An Early Innings for Artificial Intelligence

Artificial Intelligence (AI) will continue to be a major focus, with strong capital investment expected in the second half of 2024. Major cloud providers are increasing their spending on AI servers due to high demand that outstrips supply.

Morgan Stanley believes the monetisation of the AI ‘compute cycle’ will likely shift from semiconductors to infrastructure and then to software and services. Semiconductor companies (e.g. Nvidia) and infrastructure (e.g. hardware and data centres) are viewed as early beneficiaries in the cycle.

Currently, semiconductor and data centre investments are still in their early stages and are expected to grow over the coming years.

Semiconductors (or chipmakers) offer significant opportunities due to their essential role in AI. They provide the computing power needed for AI, making them a crucial part of the AI ‘boom’. These companies are expected to be the primary beneficiaries as AI continues to grow.

The semiconductor sector has seen strong performance over the last 10-15 years due to global technological progress and this track record will likely extend further as the progression of AI drives strong demand for newer, more advanced microchips.

Major global technology companies such as Amazon, and Microsoft have recently accelerated their investment in cloud computing and AI and are responsible for much of the demand for AI ‘compute’. For example, recent earnings releases showed capex budgets across these tech giants increasing by around 44%, with a majority of the capex being invested to build AI models.¹

While Nvidia remains the current dominant provider of AI compute power for training AI models, Broadcom has an apparently successful business in designing ‘custom’ AI chips for large technology companies which are required for more specific AI workloads² . Investment into AI model training has been rising and underpins an attractive outlook in earnings for chipmakers. The supply of AI graphic processing units had initially been constrained due to the lack of technology able to keep pace but is expected to improve as production and supply increases. A more significant phase of AI is expected as models perform high volumes of everyday tasks in the real world i.e. adoption of AI is widespread. This evolution will require more compute power and associated infrastructure overtime.

Data centres are emerging as a compelling investment thematic due to the world’s hunger for data and adoption of AI which means data centres will exist as potential long-dated investment opportunities. It is estimated that data centre demand will grow at a ~20% compound average growth rate from 2024 to 2027.³ The data centre build required to meet this growing demand is substantial and there is an urgency around delivering the infrastructure for the growing global data needs, with AI expected to accelerate this trend.

There are a vast range of date centre beneficiaries, most of which have direct or potential exposure to the rapidly expanding need for data centres. Companies with direct data centre ownership as well as companies providing the compute power, connectivity and the supporting hardware infrastructure will be well positioned as beneficiaries.

Data storage is a major area of focus as chief information officers (CIOs) upgrade and monetise their data centre infrastructure to accommodate generative AI data and AI model development. Storage is showing attractive growth signs, with a CIO survey pointing to strong upward revision to storage growth expectations in 2024.⁴

Edge AI is the use of AI applications in devices (e.g. in smartphones and PCs). Recently, there have been several AI PC announcements that present an opportunity for hardware companies to increase their selling prices. Dell Technologies Chief Operating Officer, Jeff Clarke has called Edge AI on PCs as the ‘killer app’, suggesting the opportunity that lies ahead. AI applications in smartphones will also raise the potential for an acceleration in the phone replacement cycle.