Stock market: which companies are best placed in the AI value chain?
Who's best placed in the AI value chain (between semiconductor manufacturers, software publishers, the generative AI companies themselves, the GAFAs)?
As technology investors, our focus is on maximizing the benefits our portfolio can derive from
the current influx of investments into generative artificial intelligence by tech companies, and more importantly,
the long-term trends that will emerge from the successful integration of generative AI into everyday life by workers and consumers.
We propose a multi-phase investment philosophy, with each phase building upon the previous one. This approach allows us to identify opportunities and understand the inherent risks associated with investing in generative AI related companies. Despite the rapid consumer-level adoption, there are still uncertainties regarding the technology's capabilities and whether enterprises (and consumers) worldwide will see sufficient benefits to offset the costs of deploying and scaling generative AI models. For instance, it is reported that training GPT-3 cost $4.6 million, while semianalysis.com points out that ChatGPT's weekly operating cost for inference surpasses its training cost - making it clear that the ongoing cost of running models like GPT are significant.
Given this uncertainty, we believe that key milestones, such as the successful commercialization of generative AI-powered tools and features, will play a crucial role in securing strong IT budgets for 2024 and 2025 which will be needed to maintain robust investment in this field. For example, the successful launch and adoption of Microsoft's 365 Co-Pilot, an AI assistant for Microsoft productivity tools (Outlook, Word, Excel, PowerPoint, etc.), will be critical in the year following its general availability date, November 1st, 2023.
Over time, the successful commercialization of generative AI-powered tools will stimulate further investment in infrastructure and data technologies in industries beyond enterprise technology. However, as long-term investors, we believe it's crucial to not only select the best players in each phase and subsector from a financial returns perspective but also consider the long-term sustainability of these businesses to mitigate potential negative externalities that could impact both the fund’s financial performance and other stakeholders. Generative AI does indeed present inherent risks related to its use and development, such as:
Model ethics and transparency: audit of integrated external AI models, explainability, biases, reliability, governance, IP protection, balanced & safe dataset, algorithm restriction and moderation, data privacy, applications and end uses.
Human rights: labor and human rights in the supply chain and conflict minerals.
Potential labor impacts: potential job losses linked to automation.
Environmental impacts: higher energy consumptions than more traditional computing.
In considering our investment philosophy for generative AI, we identify the following phases as the core opportunities for public equity investors:
First Phase: Deep tech investments (the “picks and shovels”)
The first step in building strong AI capabilities is to build the infrastructure to train and operate generative-AI technologies. The infrastructure of AI is built on a combination of hardware and software, which we’d categorize as “deep tech”. Deep tech names are the “picks and shovel” of generative AI and are companies that are already benefiting (or will in short order) from the significant generative-AI investments made mostly by big tech companies and startups.
Timeline: Currently ongoing
Best positioned sectors: Semiconductors, IT Hardware and Cloud Infrastructure.
Best positioned stocks: Nvidia, Microsoft, Amazon, Google, Broadcom, AMD, Taiwan Semi, Micron, SK Hynix, Arista Networks, Quanta Computers, Wiwynn, Baidu.
Second Phase: Gen-AI-powered applications
The next leg of generative AI benefactors, in our view, will be for application companies who are market leaders in their respective markets. These companies, if they are running cloud-native or at least more modern data architectures, should be able to launch productivity-targeted features and capabilities into their core products. This should help them differentiate against competition and drive higher pricing and growth.
Timeline: Co-pilot launch (Nov 1st, 2023) and the next 12 to 18 months.
Best positioned sectors: Application software, Tech-enabled consumer applications.
Best positioned stocks: Salesforce.com, Intuit, ServiceNow, Procore, Duolingo, Gitlab.
Third Phase: The great infrastructure modernization
In our view, if the enterprise and consumer application businesses can prove strong-enough ROI from the use of generative AI, then businesses globally will rush to modernize their IT infrastructure in order to be able to build and run generative AI on their own data and workflows. This would be the “great unlock” moment for AI. In this case, we would expect an acceleration in cloud migrations but also in modernization of IT systems. Accenture, in their last earnings call mentioned that less than 10% have mature enough data infrastructure for AI, leaving a long runway of modernization ahead.
Timeline: 12-18 months out, depending on phase 2.
Best positioned sectors: Infrastructure software, public cloud infrastructure, and IT services.
Best positioned stocks: Stocks from the first phase, Accenture, Palo Alto Networks, MongoDB, Confluent, HashiCorp, Snowflake.
Fourth Phase: Global cross-industry adoption
The real world return on invested AI capital will take time to show up in the financial profiles of non-tech businesses. But once they do, we expect better growth from generative AI-enabled companies, with higher margins as workers become more and more productive.
Timeline: 2 to 3 years out, at least.
Best positioned sectors: Sectors/stocks from phase 1, 2 and 3 as well as other sectors such as retail, healthcare, and education. In the next section, we discuss the industries we expect to be most impacted.
The opinions and estimates herein are based on our judgement and may change without prior warning as may assertions on financial market trends which are based on current market conditions. To the best of our knowledge, the information herein is reliable but must not be considered as exhaustive. This document is not an offer or a solicitation to buy or sell any financial instrument whatsoever. References to specific securities or their issuing companies are merely for illustrative purposes and should not be construed as recommendations to buy or sell these securities. Past performance is not a reliable indicator of future returns. Opinions and strategies described may not be suitable for all investors. Returns and valuations for investments in any funds that might be mentioned may rise or fall and investors may receive more or less at redemption than the sum initially invested. Investors are warned that they could suffer capital losses.