3 observations on artificial intelligence (AI) and a winning strategy for investors

This article was syndicated by Endowus in partnership with Thematics Asset Management, an affiliate of Natixis Investment Managers.

‍

If we were to envision the development of artificial intelligence (AI) as a blockbuster movie, how would it be portrayed? 

According to Natixis Investment Managers’  latest survey of institutional investors, most associate AI with a future that is depicted more accurately in the film Moneyball (50%), than in The Terminator (6%). Of course, this is not influenced by their preference for Brad Pitt over Arnold Schwarzenegger, as the respondents reckon that AI is “just a new tool for analysing data rather than a key opening door to a dystopian future.” Also, 35% of them draw parallels between AI and the Sci-fi thriller War Games, expressing concerns that hackers could set off geopolitical, economic, or social turmoil. 

After an exciting year for AI and peripheral themes in 2023, is there value left for investors to mine further? What would our experience in robotics and automation in the past decade tell us about the future of artificial intelligence? At this juncture, three observations come to mind:

1. The initial hype may subside before adoption starts to take hold
2. The most exciting breakthroughs sometimes come from the least obvious of applications
3. Big tech leads innovation for the near term, paving the way for the success of their small peers 

1. The initial hype may subside before adoption starts to take hold

While the long-term opportunities related to AI are significant and far-reaching, there are always risks that the hype associated with the underlying technologies subsides before those same technologies are widely adopted. 

The newsflow surrounding generative AI and large-language models in the first half of 2023 has been unrelenting and the speed at which the associated acronyms have become a mainstay in seemingly every corporate earnings call, no matter the industry or sector is astounding. 

The reality is that it will take time for upstream capital expenditure to be converted into durable company profits for a whole ecosystem. Downstream applications or adoptions of AI, even the first-movers, are not currently expected to contribute materially before the second half of 2024 or the first half of 2025 at the earliest, as CFO Amy Hood said of Microsoft’s AI services and co-pilots such as Azure AI. 

At Snowflake’s second-quarter earnings, Mike Scarpelli, the data cloud company’s CFO, told analysts at the second-quarter earnings call that “It’s going to take some time for AI [...] people are still struggling to get graphic processing units (GPUs) and there is a time lag between when a chip manufacturer sells the chips to [when] it gets built into the hardware that gets deployed on the rack, in a data centre and gets deployed to customers¹.”

This isn’t to deny the long-term potential of generative AI's conversational abilities as the technology democratizes future innovations and applications, allowing anyone with a computer to write code, develop apps, direct movies, or write novels. Technical hardware is no longer a significant barrier, as AI applications can run on existing devices through cloud hosting.

Excitements aside, in the world of investment, one must put today’s market environment and valuations into perspective, even as investors make bets on what could be tomorrow like, rather than what is today. That’s why, we are posing three questions for companies and investors like ourselves to reflect on:

1. Will the fundamentals ever be able to match their elevated valuations?
2. Once monetisation takes hold, will they be able to recoup their initial capital outlay and turn a profit?
3. Once the mainstream catches up and the use of AI becomes ubiquitous, how sustainable will those profits be?

2. The most exciting breakthroughs sometimes come from the least obvious

It’s often the most creative and fantastical applications of AI that make the headlines. To name a few, generative AI is used to produce personalised movies in which audiences can pick their own actors or even insert themselves into the storyline². Declining honey bee populations³ is worrying and the impact can be alleviated by AI. Robotic bees are tasked to support pollination and crop development to help stem biodiversity loss.

Behind the news headlines, AI is being put to work in an even wider range of impactful areas. Examples include drug development and chip design. 

In November 2020, Google DeepMind's AI technology, AlphaFold, successfully solved a long-standing biology problem. Building upon the foundational techniques used in AlphaGo, a bot renowned for its victory against the legendary Go player Lee Sedol in 2016, the AI model accurately predicts the 3D structure of complex protein folds. This breakthrough has promising implications for speeding up drug development, increasing success rates, and reducing costs in the field of biology.

Nvidia, AMD, and even Alphabet utilise AI in their processors, crucial for large language models and machine learning systems. In fact, the integration of AI streamlines chipmaking itself, reducing physical implementation and verification time from months to hours. Even better, AI models continuously learn from past designs, empowering engineers to rapidly simulate, develop, and optimise new chip architectures, enhancing overall efficiency.

3. Big tech leads short-term innovation, paving the way for small peers’ success

Experience teaches us that when a breakthrough technology emerges, a flurry of new start-ups, spin-offs or company pivots follow. 

Given the investment needed in infrastructure, it will likely take time before we see a wave of disruptive innovations emerge outside of the incumbent tech firms. Foundational models, those with hundreds of billions of parameters, require significant ‘compute’ power and typically cost over USD 50 million to build and train such models⁴.

Just as in the metaphor popularised by Isaac Newton: “If I have seen further, it is only by standing on the shoulders of giants.” Smaller and more nimble firms can leverage the work of established players and build smaller, specialised models. A reverse example of this is Nuance Communications. 

Since the late 90s, Nuance Communications has been at the forefront of voice recognition systems. Their early application, Dragon laid the foundation for the virtual assistants found in today’s smartphones and smart TVs. Nuance's speech recognition and text-to-speech technology find widespread use across various industries. It facilitates healthcare professionals in dictating patient notes, aids in fraud detection and prevention in finance, and enables personalised customer interactions through automated chatbots in retail.

In July 2022, Microsoft acquired Nuance Communications for US$17.2 billion⁵. This acquisition enabled Microsoft, which already has a stake in OpenAI, to merge the expertise of Nuance with OpenAI's GPT4 model and computing capabilities.

A development that transcends regions and sectors

Robotics and automation is not merely a lens for us to gauge what the future might hold for AI. Combining these two areas can present a novel opportunity set for investors. 

Targeting companies deeply involved in AI and robotics, the approach Thematics AI and Robotics Fund adopts is to balance the fields which are currently at different stages of their development and adoption and with different cyclicality and valuations. For investors, the combined approach builds on a myriad of mutual and self-reinforcing benefits. 

AI can be a theme in itself but the adoption of AI can be captured by other themes. For example, the umbrella of the subscription economy is a theme that the Thematics Meta Fund is capitalising on. Sprawling across software, media, and data, the subscription economy is set to be enhanced and optimised by a more mature use of artificial intelligence. 

Opportunities abound but we should never overlook risks and Safety, another theme in the Meta Fund, is tied to the larger story as well. The development of artificial intelligence also opened a “Pandora’s box” of cybersecurity. Wide adoption of the technology unlocks new challenges as it facilitates the deployment of highly effective phishing scams, reduces attack costs, and empowers agile hackers. The persistent demand for solutions in response to these threats presents a multitude of investment opportunities. 

Conclusion: Active management and valuation discipline are essential

As innovations akin to ChatGPT roll out and grab the attention of the general public, we expect to see flare-ups of enthusiasm as well as occasional periods of slow progress or roadblocks. All in all, this field has shown its exponential nature of advancements since the 1950s, when American engineer Claude Shannon created the maze-solving robotic mouse Theseus, the world’s first shown AI system. 

In a very dynamic and rapidly evolving investment opportunity set, investors have the challenge and the opportunity to identify the long-term winners early on. More importantly, though, we should avoid “over-hyped” areas of the market. These typically are materially detrimental to performance. 

Indeed, we believe that the true value lies in the enablers of artificial intelligence rather than the adopters. Navigating through relevant themes and the fundamental side of AI requires robust portfolio management, an active mindset, as well as valuation discipline. 

The winning strategies of tomorrow could be one that maintains consistent and long-term exposure to the most promising opportunities across various markets while adopting a strict valuation discipline to ride through numerous inevitable ups and downs along the journey.

Thematics Meta Fund: An active strategy of five themes

Thematics Meta Fund offers diversified exposure to multiple themes that have the potential to grow at a rate superior to that of the broader global economy, due to the long-term secular growth drivers that underpin them.

The fund invests in five thematic strategies, namely AI & Robotics, Safety, Subscription Economy, Water, and Wellness on an equal weight basis. The portfolio is rebalanced on a monthly basis. 

As the fund identifies new themes over time, they may get incorporated into the Fund gradually. The fund is also not constrained to a specific region, allowing investors to gain exposure to equity markets on a global basis.

Thematics Asset Management (an affiliate of Natixis Investment Managers) is a dedicated equity manager in innovative thematic strategy. Its overarching approach to theme definition is clear with a strong investment intuition – they start by understanding the primary forces behind different secular growth trends, then identify the most attractive opportunities where multiple forces converge and reinforce one another to give rise to a theme. The themes they identify should provide the opportunity to earn a sustainable long-term premium over the market.

Endowus has four funds from Thematics Asset Management, an affiliate of Natixis Investment Managers (as of 15 February 2024), including the Thematics Meta Fund and the Thematics AI and Robotics Fund.

Get started building your own portfolio with these funds on the Endowus Fund Smart platform. 

<divider><divider>

1 Source: Microsoft & Snowflake corporate earnings calls, July 2023

2 Source: Could AI replace Hollywood with personalized movies?, Forbes, April 2019 

3 Source: Autonomous drones based on bees use AI to work together, University of Maryland, December 2022

4 Source: Thematics AM, August 2023

5 Source: Microsoft in Talks to Buy AI Firm Nuance Communications, Bloomberg News, April 2021

<divider><divider>