AI: Over-hyped or Under Appreciated?
By Rob Oellermann, Portfolio Manager, Laurium Capital
If you have not been bombarded with breaking news and views on the subject of Artificial Intelligence (AI) since the end of 2022 you must have been living under a rock. This article assumes you have been alive and well and living in the real world, occasionally dipping in and out of news or thematic articles about this new age of accelerated computing power and Machine Learning (ML). So perhaps you know what AI stands for, but now you want to know a little bit more – what makes it different, for instance, from being just another search engine, like Google? How could AI affect our daily lives? And very importantly for those thinking about their life savings from time to time: what are fund managers thinking about the investment opportunities arising from this development?
At a high level, generative AI applications are simply much smarter than linear search engines. Analogy: instead of teaching a friendly lapdog (called Google) to sniff out a bacon sandwich from under your couch, a function like ChatGPT or (Google’s Bard) puts a whole pack of Border Collies at your disposal, just waiting for an idea. If you want a gourmet meal, an AI tool can provide a complete menu based on your preferences, find the right recipes, and then make a shopping list. It is safe to assume that, very soon, you will be able to have your personalised AI tool place an order for these items from your local grocery store, operating within a budget, working to a schedule, maximising your loyalty points and discounts. This is an altogether much smarter pack of dogs! This simple example is merely the tip of the iceberg.
What exactly has facilitated this newfound “brainpower” at your fingertips? In very simple terms there has been a huge leap forward in accelerated computing architecture and machine learning capability in just a few years. We now have massive computing power running on focused GPUs, at a high pace, training themselves to find complex meanings and inferences that lie behind single words, sentences, patterns, and even pictures. It is all about using this parallel processing power to train a computer to “fill in the gaps” and make informed choices based not only on what it knows but also on what it can infer. These “large language models (LLMs)” literally fill in the gaps between 100bn+ individual data points (so-called “parameters”).
We can thank the gaming industry for developing the basic hardware that operates in this parallel, accelerated world of training and learning. This was the main focus area for Nvidia, whose technology know-how and GPUs have become the key enablers for LLM training and hence all things AI. These LLMs are literally the classrooms for generative AI tools. We see hardware as the first and strongest initial investment opportunity arising from AI.
At Laurium Capital, managers of the PPS Stable Growth Fund*, we have found several other companies focused on servicing upgrades needed in the data centres, as attractive investment opportunities, notably Broadcom. Broadcom has not attracted the headlines in the same way that Nvidia has. It does not benefit quite as materially as Nvidia (whose CEO called out a $1tn revenue potential!), but its advanced network tools as well as its role as a key vendor to Nvidia and Google (despite recent news articles) means that it has a solid pipeline of revenue growth into 2024 and beyond. Yet, it is still trading at a more modest mid to high teens PE multiple. Broadcom has been a substantial contributor to our strong performance this calendar year and remains a holding in the PPS Stable Growth Fund as of 31 August 2023.
The bigger uncertainty facing investors has been thinking about AI further down the line. We have been asking ourselves the question:
“How will AI be commercialised for financial gain?”. Out of interest, I posed this exact question to ChatGPT, and received the following answer:
AI, driven by machine learning and LLMs, offers numerous avenues for commercialisation:
- Product Development: Companies can create AI-powered products and services, such as recommendation systems, virtual assistants, and fraud detection tools, to address market needs.
- Data Monetisation: Organisations can leverage AI to extract insights from data, enhancing decision-making processes and creating value through data-driven services or targeted advertising.
- SaaS Offerings: AI can be packaged as Software as a Service (SaaS) platform, allowing businesses to access AI capabilities on-demand, reducing the need for in-house AI expertise.
- Custom Solutions: Developing bespoke AI solutions for specific industries or business problems can be a lucrative avenue. Tailored AI systems can optimise processes, improve efficiency, and solve unique challenges.
- Consulting and Integration: AI consulting firms can assist businesses in integrating AI into their existing operations, ensuring AI technologies are effectively adopted.
- AI Marketplaces: Companies can create marketplaces for AI models and services, enabling developers to monetise their AI creations while providing businesses with a diverse range of AI solutions.
- Data Annotation and Preparation: AI relies on high-quality labelled data. Companies can offer data annotation and preparation services to AI developers, ensuring they have the necessary training data.
- AI Hardware: Given the increased hardware demands of AI, manufacturers of GPUs, TPUs, and other specialised hardware stand to benefit from the growing AI ecosystem.
It is fairly obvious that there is huge scope for new business ideas to emerge and flourish with AI. The research and development process is well underway in almost every company that we engage with. However, the most advanced commercial AI thinking to date has emerged from the large platform companies themselves (Alphabet/Google, Amazon, Tencent, Microsoft – all of which are holdings in our funds).
Microsoft has very impressive commercial initiatives around AI and leads the pack. Microsoft actually kicked off the investment ‘frenzy’ by investing a rumoured $10bn in OpenAI, the company behind ChatGPT, in January 2023. The recent launch of Microsoft’s CoPilot AI add-on for its Windows 365 and Dynamics 365 products, both at a significant uplift in subscription fees, contributes to the investment appeal of Microsoft. Each of the other platform players mentioned has demonstrated strong capabilities but no clear commercialisation goals to date, we continue to watch this with keen interest.
In summary, this article only scratches the surface of the recent news flow and the potential still to come from AI. Yes, there is a lot of hype around AI. However, the rapid developments underlying the technology shift are real and will be far-reaching…. and there is significant real spending involved. At Laurium Capital, we are intensely monitoring this complex, fast-moving, and evolving landscape and feel fully justified in retaining significant exposure to these stocks in our funds.