netherlands #3 Norway #4 United States #5 Finland #6 Sweden #7 South Korea #8 United Arab Emirates #9 United Kingdom #10 Denmark #11 Japan #12 Canada #13 Taiwan #14 Germany #15 Australia #16 Israel #17 New Zealand #18 Austria #19 France #20 China #21 Belgium #22 Spain #23 Czech Republic #24 Italy #25 Hungary #26 Russia #27 Chile #28 Mexico #29 India #30 Brazil Source: HSBC, KPMG Chart 11B. Most AV friendly policy wise Brazil Mexico India Italy Russia Hungary Chile Spain Czech Republic China Belgium Israel Japan Australia South Korea Sweden France Taiwan Denmark Austria Norway Canada United Arab Emirates Germany United States New Zealand Finland The Netherlands United Kingdom Singapore 1.53 2.00 2.16 3.78 3.82 3.93 3.96 Non-AI software 4% 4.65 4.89 5.07 5.17 5.27 Physical Systems 4% HD Mapping 4% 5.78 6.02 6.14 6.22 6.24 6.33 6.34 6.37 6.64 6.74 6.79 6.81 7.02 7.08 7.10 7.17 7.23 AI 5% Source: HSBC, IBM 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 Other systems (steering, chassis components, etc.) 3% Ridesharing App 1% Microchips (GPU's, ECU's) 49% Auto Electronics 10% Sensors 20% 7.82 0.00 Investments for AV technology Chart 12B. Where AV investments are going (2016-2022) 9.00 Source: Frost & Sullivan According to Frost & Sullivan, 2016 was the year when investments started properly into AV technology. They estimate that between 2017 and 2023, nearly 50% of investments deployed will be into microchips (GPU, ECU) for AVs, 29% in sensors and 10% into auto electronics. Surprisingly AI software is only estimated to take 5% of the investment and non-AI software another 4%, so in total software will be about 9%, similar to the total amount for auto electronics. See chart 12B. Chart 13B shows us the hardware and software combinations of both OEMs and disruptors. Most of them appear to be developing the software in-house, with partnerships for hardware sensors and chips. Moreover, chart 15B gives us an overview of the various sensors, their vision reach, approximate costs and data transfer rates. 52 Thematic ● Global November 2020 What about AV commercial trucks? Although much of the public buzz for AVs over the years has been about passenger vehicles, the actual implementation is proving to be trickier than expected for a number of reasons and current COVID-19 may make things more difficult from an economic point of view (selling expensive new AV cars and possible pressure on ride-hailed AVs due to shared spaces). Table 13B. AV hardware and software partnerships OEM/Disruptor H/W Apple N/A S/W In house BMW Mobileye, Nvidia, Intel IBM Watson Daimler Qualcomm, Mobileye In-house Ford Velodyne, Nvidia In-house GM Mobileye, Delphi Cruise Automation Tesla Nvidia In-house In-house Uber Velodyne In-house VW Mobileye, Nvidia, LG In-house Volvo Velodyne In-house Waymo In-house, FCA, Lyft In-house Source: HSBC, Frost & Sullivan However, we believe that the B2B AV market may prove to make more sense, especially in the commercial truck and heavy-goods vehicles (HGVs) markets, in particular as there is due to be a refreshing of HGVs and other trucks over the coming few years, with new greener fuel (battery electric and hydrogen) based on electric drive-trains, as we have written extensively within HSBC Global Research. These trucks will be driven on pre-determined commercial highway routes, away from dense human populated city streets, making it safer from a practical point of view. Table 14B. Freight truck savings Driver costs Fuel Emissions Safety Net new traffic _________________ 2025 __________________ _________________ 2040 __________________ Freight Passenger Freight Passenger 91.1% 77.3% 87.3% 73.8% 1.8% 10.8% 3.1% 9.2% 1.2% 4.3% 2.2% 3.1% 3.5% 3.5% 3.9% 9.8% 2.4% 4.1% 3.5% 4.1% Source: Linkoping University Chart 14B compares passenger and freight AVs and the expected gains from both types of vehicles from automation. For example, in 2025 it’s expected that in terms of driver costs from automation, fright trucks will save 91% of driver costs against 77% for passenger vehicles, according to Linkoping University. This is a significant savings in terms of hiring drivers, for businesses. Chart 16B shows us the average cost per ton-mile of transport by load type. Air freight is the costliest at USD125, truck cargo sits at USD18 and rail at USD4. 53 Thematic ● Global November 2020 Table 15B. Sensor specs for AVs (diagram of sensor range) LIDAR Measurement distance (m): Up to 250 Cost ($): 1,000-75,000 Data rate (Mbps): 20-100 RADAR Measurement distance (m): 0.2-300 Cost ($): 30-400 Data rate (Mbps): 0.1-15 Camera Measurement distance (m): 0-250 Cost ($): 4–200 Data rate (Mbps): 500-3500 Ultrasound Measurement distance (m): 0.02-10 Cost ($): 30-400 Data rate (Mbps): < 0.01 Source: HSBC, Frost & Sullivan So truck is four times more costly than rail to transport goods. Though is mostly because of labour costs, which is about 40%+ of per ton-mile for trucks. Also one must account that trucks can be direct A to B, at any time (convenience) and air freight has more complex pathways for getting to the end destination, likely involving road too, thus adding more time to the journey and so forth. Chart 17B shows us that the testing phase for AV trucks is supposed to be from now till 2022, with more automated truck AVs on the road from 2023 onwards including tele-operation. Source: HSBC, Umlaut 54 Domestic Waterways Level 4 Source: HSBC, Idtechx Level 5 2037e Rail Cargo 2036e Truck 3 2035e Air freight 4 2034e 0 2032e 18 20 2029e 40 2028e 60 2025e Labour costs is approx 40% of truck 80 2024e USD 100 L5 penetration accelerates L4 automation penetrates and peaks L4 automation in semicommerical trials 2023e 120 50% 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% 2022e 125 2021e 140 Chart 17B. Expected timeline for AV trucks 2020e Chart 16B. Avg cost per ton for transport (2019) Thematic ● Global November 2020 3. The 3D printing story: production on demand Tea. Earl Grey. Hot Captain Pickard, Star Trek: The Next Generation In the early 2010s, consumer 3D printing made a lot of headlines due to key patents expiring, and encouraged a swathe of B2C companies to create 3D printers for consumer market. However, this wave of hype didn’t materialise to everyone having a 3D printer in their home, like everyone has a computer in their homes. Chart 18B. Global 3D printing revenues 10,000 9,000 Global 3DP revenues (in million $) 3D printing for consumers failed but not for industrial applications 3D printing – the real life equivalent of the replicator from Star Trek We don’t quite have Star Trek’s Replicator device from the science fiction television series, where people can order at will anything from this device (like tea), and it will create. However, back in the real world, the first 3D printer patent was issued in the 70s, and today 3D printers or additive manufacturing (AM) exist for a number of fields, in various stages of technological progress and application. 3D printing or AM “prints” layer by layer from a 3D model and can create structures and shapes that would be very difficult or even impossible using traditional “reductive” methods of manufacturing. 8,000 7,000 6,000 5,000 4,000 3,000 2,000 1,000 0 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 Products Services Source: Wohlers Today, AM manufacturing is making headlines again, especially industrial uses of 3D printing, in part due to the pandemic highlighting that some of our supply chains are not as resilient as we assumed previously. In Italy during the height of the first wave of the pandemic, a local company 3D printed components for ventilators for EUR1 each, when the global supply chain was disrupted, helping to save lives. We expect 3D printing to go beyond the typical use of rapid prototyping and become more embedded in localisation of supply chains. 55 Thematic ● Global November 2020 Chart 19B. 3D printers by country of use (2019) Russia 1.5% Chart 20B. 3D printing market share by type (2019) Research Post- Institutions processing 5.2% 4.8% Other 13.9% US 35.3% Taiwan 1.5% Metal materials 10.8% China 10.6% Turkey 1.3% Italy 3.3% UK 4.0% Canada 1.8% Sweden 1.1% Germany France 8.3% 3.1% Source: Wohlers Ceramic machines 3.9% In 2019, it was estimated that the AM sector was worth nearly USD10bn but this pales in comparison to the US12trn of the whole manufacturing sector.38 Today 35% of AM printers see their home in the US, followed by 10% in China and then 8% in Germany – chart 19B. The 3D printing market ecosystem is composed of hardware manufacturers (metals, polymers, desktop, composites, ceramic and electronics), software vendors (design and simulation, slicer and data preparation, workflow, security), materials suppliers (metals, polymers, composites) and post processing manufacturers. From an investing point of view, it is interesting to see that hardware makes up 56% of the AM market, 23% by software, 20% by materials and 5% by post processing. Chart 20B shows that metal printers are the most popular product in the AM market, at almost 23%. Bracket Seat buckle Bionic bracket Resource production Manufacturing Engine cover Door hinge Fork fitting Distribution I Distribution II Source:Procedia Manufacturing 0 0% Oil and Gas 0 5% Shipping 0.75 10% Trucking 750 15% Aviation 1.50 20% Residential Heat 1500 25% Power Generation 2.25 1.6% 1.4% 1.2% 1.0% 0.8% 0.6% 0.4% 0.2% 0.0% Industrial Heat 2250 Chart 22B. Additive manufacturing can reduce CO2 within industry (Global CO2 savings %, CO2 savings within category %) Passenger Cars Energy Consumption (MJ) CM energy intensity AM energy intensity Energy Intensity (GJkg)) 3.00 3000 Source:Thurun Energy The high degree of freedom in design, means components can have significant reduction in weight and generally means that one can reduce the amount of energy in the life cycle of the product made. For example, using traditional manufacturing machines an aircraft bracket, studies have shown that over 200,000MJ is used but using EBM (electron beam melting technique), a AM process, under 80,000MJ was used. There is not only a reduction of weight of the final product (1.09kg vs 0.38kg), through using less ingot (8.72kg vs 0.57kg) and raw ______________________________________ 38 AFMG 56 Metal machines 22.5% Source: Wohlers Chart 21B. Additive manufacturing uses less energy than conventional machining Energy and CO2 savings from AM MES and Workflow 2.2% Slicer and Data 2.6% Design & Stimulation 7.4% Composite Electronics machines machines 4.3% 2.2% Manufacturing Quick AM market overview Desktop machines 8.7% Polymer materials 9.5% Korea 3.9% Japan 9.2% Spain 1.2% Polymer machines 14.7% Thematic ● Global November 2020 material (8,000MJ vs 500MJ) but also less energy used for the manufacturing (900MJ), transport (40MJ vs 14MJ) and use phase (217MJ vs 76MJ). The usage of AM, it has been shown, can save 7-173 million GJ/year by 2050 for say, lightweight metallic aircraft components. Chart 21B shows us the significant savings in energy and energy intensity during the AM manufacturing process for aircraft: brackets, buckle belts, bionic bracket, and engine cover door hinge and for fork fitting. Chart 22B shows us that this kind of energy savings in the life-cycle of components can translate into up to 25% of CO2 savings within certain industries, manufacturing gaining the most. It also shows that nearly 7% of global CO2 can be reduced by using an AM process. Future of 3D printing… 3D printing in industry is clearly in the real applications stage of our disruption framework but there are more spaces where 3D printing is in earlier stages of development. For example, 3D printing is being used to print shoes by Adidas39 today and exploration of making human organs40, PCBs41, food42, homes43 and more, layer by layer. Thematic 1st, 2nd tier and ESG implications for automation Replacing pickers in warehouses has an impact on jobs Logistics ESG: Robotic arms in warehouses impact DM today, EM tomorrow: Warehouse automation is set to grow over the coming decade. The global logistics industry is valued at USD5trn, with US companies employing 2m workers to carry out stock and order fulfilment. 90% of warehouse picking is currently done by hand. Automated picking will give gains of 2-3x productivity (vs pick conveyer operations), and 5-6x productivity increase when compared to manual pick-to-pallet fulfilment centres. As consumption of physical goods grows to 50% of the market by 2030 in the DM, in order for the economics to make sense from a business point of view, we believe that the picking part of the warehouse will be increasingly vulnerable to automation. This physical task which was formerly in the domain of the human hands (quite literally), poses a threat to 2m jobs in the US, and countless more in warehouses in the rest of the DM. These automated pickers will increasingly use computer vision technologies in conjunction with soft-grippers. If the cost of these technologies fall, sooner rather than later, similar issues could arise in the EM (as their use grows), especially in regions where wages are rising. The question is what percentage (if any) of the displaced workers can re-train for different jobs. ESG: Automated warehouses are more resilient to labour issues but are vulnerable to cyber-attacks: Currently only 5% of warehouses globally are fully automated. These are the warehouses which would be most resilient to labour related shocks (e.g. pandemic or otherwise) due to a prudent reduction of humans in indoor space, allowing social distancing of smaller number of workers. This resiliency will be helpful should there be any future flare ups from pandemics or employment issues, like rising wages. However, full automation also comes with possibility of cybercriminals targeting these fully automated warehouses to disrupt the company and distribution of goods in the economy. Warehouse automation could be deflationary for online goods: Companies which deploy more automation logistics, by removing human labour costs, naturally create the possibility of passing on the savings to the consumer. Naturally this is deflationary, if it breaks the online conundrum, making low margin products like groceries profitable to sell online. Warehouse automation has labour advantages but is vulnerable to cyber-attacks Automation to make goods cheaper ______________________________________ 39 "Adidas to release a new version of 3D printed show, Alphaedge 4D!, 3D Printing Industry, May 2019. 40 "3D printing offers hope of building human organs from scratch", FT, December 2019. 41 Hensoldt and Nano Dimension create 10-layer 3D printed circuit board", 3D Natives, May 2020. 42 "Meat grows in space with 3D printer on ISS", Business Insider, October 2019. 43 "Scientists create 3D-printed buildings from soil", The Guardian, August 2020. 57 Thematic ● Global November 2020 ESG: Automation and alternative incomes: Automation of physical labour as in warehouses and digital factories will naturally bring to the forefront further discussions about alternative income models like UBI or robot dividends. As we outlined before, if picking alone is automated in warehouses, then that is a threat to 2m jobs. Many other tasks in digital factories can also be automated over time. Automation creates higher paid jobs: Whereas physical automation in industries like logistics will reduce previously un-automatable jobs due to advances in machine learning, computer vision and robotics – this next generation of automation is likely to create some jobs too. For example, this kind of automation will require the hiring of knowledge economy workers who are highly qualified and innovative, with expertise in software and hardware, and computer systems engineers who are on the cutting edge of technology. These will naturally be higher paid roles too. They will also likely be kept on the books to continue to maintain the equipment over time and improve it also. ESG: More urban farming and organic farming tech to feed the world: As populations grow and there is likely to be more demand for greater variety of foods from the EM. Moreover, climate change, shrinking farmland, water security, and overfishing are likely to place pressure on global nutrition supply and create biodiversity issues. Add to this, people are living longer and it’s expected that by 2050 the global population will hit 10bn and each person will consume 12% more than they did at the turn of the century, with total food consumption increasing by 70% by the midpoint of the century, according to the UN. To generate an additional 7,400tn calories by then would in theory require twice the landmass of the size of India.44 Note that there are 570m farms globally, 90% run by families or relying on family labour, producing 80% of food worldwide. UBI or robot dividends for lost warehouse jobs? New jobs created to automate Intensive farming helped with smart farming To meet the task of supplying the increasing demand for calories, global trade and technology will play an important role in keeping up yields to demand. Part of this will be making farming more efficient. So we expect global food supply chains to deploy more smart farming tools to help support yields. But using more land for food is complicated, placing pressure on forests, peatlands, wild areas and thus releasing carbon stored within them, adding to climate change. This also impacts biodiversity. Could organic (2% UK food sales only and 5.5% in the US) farming be preferred in some places over intensive farming techniques, but assisted with smart farming tools? Urban farming (makes a fifth of the world’s food today) technologies could also be part of the solution to make use of spaces in cities, which may be cleared up as consequences of WFH trends going forward.45 Green power for drones Moving over distance ESG: Can hydrogen eventually power drones safely? As a market and society, we have already discussed and funded the pipeline of making passenger vehicles green, meaning battery electric powered. The next transport category that is currently the hot topic is decarbonising HGVs, through either hydrogen or battery electric. However, what is not widely discussed yet is what will power drones, and this may become an important discussion point if drones are going to become a USD43bn market by 2025. There are a few options on the horizon for drones, including gas, battery and hydrogen. Gas will be able to get the drone further, in terms of distance, but isn’t green, so should probably be discouraged from the outset. Battery electric power drones are the best bet in the medium term. However, there is also the potential of (safe) hydrogen-powered drones looking out into the future a little. A team called H2Go Power based in Imperial College London are busy developing the tech (stable hydrogen without compression) and this is a ______________________________________ 44 World Resources Institute, 2018. 45 "Can we ditch intensive farming - and still feed the world?", The Guardian, January 2020. 58 Thematic ● Global November 2020 space to watch, as a hydrogen-based economy continues to be a popular theme for investors. H2Go are partnering with Ballard Power to see if they can deploy the technology for drones.46 Cheap drone delivery Drones could make deliveries cheaper: Drones may help lower the cost of local deliveries, if the technology is scaled and collected from fulfilment centres in close proximity to metropolitan areas. For example, in the US, Amazon operate most of their fulfilment centres within 5 minutes of big cities. A study looking at the economics of truck vs drone delivery for Amazon estimated that it costs on average USD1.20 for truck delivery vs USD0.36 for drones.47 Another study looked at drone vs e-van delivery costs for biomedical samples, food deliveries and parcel deliveries. It found that transport of biomedical samples could be cheaper for drones (EUR1.92 within 15 minutes) vs e-van (EUR4.59 within 42 minutes) in Rouen, France. For food delivery it estimated drones (EUR0.89 within 6 minutes) vs e-bike (EUR1.90 within 13 minutes) in London. For parcels it estimated drones (EUR0.59 within 7 minutes) and e-van (EUR0.14 but within 352 minutes).48 Trucks to go AV soon as they get new fuels too? ESG: AV passenger vehicles vs AV trucks: Since about 2016, the investments in AV technology have been increasing, as we highlighted previously. On top of this, there has been a lot of hype about passenger vehicles in the media about the development of disrupters, such as Waymo, Uber and Tesla. However, the reality is that today we don’t necessarily have level 4 autonomy passenger vehicles on the roads. The reason for this is partly technological, meaning it’s fairly difficult to navigate AVs around densely population regions safely. The other issue is similar to the drone problem of getting the right regulatory support, with high profile outlier AV accidents (note Uber’s 2018 crash with their “safety driver” now being charged with negligent homicide this year49) bringing negative media attention and headlines. The high AV accident occurred with Uber in 2018, where the back-up driver was recently charged with negligent homicide for the death of a pedestrian in Arizona. The safety driver was streaming a TV episode during the time of the accident. It’s important to note that Uber didn’t face criminal charges but it meant that they ended the testing of the technology in Arizona at the time.50 Although we are not ruling out passenger AVs rolling out over the coming few years like Tesla FSD (full self-driving) or through ride-hailing services (Waymo started fully AV trips in Arizona in October 2020) or supporting last-mile commerce services like with Amazon’s Zoox purchase, we believe from an investor point of view, commercial truck AV might the angle which becomes a reality somewhat more easily with scale. AV trucks would reduce the cost of drivers, have pre-determined commercial routes on highways, be away from dense urban areas and be entirely a B2B model. Moreover, with HGVs now on the global decarbonisation agenda, with hydrogen/battery electric fuel input, new trucks can be deployed with AV technology. Other commercial AV applications can also be attractive to companies, to improve safety, costs and efficiency – like within the mining industry. 500 AV trucks currently operate at surface mines around the world with 80% of them in Australia. It’s expected that this global number will triple by 2023, according to GlobalData’s Mining Intelligence Center.51 ______________________________________ 46 "H2Go Power seeks to power drones with 'happy gas'", BBC, January 2020. 47 "A cost analysis of Amazon Prime Air (drone delivery", Journal for Economic Educators, 2016. 48 "Comparing the cost-effectiveness of drones v ground vehicles for medical, food and parcel deliveries", Unmanned Airspace, November 2020. 49 "Uber's "safety driver" during deadly self-driving car crash charged with negligent homicide", Futurism, September 2020. 50 "Uber's self-driving operator charged over fatal crash", BBC, September 2020. 51 "Australia dominate global autonomous haul truck use with numbers set to triple", Mining Technology, March 2020. 59 Thematic ● Global November 2020 Automating trucks can put pressure on the supporting economy ESG: Negative impact on jobs from automating trucks: The push for automating trucks is not only for technological awe but there is a practical reason to do this also. For example, in the United States there is currently a shortage of truck drivers, which is driving driver wage inflation. In 2018, it was reported that there was a gap of about 51k drivers to meet the demand from companies like Amazon and Walmart. If this problem is not solved, then it could mean deliveries getting delayed and the price of goods going up. The reason it’s a struggle to hire truck drivers is that it’s said that the lifestyle is tough, meaning a truck driver does not see family often, is more likely to gain weight gain from sitting all day and lack of healthy food on the road, and so forth. Going forwards will the digital natives want such time consuming, anti-social and potentially repetitive work, even with a median pay of USD42k in the industry, which doesn’t require more than a high school education.52 According to the BLS there are more than 3m truck driving-related jobs in the US alone. Furthermore, there are the usual support infrastructures that truck drivers will utilise, especially long-haul HGVs, at cafes, motels and other rest stops – this part of the support economy could be negatively impacted too. An interesting comparison is to observe what happened when the knowledge economy workers worked from home during the pandemic and shops which supported the office economy were hard hit. Real life AV testing important but comes with reputational risks Risks from AV testing: In technology, being an early mover has advantages but can also pose risks. Deploying AVs on real roads gives a company a dialogue with regulators, and ability to learn from miles driven – both being positives. Waymo have said they have driven 20m miles (800 times around the globe) on public roads over 25 cities, adding 10m miles over the most recent year. Yandex and Baidu have done 1m miles each. GM’s Cruise AV entity is said to be adding 1m miles on monthly basis. 53 Tesla have said they have done over 3bn miles in Autopilot mode.54 Uber have said they have done “millions” of AV miles in testing and “tens of thousands” of passenger journeys. This kind of real road data can improve the AI/ML technology, as to make it come across real life situations that it may have not modelled and so can learn from them. However, the downside with AV today is that road testing has the risk of leading to accidents and thus bringing a host of associated public reputational risks. Additionally, the current AV regulation frameworks around the world, it could be argued, are very fragmented, where no international standards have been agreed upon. This poses further risks for AV companies in terms of roll-out. Fewer vehicles needed on road with AVs, but impact on metals and insurance? ESG: Impact of AVs translating to fewer vehicles on the road: If AV vehicles take off at some point for passenger vehicles, there is the potential due to fleet sharing efficiencies for ride hail services, traffic mapping algorithms and others, that fewer vehicles may be required on the road when compared to current traffic patterns with traditional human-driven vehicles. For example, a study shows that within an American city like New York, it could reduce the volume of cars by 76% - therefore reducing emissions and improving quality of life for citizens in urban areas.55 In theory, fewer vehicles on the road naturally means that there could be an impact on the demand for metals. On the flip side, as these vehicles will be more efficient, they will do more miles than human driven ride-hailing or privately owned cars, so could need replacing ______________________________________ 52 "America has a massive truck driver shortage. Here's why few want an $80,000 job", Washington Post, May 2018. 53 "Waymo's autonomous cars have driven 20 million miles on public roads", Venture Beat, January 2020. 54 "Tesla drops a bunch of new Autopilot data, 3 billion miles and more", Electrek, April 2020. 55 "On-demand high-capacity ride-sharing via dynamic trip-vehicle assignment", Javier Alonso-More, Daniela Rus et al, National Academy of Sciences of the United States, November 2016. 60 Thematic ● Global November 2020 more frequently due to wear and tear. On the insurance side of things, as the AVs will contain more computing and sensing equipment, the insurance per AV could be high when compared to traditional vehicle sensor content. Can 3D printing reduce CO2 and inadvertently jobs in some regions? 3D printing ESG: Pressure for 3D production to drive CO2 down: One of the big theme discussions to come out of the pandemic is whether some global supply chains should be re-localised. Automation technologies like 3D printing or AM (additive manufacturing) could play a role if this becomes more of a trend, to make businesses more resilient. We believe the economic reason is not the only reason to re-localise some parts of global production. It’s estimated that using AM manufacturing can reduce CO2 emissions by 25% CO2, according to Thurun Energy. There is the potential for localised production of goods through technologies like AM to place pressure (e.g. from investors or through regulation) on some companies to adopt the technology, due to the ESG gains. It’s estimated that if AM is used by a variety of industries such as manufacturing goods, passenger vehicles, industrials/power generation, aviation, trucking, shipping and O&G, it could reduce global CO2 by almost 7%. Another knock-on implication of 3D printing locally is the re-shoring of manufacturing jobs from overseas. 3D printing gives many efficiencies ESG: 3D printing in industry can reduce waste, cost and time: Many industries like automotive keep spare parts in stock. Not only does this add to the cost but also creates potential waste. 3D printing using AM techniques has the potential to reduce this waste going forwards, with lower volume production (say vs. injection moulding as an example). There can also be cost savings associated with optimised printing from single builds. For example, a company called Betatype used laser powder bed fusion (LPBF) to create 300+ metal auto parts. Using the optimised system, they produced each part for USD4 vs. USD40+ un-optimised 3D printing. They were also able to reduce the lead times from 444 to 34 hours.56 In another example GE used 3D printing to reduce the part count from their catalyst turboprop engine from 845 to only 11 – meaning fewer parts to certify. It not only reduced the cost but also fuel consumption by 20% and reduced the overall engine weight.57 Revise IP for 3D printing era? ESG: 3D printing and IP issues: The ability to 3D print a host of things, anywhere in the world, raises issues of IP. For example, an object can be reverse engineered and made into a 3D computer model (e.g. an STL file), from using a combination of 3D scanning tools and design work. This means that teams with skills can replicate almost any object thus lowering the barrier to entry for manufacturing a variety of widgets. There was an interesting case in the United States, where a company (ClearCorrect) which made 3D printed orthodontic aligners, had their 3D model sent over the internet from their overseas internal designers. Another company (Align Technology), also operating in the United States, filed a patent infringement against ClearCorrect. The Federal Circuit overturned the initial US ITC (International Trade Commission) decision that they had no jurisdiction over digital data, including 3D printing files sent into the United States from overseas, as they can’t be considered ‘material things’.58 In some areas like medical equipment, countries may face a dilemma going forward; should existing IP be respected, or should IP laws be changed to adapt to new technologies, to lower cost for health-related products, for the public good. This kind of easing of patents has happened previously during wars and also during the current pandemic (for potential COVID19 vaccinations). Going forward governments will have to re-visit IP in the digital economy to balance the public good and the ability to foster investment in innovation via patents – perhaps even enabling pooling of patents for easier use. ______________________________________ 56 "How real value of metal additive manufacturing can be leveraged for automotive production", September 2020. 57 "6 ways to reduce manufacturing costs with 3D printing", AMFG, September 2018. 58 "ITC connot police 3D printing data transmission in ClearCorrect case", 3Ders, November 2015. 61 Thematic ● Global November 2020 Death, taxes and 3D printing 3D printing, trade and tax… An interesting question to ask is what are the implications of reduced trade flows, if it becomes a normal business model for goods or parts to be printed for a company locally rather than sourced from abroad? 3D printing has trade and tax implications, which are very complicated issues, and can vary regionally. Currently, digital flows including 3D printing files are not subject to tariffs. In fact, World Trade Organization (WTO) members have periodically renewed commitments to ban customs duties on electronic transmissions (see Taxing digital trade). The moratorium is due to expire to expire in mid-2021 unless renewed further. However, some countries such as South Africa and India are concerned that developing economies will lose out on future tariff revenue as goods become increasingly digitised (e.g. via 3D printing). The ECIPE present the other side of this discussion, and estimate that imposing tariffs on digital flows could lead to an annual GDP decline of USD10.6bn for developing economies. Removing the moratorium could also affect public finances via tax erosion. For example, they calculate that India could lose USD2bn in VAT, as well as corporate and income taxes if duties are imposed. Moreover, according to the World Economic Forum, digital duties could slow down 3D printing adoption due to the higher trade costs and suggest that VAT or sales tax could be a better way to approach this. They suggest that 3D printing could also foster innovation and level the playing field between countries in production.[1] If there is a revenue gap due to loss in imports, how might governments bridge the revenue gap? There are several good open questions. In 2019, the Federal Revenue Service (FRS) in Brazil ruled that a 3D printer is considered a device which manufactures products, taking raw materials and morphing it into a new real product. The implication of this is that if a retailer 3D printed something for a customer, they could be liable for an indirect tax on the sale of goods – paid by the merchant, who passes it to the consumer through the price of the product. 59 ______________________________________ 59 "Brazil rulign on tax treatmet of 3D printing could have global implications", 3D printing media network, July 2019. 62