Artificial intelligence (AI) is the foremost disruptive technology in today’s economy. The scope of its potential is vast – offering the promise to revolutionise the way we work, the products we use and the services we consume.
Yet, part and parcel of such an emerging digital force are its potential risks, uncertainties and creative destruction that could disturb the security of our everyday lives.
For investors, alongside the myriad opportunities for those best able to harness the capabilities of AI lurks a complex investment landscape. An active and flexible approach will therefore be necessary if investors are to successfully navigate the era of AI.
AI: an investment opportunity that is hard to avoid
The launch of ChatGPT in November 2022 was the innovation trigger that pushed AI into the mainstream spotlight. Two years on, we have witnessed many businesses race to take advantage of this technological advance. Worldwide spending on AI services is expected to more than double by 2028, when it is forecast to reach USD 632 billion1.
Right now, those dominating the AI terrain are primarily the providers of foundational technologies which are driving the necessary infrastructure roll-out. But, over the medium and longer term, Al’s potential as a multi-purpose technology offers a much wider array of applications. Its diverse use cases will likely impact nearly every sector of the economy – both positively by boosting productivity, but also negatively by potentially eliminating jobs.
With so many differing companies investing in the AI story, there will likely be vastly different levels of return on these investments, presenting challenges and opportunities for investors.
Foundational leaders
The initial leaders advancing the AI theme are either those progressing innovation in AI or the ones who are providing the tools to drive the technology forward.
The development side is relatively concentrated and includes the creators of Large Language Models (LLMs) and machine learning algorithms, as well as companies with proprietary data sets that can be used to train AI models. From an investment perspective, it is important to note that not many developers currently trade on public equity markets.
The infrastructure build-out side is much broader and, arguably, more accessible for investors. Cloud service providers offer a gateway to AI. Thanks to their vast data centre footprints, cloud providers have the bandwidth to host AI models. Semiconductor manufacturers produce the Graphics Processing Units (GPUs) and Application Specific Integrated Circuits (ASICs) essential to support AI training and inferencing. Software development tools and database software are also supportive components for AI initiatives – AI algorithms are software and can be developed and managed with the same tools used for traditional applications. Interestingly, AI can also be applied to improve these software tools! Meanwhile, cybersecurity is key to protecting algorithms and data from attacks; cybersecurity systems can also incorporate machine learning techniques to better identify and thwart threats.
AI enablers also encompass semiconductor capital equipment and materials suppliers, foundries and providers of networking equipment and data storage systems. In addition, energy efficiency and alternative energy solutions are integral to AI. In fact, several leading technology giants are funding nuclear power projects to ensure an alternative source of low-carbon electricity for their energy-hungry data centres.
Future beneficiaries
The pinnacle for AI productivity is certainly some years away, but AI is already being successfully integrated into a range of applications. In particular, AI is a key enabler of versatile tools such as recommendation engines, manufacturing automation and document production that can be utilised in all areas of the economy.
Within sales and marketing, chatbots are used to interact with customer service requests and to simplify workflows, which are leading to productivity improvements. Scientific applications include pattern recognition and developing new protein structures for new drug discovery and materials science, while complex computer simulations can be used to enhance weather models. In the industrial sector, manufacturers and equipment providers are implementing AI systems to enhance process controls and virtual digital twin models. In healthcare, AI is being applied to medical image analysis to classify tumours, which could improve the decision-making of surgeons during brain procedures. AI copilots are also helping to write summaries of medical reports and patient visits, which is having a positive impact on doctor and pharmacist productivity.
With such diverse potential, AI will be a key growth driver moving forward. The companies that should reap the most rewards from AI are those able to leverage proprietary data or gain a competitive advantage for products or services that can be widely adopted.
Challenges and risks
With AI being adopted by an ever-widening proportion of the economy, attention must be diverted away from focusing on its promise to properly evaluate the risks – albeit often unintended – of its deployment.
With AI-generated ‘synthetic’ content becoming more mainstream, users must be able to easily discern what is real from what is fake. Copyright and intellectual property of model inputs and outputs are becoming key topics of debate. In addition, large language models have proven susceptible to producing ‘hallucinations’ or outputs which have incorrect or misleading information.
In a world where our security is already threatened by irresponsible actors manipulating the use of disinformation and fake news, the implementation of AI-tagging could better identify the source of content in future. Inbuilt bias could be mitigated by using more diverse and representative training data. And developers could take greater steps to ensure high-quality datasets are used as a default and be transparent about sources; after all, AI will only ever be as good as its inputs.
There are also fears that AI could be abused for surveillance purposes; while hacking and cybercrime could also take advantage of AI to perpetuate more sophisticated criminal activities. Yet, perhaps most consequential is AI’s potential to create and displace jobs. The IMF has projected that AI will affect almost 40% of jobs around the world2 – such a transition must be handled fairly and justly to avoid deepening inequality and sparking social unrest.
The world’s policymakers are rightly turning their attention to AI, but there is a fine balancing act between legislation that leverages the potential of AI for the benefit and protection of society, and rules that threaten to stymie innovation and curtail economic progress.
Capturing the broadening AI opportunity
As AI becomes more embedded in the global economy, the opportunities for investors will broaden. Right now, the main beneficiaries have been the relatively small universe of enablers, but this will evolve as the technology matures both within and beyond the technology sector.
With AI’s potential to creatively expand and disrupt the economic landscape, investors will need to adopt an active and flexible approach so they can capitalise on the positive outcomes while avoiding the pitfalls. At BNP Paribas Asset Management, our Disruptive Technology strategy is focused on digital transformation. We look to invest in the leaders and beneficiaries of digital innovation, while avoiding those companies that are unable to adapt. We believe the AI revolution is one of the most exciting, durable and expansive themes in technology investing and its economic impact will only increase over time. Our unique process of examining the AI proposition from multiple perspectives means we can capture the full scope of AI’s benefits now and into the future. This means a third of our investments are already in companies outside of the traditional technology sector, placing our strategy in an optimal position to capture the full benefits of AI implementation.
