Investing in Artificial Intelligence

Our resident EasyAssetManagement Chief Investment Officer Shaun Krom shares his views on the AI industry and some key companies that are influenced by it.

It is not possible to turn on the financial news without mention of the latest developments in Artificial Intelligence (AI), and which companies people believe will be the winners/losers in this new market paradigm.

When we think of investing in the theme of AI we can start by thinking about how AI may disrupt future industries. On a high-level, AI has the ability to automate tasks, whether white collar or on the factory floor, which can create efficiencies and boost productivity and profitability. It can allow companies to utilise more of their data thus enabling them to make more data-driven decisions which can improve design, marketing, customer service, self-driving etc. New technologies drive innovation and can birth new products and services; or companies that can tackle a problem in a whole new way or provide a solution to a problem that you hadn’t previously considered.

AI has the potential to disrupt many incumbents and create new winners; on the other hand, it might be embraced by current stakeholders thus entrenching them even further.

Since AI has the potential to disrupt so many industries and companies, when looking for companies to invest in, it’s worth investigating companies throughout the different layers of the AI value chain. From the infrastructure layer to model creation, model deployment and everything in-between. The below is not an exhaustive list but it provides a framework of how an investor might want to tackle their own AI investment strategy.

Infrastructure Layer

AI deploys Large Language Models (LLM) which rely on Graphical Processing Units (GPUs), due to the need to run processes in parallel with high bandwidth memory. The clear leader, which everyone by now knows about, is Nvidia (NVDA). Advanced Micro Devices (AMD) is a company that is biting at their heels – their legacy business is still the biggest part of AMD and even while they are taking market share from Intel on the server business, they are scaling up their GPU business.

But to make a chip you need to have an advanced physics engine. Think of how architects use CAD to design buildings; in the same way, chip designers use Electronic Design Automation (EDA) tools to design chips. This is not only used for leading edge GPUs but any chip design that happens at Google, Microsoft, Apple or wherever. EDA is a very complex field and as such has divided itself into an oligopoly with a few key players owning the market. Thus, they have a wide moat. There are only two companies listed that are pure play EDA companies: Synopsis and Cadence. They charge a software as a service (SaaS) fee with different modules and add-ons.

These chips need to be made in a foundry of which the Taiwan Semiconductor Manufacturing Company (TSMC) is the leader. But TSMC needs to buy the machines to deploy in the foundry to make the chips; examples of this include ASML Holding and Lam Research.

Model Creation and Cloud Deployment

The listed companies that dominated model creation are the big three cloud providers of Alphabet, Microsoft (which includes OpenAI), Amazon and Facebook (which doesn’t have a cloud offering). Meta has taken an interesting approach here. They open source all of their coding as they do not have a cloud to monetise. So they have gone open source so that they can get maximum take-up, which they can then use to enhance their current offerings.

Companies that create LLMs need to train their models on the large public clouds. They then deploy those models on new data, called inference, which also utilise the public cloud providers, although Cloudflare (NET) offers an interesting alternative to inference at the edge.

The above companies can be thought of as the picks in a “pick and shovel” strategy. Next, we discuss some of the shovel companies. Some of which are mentioned in this article on cybersecurity.

Data

MongoDB (MDB) offers the data foundation, Elastic (ESTC) the analytical tools, and Snowflake (SNOW) the collaborative stage for AI utilisation. These three companies create a powerful data pipeline that fuels the LLMs' growth and enables them to perform complex tasks.

MDB provides the flexible database infrastructure needed to store and manage the massive datasets LLMs rely on for training and inference. Elastic search allows companies to search through mountains for data for LLM to digest. While Snowflake's cloud-based data platform stores, analyses, and shares structured and unstructured data in a secure, scalable way for feeding an LLM's data requirements.

AI Powered Companions for Modern Business

There is software that allows companies to use AI to understand their workflow and understand their customers and employees. Companies like Workday, Salesforce and ServiceNow empower companies to connect with customers and employees on a deeper level. Whether it's predicting customer churn, analysing customer interactions, or viewing IT health; they allow for streamlined operations, boosting productivity and delighting customers.

Retail

AI is transforming the retail landscape, and these three giants exemplify its potential:

Shopify leverages AI to create personalized shopping experiences such as product recommendations, customer requirement predictions, and personalized marketing campaigns. This creates a seamless shopping experience and drives higher sales.

Amazon uses AI in a similar way to Shopify, with the added function of using it to optimise its and inventory management, predict demand; and customise that for its shoppers.

Nike uses an LLM to generate personalized emails for its customers. The emails are tailored to the customer's interests and purchase history. Nike together with Unilever use LLMs to generate ideas for new apparel and food products.

Conclusion

There is a universe of companies related to the AI industry that can be considered for investing in. It helps to outline how much AI will deliver in terms of earnings (think of Nvidia vs Shopify); where the growth may come from; and whether a company’s potential growth is sustainable over the long/short term (i.e. what your investment timeframe is for each stock). This can provide a framework for your investments. Following that you may want to consider company specific valuation metrics before you decide to invest.