Beyond near-term opportunities in software, we also highlight internet as another medium-term beneficiary as AI becomes integrated in consumer applications like gaming, entertainment, and advertising. (UBS)

Given the broadening AI demand trends, how should investors position in AI?
We believe the strong year-to-date rally in semiconductors and hardware stocks has fairly discounted the solid growth prospects in the AI infrastructure segment. Given the rich valuations, we are waiting for a pullback to turn positive on the segment again. Meanwhile, the risk-reward remains more attractive for software stocks, which, in our view, are well positioned to ride the broadening AI demand trends. We see significant opportunities over the next few quarters, such as in the integration of AI “copilots” in office productivity software, rising demand for big data analytics, and AI integration in image/video and other enterprise applications.

As a result, we believe software companies are well positioned to not only benefit from a potential expansion of the addressable market as AI-related software spending sharply rises across industries, including healthcare, media, autos, and other traditional industries, but also from repricing tailwinds and other cost efficiencies. We also see 15% of new applications globally to be automatically generated by AI without any human involvement by 2027 from 0% today. This, in our view, represents a large boost to the software industry as innovation and time to market are accelerated.

Our positive view on software is consistent with our tech playbook, which calls for a gradual rotation away from early cyclicals like semiconductors and hardware into mid-cyclicals. Beyond near-term opportunities in software, we also highlight internet as another medium-term beneficiary as AI becomes integrated in consumer applications like gaming, entertainment, and advertising.

What has AI’s contribution been to the year-to-date tech rally?
Globally, the five industries in tech—semiconductors, hardware (both early cyclicals), software, internet (both mid-cyclicals), and IT services (late cyclicals)—have a combined market cap of USD 26 trillion and account for roughly 32% of the MSCI AC World Index.

Year-to-date, the combined global tech market cap has climbed by USD 6 trillion. While rising interest in AI has been a key driver, other important factors are also at play, such as bottoming earnings revisions, market expectations of peaking rates, ongoing tech restructuring efforts, and solid share buybacks. However, based on our bottom-up calculations and looking at the most AI sensitive stocks and industries, we think AI-related stocks have contributed roughly USD 2 trillion to the combined market cap or about one-third of the global tech rally. While CIO doesn’t think AI is a bubble given clear use cases and solid long-term visibility, we have become more selective on AI-related stocks given the strong year-to-date returns and premium valuations, and recommend investors consider companies with clear monetization trends.

How can AI companies monetize?
Before we size up the global AI market, it is imperative to understand how AI companies can monetize their value/assets. There are two broad avenues for AI-related companies to generate revenues—through the infrastructure layer and the applications and data models layer.

The infrastructure segment mainly focuses on computing, networking, storage, and other related areas required to run huge datasets and AI-based applications. This segment is dominated by semiconductor and hardware companies, as well as infrastructure-based data center companies. The other way to monetize is through data models and applications. Companies can license their huge data sets or run big data analytics in the model layer and AI-specific applications in the application layer across the spectrum, from enterprise-based applications such as office productivity, customer resource management, front/back office software, etc. to consumer applications like advertising, gaming, etc.

Currently, we believe most of the AI-related spending is concentrated in the infrastructure layer given the need to build and train huge datasets. But with more signs of broadening demand, we expect the applications and models segment to emerge as the dominant force in the medium to long term.

For more, read the full report TechGPT: Raising AI growth estimates on broadening demand trends 24 July 2023.

Main contributors: Sundeep Gantori and Kevin Dennean