by Justin Leverenz, CFA, Chief Investment Officer, Developing Markets Equities, Invesco
Key takeaways
- An evolving story. - We’re seeing more nuanced debates about the downstream impact of AI, which may be important in discerning the long-term winners in this space.
- The EM connection. - A handful of emerging market companies and their underlying semiconductor technologies have been crucial to enabling generative AI solutions.
- What lies ahead? The EM world is also ripe for new waves of hardware innovation and chip design prowess that dovetail into AI products.
Emerging market (EM) tech firms play a key role in the AI story
Season One of the artificial intelligence (AI) saga may be remembered for a rather uncomplicated storyline: Markets quickly rewarded semiconductor and hardware companies for powering AI (some deservedly, some not), particularly as the AI wave happily coincided with a cyclical bottoming of inventories in the tech supply chain. We believe Season Two is expected to feature more nuanced debates about the downstream impact of AI, which may be important in discerning the long-term winners in this space. As emerging market investors, we’re following this story closely as a handful of EM companies seek to expand their competitive edge in enabling generative AI solutions.
AI storyline #1: Enterprise versus Consumer
The use of ChatGPT has become ubiquitous, Microsoft1 has raced ahead with tools such as its Copilot “AI companion,” and IT service firms claim a rising number of AI projects. One might think businesses should see immediate value in AI adoption, but the enterprise world is so beset with disparate data, privacy and regulatory constraints, and a risk-averse outlook to seismic change, that sizeable efficiency gains may be hard to achieve.
Past technology revolutions have been driven by eager consumers, but here too, we need to see truly useful AI applications that can function on our devices. Smartphones may make a comeback if compelling use cases drive a global replacement cycle. We are seeing PC makers call some of their products “AI PCs” – do we need a new PC cycle?
AI storyline #2: Offense versus Defense
In the fiscal year ended January 2023, Nvidia1 reported some $15 billion in datacenter revenues, which is where its AI business is housed.2 For the year ending January 2025, Nvidia's customers are expected to spend close to $100 billion.3 On AI chips. What will the return on investment look like as these spending plans seek revenue tailwinds?
- In our opinion, if we assume revenues from search and social media are AI-enabled, then the capex is merely defensive and may allow existing platform giants such as Meta and Google1 to continue their dominance.
- However, in our opinion, to sustain such elevated spending, companies need to go on offense creating new revenue streams. These could be through material cost reductions, such as the automation of customer service. Customized software development might accelerate, impacting Software as a Service (SaaS) companies, or accelerated medical drug discovery might transform health care.
The space is evolving rapidly, and moats may not be as wide as they appear. Therefore, optimism about future possibilities needs to be tempered with the risk of disruption. Amid these possibilities, it is clear we are heading into the great unknown.
Emerging market companies have played a foundational role in AI enablement
In the EM world, a handful of companies and their underlying semiconductor technologies have been crucial to enabling generative AI solutions, including Nvidia’s datacenter graphics processing units (GPUs).
- Taiwan Semiconductor Manufacturing Corp1 looms over all things cutting edge in global semiconductors. With generative AI, its competitive edge has now expanded beyond wafer production to include advanced packaging that effectively “stiches” together the memory and graphics processor chips in a way that helps to achieve a generational leap in performance while minimizing power consumption.
- SK Hynix and Samsung Electronics1 have helped pioneer the market for the high bandwidth memory that vastly improves memory throughputs for GPUs. It is perhaps the singular most significant semiconductor technology in enabling the acceleration of GPU’s massive parallel computing performance.
We believe the EM world is ripe with new waves of hardware innovation and design prowess that dovetail into AI products. These spaces are more complicated, but we need to keep an eye on the evolution of smartphone-related semiconductors, where MediaTek1 operates, or specific original design manufacturers who might win new business from the chip design efforts of large platforms, including the likes of Google, Meta and Amazon.1
A future for the curious
AI’s first season ran on the theme of “see it, believe it.” Season Two is expected to be more layered and may result in wildly different outcomes for stock prices. In the face of all the hyperbole, Invesco Developing Markets Fund remains a long-term investor in highly differentiated businesses that have durability of growth, sustainable competitive advantage, strong governance (and thereby capital allocation) and a host of real options embedded in their franchises that we believe will emerge over many years.
Footnotes
1 As of March 31, 2024, Microsoft, Meta, Google, Nvidia and Amazon, represented 0% of the Invesco Developing Markets Fund. Taiwan Semiconductor manufacturing Corp., Samsung Electronics, SK Hynix and Mediatek represented 9.3%, 6.4%, 0.8% and 0.2%, respectively.
2 Source: Nvidia Annual Report March 2023
3 Source to Bernstein Research May 2024
Copyright © Invesco
