Outlook 2023 - AI & Robotics: Applications ready to steal the show

With advanced applications coming in with a bang, Artificial Intelligence and automation are in a steep ascending phase. Their impact on societies and economies is becoming tangible, and the potential remains immeasurable.

Bottom line

After a complicated year in 2022, impacted by supply-chain problems and the collapse of the consumer segment, we believe 2023 will be a foundational year: the incoming recession may open a period of crisis, but this is precisely the kind of period that prepares the deepest transformations. Artificial Intelligence is no longer hiding in plain sight. Although the infrastructure phase is far from complete, advanced AI applications are rapidly exiting research departments. Despite their early stage, their capabilities are already transformational. AI's colossal potential will lead to seismic shifts within our modern economies and societies. Infrastructure will keep being rolled out to address the ever-increasing demand fueled by the rapid iteration of advanced AI applications. Our portfolio stands ready to capture these trends.  

Table of contents

Portfolio Snapshot
AI & Robotics Overview

Computing & Automation Systems
Artificial Intelligence Platforms
Productivity AI
Data Infrastructure
Data Management
Robotics
Autonomous Mobility
Semiconductor Manufacturing
Virtual Universes
Augmented Reality
Machine Learning
3D Printing

A glance in the rearview mirror
Catalysts / Risks

Portfolio Snapshot

Overview

The twenties will be roaring

Applications are there, with a bang

2022 will likely be remembered as the year when generative AI became a thing. But despite the deafening buzz and the already incoming controversies, the technology has a long way to go to reach maturity and its full potential. Yet the perspectives are compelling, both for what is now possible to do and for all that is implied behind the scenes for the supply chain. And behind generative AI, many other applications, some maybe less sexy but still with transformational potential, are quietly brewing. We strongly believe that AI is not far from its "iPhone moment": nobody expected it, then nobody understood it, and finally, the world was never the same anymore. One more reason to hop on the train.

The great divide

Opposite dynamics are expected to keep tearing apart the theme next year. On one side, consumer spending is stalling, shaken by inflationary pressures and bracing for the impact of the incoming recession. On the other side, the demand for infrastructure and AI business applications has never been stronger now that the capabilities of AI are becoming obvious. For those infrastructure players, the question is not whether there will be less demand but how they will be able to address it. Turbulences are still possible over the short term but will be remembered just as a blip, considering the scale of what is to come.  

What impact from a possible recession?

Apart from the B2B/B2C divide, we expect various impacts on our subthemes. Segments enabling efficiency at a reasonable cost (e.g., Robotic Process Automation) or which will be a hedge for salary inflation (warehouse automation) are likely to witness an acceleration. On the other hand, segments requiring massive CAPEX with distant ROI (industrial robots) or without having a defined business case (robotaxi services) are likely to suffer more. But as we previously said, the biggest threat to the theme is geopolitical: the invasion of Taiwan by China. Paradoxically, the Ukrainian conflict makes it less probable, as it has shown the difficulties of invading a well-defended country. But, alas, the same conflict showed that logic does not always prevail.

-> Table of Content

From stars to stables… to stars again?

A variable macroeconomic impact

The ripple effects of an economic contraction in supply chains will inevitably impact the segment. This will especially be true for consumer-dependent players, such as suppliers of components for industrial robotics or DRAM makers, which will have to cope with the double effects of a pricing downcycle and a collapse of the consumer electronics market. However, we expect players with more diversified or specialized exposure to better withstand the hiccup, such as players providing specialized chips dedicated to AI workloads.

The year of rebound for chips?

From a stock-market perspective, the deceleration has been brutal for semiconductors players. However, this fall has hidden two different dynamics: on one side, the consumer segment has burdened the sector; on the other, the demand for AI chips remains strong. With the advent of a new manufacturing node at both Samsung Electronics and TSMC, as well as with innovative manufacturing technologies (e.g., chiplets and 3D-stacking) benefiting performances and costs, we expect an inflow of specialized chips to build the highly-demanded infrastructure for advanced AI applications. With the race for AI supremacy raging, we do not see how dedicated semiconductor players could not benefit from the trend.

Next-generation computing to remain in the labs

Computing scientists are already working on next-generation architectures, such as quantum or neuromorphic chips, which offer great promises and regularly trigger bursts of enthusiasm. However, they are far from ready to play a significant role, and we expect them to remain at the prototype stage. Much work remains to be done at the hardware and software levels. The technology will rely on a brand-new ecosystem, which will likely not be completed before the decade's end. A revolution is in the making, but we will have to wait!

-> Table of Content

 

Get ready to question everything you see

Content generation keeps improving and triggers its first major controversies

Automatic image generation through a text prompt has undoubtedly been a considerable breakthrough among AI applications in 2022. DALL-E is the perfect poster child. Models will keep improving, leading to better out-of-nowhere images, videos, texts, and sounds and revolutionizing the creative sector. However, the first snags are already appearing, with controversies regarding the origin of data used to train these models. On top of that, and despite the implemented safeguards, it is only a matter of time before a significant controversy caused by the use of an AI-generated deepfake comes to the fore.

The consumer segment, an improbable driver?

The market is witnessing the gradual arrival of specialized AI platforms, which enable the democratization of AI technologies in specific verticals. It is mainly the case in marketing, where AI can increase the probability of closing a sale at a marginal cost. This may be of particular interest for companies facing a dip in consumer spending and may materialize a particular tailwind for the segment despite the gloomy economic context.

Strongest traction in Healthcare

We believe the most significant traction for AI platforms will be in Healthcare. As demonstrated by Alphabet 's Deepmind giving the world the structure of 200mn proteins, the technology can accelerate drug discovery by a significant margin while facilitating and reducing the costs of subsequent clinical trials. Finally, progress in accuracy and interfaces will greatly help propose more accurate diagnostics in the context of an aging workforce (i.e., fewer doctors) and population (i.e., more illnesses). However, given the early stage of such technology, the incoming recession will act as the tiebreaker between companies with a differentiating technology and solid financial structure and the others.

-> Table of Content

The moment of truth approaches

AI assistants must close their business case

Virtual assistants such as Amazon's Alexa are facing a wall. On the one hand, technology has pioneered a revolution in human-machine interactions, with Alexa and Siri becoming common terms. On the other hand, their parent companies cannot seem to find a way to monetize them, except for Apple Inc. Facing an uncertain economic horizon, layoffs have started. 2023 will therefore be a pivotal year to determine if the technology can be truly transformational after finding a good business model or if it will just be another gimmick.

Macro drivers open doors for Robotic Process Automation (RPA)

The incoming combination of recession and inflation favors solutions helping to cut costs, especially those arising from using a high number of low-wage workers for low-added value (albeit necessary) tasks. RPA, and its little brother hyper-automation, enable the automation of tedious office processes and hedge the effect of wage inflation. It may therefore have substantial business opportunities over the next quarters.

GPT-4 is around the corner

Transformer neural networks materialized a breakthrough in Natural Language Processing (NLP), bringing a performance leap thanks to their ability to establish contextual links. OpenAI 's GPT-3 is the best-known model using this technology. It shocked the world at its inception in 2020 with its ability to generate humanlike poetry or computer code. The next version, GPT-4, is expected to debut in 1Q23 and could bring human-machine interactions to unheard-of levels. In any case, it will steal the show and generate a lot of buzz, turning the eyes of the public toward AI once again.

-> Table of Content

 

Stronger than ever

No slowdown in cloud spending

Cloud services do look like an unstoppable train. Although a massive recession would mechanically impact IT budgets, there is currently no sign of doom in forecasts: according to Gartner, global spending on public cloud services is expected to grow by almost 21%, i.e., more than in 2022, and with segments such Infrastructure-as-a-Service clocking to nearly 30%. It is worth noting that cloud services have some key advantages in a cash-constricted context: limited monthly operating expense vs. larger one-off capital expenditure, operational savings, and business process optimization.

The great decoupling

After years of rising tensions between the U.S. and China, the hammer dropped in November 2022 with the FTC altogether banning the use of equipment from vendors Huawei and ZTE, citing national security reasons. In Europe, the idea of building a sovereign cloud is regularly surfacing. Snowflake recently made headlines in France, alas, as a showcase of the dangers of storing data on foreign servers. Such concerns will only grow in the future and increasingly lead to a decoupling: China on one side, Western countries on the other. The latter block likely dividing itself into smaller zones. As usual, the resulting voids will have to be filled, generating opportunities for authorized players.

The virtuous circle of advanced AI applications

With the increasing rollout of advanced business AI applications, business leaders can see that the technology is no longer a gimmick but indisputably a must-have. Players can choose between investing in AI capability or risk losing their edge vs. competitors doing so. As the cloud is the most direct road to deploy AI at the scale of a whole organization, such a trend will mechanically benefit the entire ecosystem, fueling a virtuous cycle of adoption and democratization.

-> Table of Content

Go with the flow

Inflating data, inflating value, inflating risk 

The amount of data generated is perpetually on the rise, having increased tenfold over 2012-20. And with our daily lives and societies increasingly embracing digitization, the trend is unlikely to reverse soon, making data management vital if one wants to exploit it efficiently. However, this is not only about how to use data: with everything becoming digitized, more sensitive data becomes accessible and usable. Users expect data to be well-secured and reliable, implying the setup of additional safety and compliance layers. And as data is more sensitive, its attractiveness grows for hackers, increasing the risk of attacks and damage in case of a successful breach.

Unstructured data gaining importance

Data used in model training or data mining was traditionally "structured", i.e., organized in a labeled database allowing for convenient analysis. This meant either this data was coming from digital processes or was painstakingly organized manually. However, new technologies are progressively opening the door to unstructured data: pdf files, text documents, images, videos, sounds, etc. All these data represent human activity and potentially entail game-changing insights for businesses. This gradual evolution, which will complement structured data, will particularly benefit players like Snowflake or Databricks, which have demonstrated their ability to manage vast amounts of data efficiently and have already initiated a pivot towards unstructured data

Hyperscalers to maintain dominance, but not hegemony

Prominent cloud players have built a data management ecosystem on top of their infrastructure, successfully generating cross-selling and reinforcing their critical mass. Given their huge footprint and inertia, this dominance has little chance of shifting in the short term. However, we believe more and more players will try to exit this customer lock and diversify their suppliers. The rise of multi-cloud, which allows the best of breeds, is set to accelerate.

-> Table of Content

Macro drivers to set the tone

Industrial robotics in uncertain waters

Industrial robotics remains firmly anchored to secular trends. The ongoing demographic transition in developed economies requires finding a replacement for workers. At the same time, the rise of collaborative robots (cobots) helps optimize production processes. While the long-term future looks bright from a business perspective, the short-term is more uncertain. Plunging consumer spending and recession fears negatively impact manufacturing lines' investments, as they ultimately remain dependent on consumers buying the goods they produce. On the other hand, China might be a brighter spot: the country has firmly engaged in a coordinated upmarket move and faces increasing demography challenges. Nevertheless, due to U.S. sanctions, it risks being hampered by increasingly complex access to components.

No stagflation in logistics

Warehouse automation may well be the relative winner of inflation. Basic logistics operations heavily rely on low-skilled workers, for whom wages are bound to increase in the current inflationary context. This creates a conundrum when combined with falling economic indicators. Warehouse automation solutions, therefore, represent an attractive alternative: increasingly capable, they are scalable, operate regularly, and do not demand wage increases. The only major obstacle we see would be the upfront investment cost, which may be a problem in the event of a deep liquidity crunch. Still, we believe ROIs are getting increasingly favorable to deploying such solutions.

Cloud over personal robotics, not necessarily over healthcare

2022 has shown personal robot makers the gap between their offerings and what the market expects of such products. The most blatant example was Amazon.com Inc 's Astro robot, which still remains elusive one year after being announced due to problems with the reliability of the robot's software. More importantly, the high price tags of personal robots will not help adoption in the context of an incipient recession. The same could be true for healthcare robots, for which the price tag is even higher. However, we believe that for the players who already benefit from a large installed base (such as Intuitive Surgical), the outlook remains positive: even if sales of new devices were to stall, such companies benefit from recurring revenues thanks to the sale of consumables. This revenue stream is directly linked to the number of surgical interventions performed, and there is a substantial backlog due to the Covid-19 pandemic.

-> Table of Content

 

No driver does not mean no direction

Services and vehicles keep coming in

2022 was marked by substantial technological progress, with players such as Waymo (Alphabet), Cruise (General Motors), and Baidu launching fully driverless services in selected cities. However, these services remain non-paying for now on regulatory grounds. At the same time, Baidu announced a level 4 vehicle with a detachable turning wheel to be sold in 2023 for a cost of ~$37k despite the staggering amount of technology embedded. Therefore, we have good reasons to expect 2023 to be the year of commercial debuts for robotaxi services. Although the scale is likely to be limited initially, this would nevertheless be a significant milestone.

The tricky question of economics

The failure of Argo AI, backed by Ford and Volkswagen, came as a major blow to the sector, given the relatively advanced position of the company. However, this is a crude reminder of the ongoing financial reality: autonomous driving remains in its infancy, requiring further substantial investments with little certainty of rapid financial return, all while the economic context is deteriorating fast. Business models will have to be squared quickly. Thus, we expect industrial accidents of this scale to happen again in the near future. Paradoxically, this will reinforce solid players, but it will nevertheless be a wild ride for investors.

What game is Tesla playing?

Once touted as an undisputable trailblazer in the segment, Tesla has been steadily losing ground. Despite several fatal crashes, the company has doubled down on its technological choices and removed every sensor, save for cameras. At the same time, all other players tend to add several layers of different sensors to ensure redundancy. With the company facing delays, several high-profile probes regarding its so-called autopilot, and regularly increasing prices with little to show for it, 2023 might be a make-or-break year for the company's ambitions in the field.

-> Table of Content

 

Turbulences ahead

Consumer demand triggers the dip

The covid-induced renewal cycle is over: households have reequipped their aging computers and no longer benefit from stimulus checks. In the meantime, inflation and fears of a recession are taking their toll on consumer spending. The result is already a historic rout in the mainstream electronic market, with PC shipments down by ~20% in 3Q22, a downturn not seen since the mid-90s. Supply has fast transitioned from shortage to high inventory levels needing clearing, which will weigh on the semiconductor supply chain next year. B2B demand seems to hold up for now but won't escape impact in the event of a significant recession, further dragging down the segment.

Politics, politics, politics

After the latest U.S. sanctions, China is more than ever aware that it can only rely on itself. A double-down on investments in a local semiconductor manufacturing ecosystem is widely expected. Nevertheless, sanctions will not make it easy for the most cutting-edge segment. Consequently, the country's ambitions regarding the future of Taiwan will be heavily monitored. Policies will also have a significant role on the other side of the new iron curtain, with Western governments massively investing ($52bn Chips and Science Act in the U.S., multi-billion European Chip Act) in reconstituting independent local supply chains, at the risk of creating significant overcapacity when these investment come online by 2024.  

The renewable hope 

Despite the worsening macroeconomic outlook, the green transition seems unstoppable. Some segments may inevitably be impacted to some extent (e.g., Electric Vehicles). Still, the underlying technological revolution will keep bringing favorable outlooks to the supply chain of next-generation materials such as Silicone Carbide (SiC) or Gallium Nitride (GaN), for which demand far outstrips supply.

-> Table of Content

Still early, but not necessarily motionless

Metaverse is not ready for prime time

2022 has indisputably shown that the concept of Metaverse, as defended by Mark Zuckerberg, was far from being mature. Limited in capacities, with disputable design choices and a frank lack of attractivity, it will take years for the public to adhere to this version of the concept, and we do not see any meaningful trigger in the short term. On top of that, despite strong insurance from Meta Platforms Inc 's founder, investors have questioned the choice to invest heavily in what appears to them as a bottomless pit. With Meta Platforms Inc 's primary source of revenue being advertising, which will suffer from a recession, one may wonder if the Metaverse will not enter its winter and hibernate until better days, much to the chagrin of its creator.

Specific platforms rather than a single hub

We believe the Virtual Reality market will develop through specific platforms adapted to a given task. A single use-case is easier to build and upgrade, which is particularly attractive at an early stage when technologies are not yet mature. This, in turn, will allow aggregating users faster, bringing the scale upon which it will be easier to build and extend. On the contrary, bigger unified platforms are likely to suffer from the lack of a technological stack and users' defiance regarding using their data. In the end, they may simply be burdened by the weight of their ambitions.

Professional training slowed by interfaces

Training platforms have considerable growth potential, especially the ones related to healthcare, where human lives are at stake. Their potential is already apparent, yet it is only at the beginning of being fully exploited. It will be leveraged in the future by maturing technologies and increasing use of real-world data. However, their development will be determined by the progress in human/machine interfaces: a surgery simulator may perfectly model a human body, but it is almost useless if the trainee cannot benefit from the same movement precision on the day the surgery will be for real.

-> Table of Content

Reasonable hopes after the unreasonable hype

Apple to enter the fray: the "iPhone moment", finally?

2023 is expected to be the year of Apple Inc launching its long-awaited Augmented Reality headset. After years of development, the public debut of the device is expected to be a significant tailwind for the sector, considering the long history of successes in hardware for the company and the ecosystem momentum that usually follows. Although details remain limited, most rumors point to a powerful but pricy device packed with over a dozen cameras, with a cohesive design and dedicated to professionals. "This is a revolution", all over again?

Digital twins to gain momentum in a challenging macro context

Digital twins, based on real-world modeling systems used to anticipate future situations, are expected to gain traction among corporates in the context of an economic recession. Such technologies indeed enable cutting development time and costs while avoiding system downtime by anticipating breakdowns. Previously limited by the high computing costs and experience curve, the segment is now benefiting from much more user-friendly platforms accessible on a SaaS basis. However, the cost-benefit threshold remains to be determined, as the segment is still at an early stage.

Integration rather than specialization

In the business applications arena, we see the future belonging to integrated platforms rather than specialized players. Such platforms enable end-to-end control of the data chain, which remains the system's core while keeping open the possibility of integrating third-party solutions for some specific tasks. In addition, they allow keeping customers locked within an ecosystem, which is particularly meaningful considering the long lifecycle of industrial projects.

-> Table of Content

 

AI for an AI

The rise of platforms

Deploying AI at a large scale remains a complex task. It still requires a heavy infrastructure and large datasets to train custom models for a specific task, or either implies accepting relying on more generic models developed by someone else. This is why we believe Machine Learning (ML) platforms, which bring turn-key solutions to implement custom AI solutions, have a highway in front of them: the rise of advanced AI applications brings a clear differentiator to the table, prompting many businesses to scramble for their deployment, without the need to have specific in-house competences. Our conviction is that the trend will accelerate in 2023, even with an economic downturn, as these platforms provide a cost-effective way to implement the most efficient solutions.

Where's my data?

Data is the key to training models, hence the importance of collecting and classifying it for supervised learning. The latter implies strong opportunities for the data labeling segment, with a CAGR estimated at above 20%. However, collecting data is sometimes harder due to its intrinsic difficulty or the rarity of data-generating events. This is why we believe synthetic data have a huge potential: datasets are artificially generated through a simulation, providing cheap yet accurate material to rapidly train AI models. In other words, one of the only successful cases of "fake it until you make it".

Staying private

Despite ML not being something new, the field remains at an early stage, having been long confined to research. As usual, the most prominent players belong to the Big Tech club but dissimulate a large ecosystem of startups quietly growing behind the scene. These players have found it easy to raise capital in the past few years and comfortably stayed private. Considering the challenging market conditions, we believe there is little chance for this situation to change in the short term, except for the weakest players that may be in desperate need of cash.

-> Table of Content

Space to go up, healthcare to go down?

Still an industrial niche

For the time being, prototyping and development will remain the central use case for 3D printing. The technology is an ideal fit with processes requiring low volumes and rapid iterations but encounters problems when requiring larger scale, either from production volume or the size of created parts, which will keep slowing down its expansion for industrial use. However, we see a rising use in intermediate use cases, e.g., the rapid manufacturing of low-criticality spare parts, as done by Deutsche Bahn.

Welcome to infinity… and beyond?

2023 will be a breakthrough year for space 3D printing. The technology is already successfully employed for specific parts (e.g., SpaceX 's Dragon capsule thrusters), but the technology is about to be taken to the next step. Rocket player Relativity will make the first flight of an entirely 3D-printed rocket, which would be a significant engineering feat. Other players, such as Rosotics, are pushing behind and are willing to shake the established order. Knowing the extreme demands of the space environment (and the trend of associated technologies to later spread out in other segments of the economy), the coming year may well represent a turning point.

Healthcare needs to see beyond the short-term

Healthcare remains a huge source of opportunities for the segment, given the highly differentiating value proposal brought by the technology. However, it remains used chiefly for non-vital parts such as dental. Organ bioprinting remains a distant dream for now. Consequently, we expect the segment to suffer in a recession, as people will likely direct their spending to more pressing healthcare needs.

-> Table of Content

-> Table of Content

Catalysts

  • Application maturation. The first iterations of advanced AI already had a major impact on users despite their unpolished status. Refinement and better integration into existing processes will enable a massive boost in their adoption.

  • Recession-induced search for efficiency. Periods of crisis are the most transformational ones, and periods of economic downturn trigger a search for higher efficiency. AI can match both criteria.

  • Better accessibility. AI and automation required a high level of technical expertise to be deployed. The emergence of low/no-code platforms will change this fact and enable a smoother rollout.

Risks

  • China invades Taiwan. Such an event would be truly catastrophic for the theme, given the importance of TSMC in cutting-edge semiconductor manufacturing. The theme would enter an uncertain winter.

  • Polemics and fear. The impact of advanced AI applications will be transformational, hence their rollout will generate friction and resistance against change. Major controversies are bound to happen and may generate some resistance, dragging on the theme.

  • Restriction on data. Data is crucial to train AI models. However, people and regulators tend to restrict the use of personal, data which could lead to models not aligning with reality anymore.

Companies mentioned in this article

Alphabet (GOOGL); Amazon.com Inc (AMZN); Apple Inc (AAPL); Argo AI (Not listed); Baidu (9888); Databricks (Not listed); Ford (F); General Motors (GM); Huawei (Not listed); Intuitive Surgical (ISRG); Meta Platforms Inc (META); OpenAI (Not listed); Relativity (Not listed); Rosotics (Not listed); Samsung Electronics (005930); Snowflake (SNOW); SpaceX (Not listed); TSMC (2330); Tesla (TSLA); Volkswagen (VOW3); ZTE (Not listed)

Explore:



Disclaimer

This report has been produced by the organizational unit responsible for investment research (Research unit) of atonra Partners and sent to you by the company sales representatives.

As an internationally active company, atonra Partners SA may be subject to a number of provisions in drawing up and distributing its investment research documents. These regulations include the Directives on the Independence of Financial Research issued by the Swiss Bankers Association. Although atonra Partners SA believes that the information provided in this document is based on reliable sources, it cannot assume responsibility for the quality, correctness, timeliness or completeness of the information contained in this report.

The information contained in these publications is exclusively intended for a client base consisting of professionals or qualified investors. It is sent to you by way of information and cannot be divulged to a third party without the prior consent of atonra Partners. While all reasonable effort has been made to ensure that the information contained is not untrue or misleading at the time of publication, no representation is made as to its accuracy or completeness and it should not be relied upon as such.

Past performance is not indicative or a guarantee of future results. Investment losses may occur, and investors could lose some or all of their investment. Any indices cited herein are provided only as examples of general market performance and no index is directly comparable to the past or future performance of the Certificate.

It should not be assumed that the Certificate will invest in any specific securities that comprise any index, nor should it be understood to mean that there is a correlation between the Certificate’s returns and any index returns.

Any material provided to you is intended only for discussion purposes and is not intended as an offer or solicitation with respect to the purchase or sale of any security and should not be relied upon by you in evaluating the merits of investing inany securities.


Contact