The Investment Case for AI and Robotics

Computational systems and software that evolve with data can solve intractable problems, automate knowledge work, and accelerate technology’s integration into every economic sector. The adoption of neural networks should prove more momentous than the introduction of the internet and potentially create 10s of trillion dollars of value.  Meanwhile, Autonomous technology and adaptive robotics are not just reshaping the transportation and manufacturing sectors but also forging new paths in healthcare, agriculture, and service industries.

Today, investors have the potential to capitalise on these advancements, which hold the promise of redefining our daily lives and fuelling unprecedented economic growth and human prosperity.

The ChatGPT moment and its impact on knowledge worker productivity

With superhuman performance on a wide range of tests, AI models like GPT-4 should spark an unprecedented boom in productivity. Building on years of progress since Google invented transformer architecture in 2017, ChatGPT catalysed the public’s understanding of Generative AI in December 2022. No longer a tool just for developers, ChatGPT’s simple chat interface enabled anyone speaking any language to harness the power of large language models (LLMs). In 2023, companies scrambled to understand and deploy Generative AI in what was the year of AI.

ChatGPT Delighted Consumers and Amazed Enterprises¹

Importantly, AI already has boosted productivity significantly. Coding assistants like GitHub Copilot and Replit AI are early success stories that have boosted the productivity and job satisfaction of software developers. AI-powered assistants are increasing the performance of knowledge works and, interestingly, benefiting underperforming workers relatively more than high performers.

AI Already has Boosted Productivity Significantly²

As a result of these efficiencies, cost declines across industries have experienced a paradigm shift. Over the past century, the cost of authoring written content has been relatively constant in real terms, representing USD $300 to $USD 400 inflation-adjusted per 1,000 words. During the past two years, as the writing quality of large language models (also known as “LLMs” – machine learning models that can comprehend and generate human language text and beyond) has improved dramatically, with the cost collapsing to under 16 cents per 1,000 words.

The Cost of Authoring the Written Word has Collapsed³

Artificial intelligence has the potential to automate most tasks in knowledge-based professions by 2030, dramatically increasing the average worker’s productivity. Software solutions that automate and accelerate knowledge work tasks should be prime beneficiaries. This acceleration of knowledge worker productivity represents a potential multi-trillion dollar opportunity. In other words, the opportunity for value accrual in artificial intelligence does sit not solely with hardware players; there is massive upside for software companies incorporating artificial intelligence smartly into their established businesses and reaping the benefits of efficiency and improved efficacy.

Accelerating the Growth of Knowledge Worker Productivity Represents A Potential Multi-Trillion Dollar Opportunity⁴

AI advancements are catapulting robotics

The leap forward in AI performance is not solely impacting the trajectory of knowledge worker productivity. Large language models and generative AI are accelerating the progress in robotics. Advances in computer vision and deep learning have improved robot performance 33-fold in seven years. Robots are already surpassing human performance by greater than a factor of two and it is unclear where the upper limit will be.

Increased Performance is Stimulating Demand for Industrial Robots⁵

Trained by GPT-4 to perform robotics tasks, a neural network performed better than human expert coders on 83% of tasks, with the margin of improvements averaging 52%. Indeed, the convergence of artificial intelligence and robotics is creating a massive opportunity set.

Large Language Models and Generative AI should Accelerate the Progress in Robotics⁶

In addition to driving outperformance, large language models should facilitate regulatory approval given that they enable text-based training, validation, and self-explanations. In other words, this transparency will be easier to analyse and understand the causes behind effects, the inputs behind outputs.

Importantly, costs are dropping precipitously as the technology is improving. Declining cost curves are key to technological progress and adoption, and similar to the other technologies ARK models and analyses, they are particularly evident in the market for industrial robots. We are seeing the costs of industrial robots fall by 50% for every cumulative doubling in their production, a relationship known as “Wright’s Law.” Lower prices are accelerating demand for industrial robots, as evidenced by the explosive growth in unit sales depicted below.

Lower Prices are Stimulating Demand for Industrial Robots⁷

Amazon, one of the world’s largest retailers, has ramped up its usage of robots which are freeing its employees from tedious physical tasks.

Many Companies are Likely to Deploy More Robots Than Humans⁸

Automation’s impact on productivity has transformed industries

This exponential demand in robots is transforming industries meaningfully. A key component of robotics is automation, which has already had a profound impact

Wasn’t 2023 the first year the no. of robots in Amazon warehouses surpassed the total no. of Amazon employees? Let’s add some context, as to why we’re using Amazon as an example here. It’s not the first company that comes to mind for most people. But the answer lies in their incredible use of warehouse robots, more than any other retailer.

Automation’s Impact on Productivity has Transformed Industries⁹

In our view, the largest automation opportunity is also the largest AI project: autonomous ride-hail (also known as robotaxi). Adjusted for inflation, the cost of owning and operating a personal car has not changed since the Model T rolled off the first assembly line more than 100 years ago. At ARK we estimate that autonomous taxis at scale could cost consumers as little as $0.25 per mile, which would spur widespread adoption. This is a direct result of steep cost declines in battery pack systems and improving LLMs that power these autonomous networks.

Autonomous Ride-Hail is Likely to Increase Access to Convenient Point-to-Point Transportation¹⁰

At $USD 0.25 per miles, autonomous transportation could serve a wider population than human-driven ride-hail does today. In the meantime, based on the value that consumers place on their time, demand at higher price points could be significant. As depicted below, ride-hail is likely to create a $USD 11 trillion addressable market, which we believe few market participants are modelling or incorporating into their investment theses. ARK’s medium- to long-term time horizon coupled with our benchmark agnostic approach afford us the ability to allocate to this opportunity set attractively.

Ride-Hail is Likely to Create an $USD 11 Trillion Addressable Market¹¹