Enabling technologies core to the AI revolution

The advent of artificial intelligence (AI) ushers in an era of potentially radical innovation and disruption across the global economy. Transformational change in some sectors looks guaranteed.

We perceive a range of investment opportunities presented by the rising adoption of AI. This technological revolution promises to deliver efficiencies and innovation across a range of sectors, helping to solve some of society’s most pressing long-term challenges – so long as the challenges posed by AI itself are managed.

Companies around the world are experimenting with the application of AI to their businesses. It can be difficult to identify those that are successfully adding value by doing so, however.

In the nearer term, we believe it makes sense to take a close look at companies whose products and services play key enabling roles in the AI revolution, for whom the explosion in technological adoption creates compelling opportunities.

AI’s potential for transformation

The ability of AI models to process and analyse huge volumes of data, and spot patterns and trends to create new insights, means that they can conceivably play important roles across most economic sectors. AI models can unlock dramatic productivity improvements, produce more accurate forecasts and provide intelligent inputs into decision-making. 

The economic impact of AI looks set to be nothing short of dramatic. McKinsey estimates that AI could contribute up to US$22tn to the global economy once potential use cases of generative AI –models capable of generating text, images, videos and other new data – are factored in.¹ Almost two in five jobs worldwide are exposed to potential uses of AI, according to the International Monetary Fund (IMF), in ways that either complement or threaten workers.²

The applications of AI are not abstract: it is already having an impact across the economy. While the public imagination has been captured by intelligent chatbots powered by large language models like ChatGPT, companies are demonstrating that the technology can deliver real-world efficiencies and innovations.

For example, connecting heating technologies with AI software can optimise temperature control and energy consumption: a pilot project in two Swedish apartment blocks reduced energy use by 20%.³ Within healthcare, meanwhile, AI-led analysis is improving the accuracy of cancer diagnosis and predicting different forms of cancer using genetic data and samples.⁴

Source: Stanford Institute for Human-Centered Artificial Intelligence, 2024

Hardware and software underpinning the AI boom

Soaring interest in the possibilities of AI has been accompanied by an upward trend in investments. So far, much of this has focused on model development. Investment in generative AI models rose eight-fold in 2023 to US$25bn, for instance.⁵

By 2025, Goldman Sachs estimates that global AI-related expenditure could reach US$200bn, including in the infrastructure and software needed to run AI applications. As more companies and organisations adopt AI, the bank estimates that AI-related investment could eventually peak at more than 2% of US GDP in the 2030s.⁶

This investment boom will support demand for the products and services that enable the AI revolution. We perceive four distinct pockets of opportunity for AI ‘picks and shovels’ companies today.

1. Chip makers

Training and running the latest AI models requires vast amounts of computational power. They rely on the latest ‘leading-edge’ semiconductors for speed and energy efficiency.

Graphics processing units (GPUs) like those designed by US company Nvidia are particularly well suited to undertaking the parallel repetitive calculations that underpin AI models. The main clients for Nvidia’s latest Blackwell chip – the world’s most powerful – include cloud service providers that support and run advanced generative AI models.⁷

Nvidia is a ‘fabless’ chip manufacturer in that it designs and sells its products but doesn’t actually fabricate or/and assemble them; production is outsourced. Peers like AMD and Qualcomm, which are both leaders in smartphone chips, are ‘fabless’ too.

AI on the edge (also known as edge AI) refers to the deployment of AI processing directly on devices rather than relying on cloud computing. This has several advantages, including lower latency (crucial for applications like augmented reality and intelligent camera systems that require real-time responses), enhanced privacy (versus cloud models that transmit sensitive information over the internet), and reduced pressure on the grid.

Taiwan-based MediaTek produces edge AI, leveraging its expertise in efficiency (crucial for smartphones) to develop chips for data centres. Focused on Application-Specific Integrated Circuits, MediaTek’s application of edge AI offers superior performance and energy efficiency versus general-purpose models. These solutions can help limit energy consumption for specific verticals, thereby working toward addressing the technology’s energy intensity. Only a few producers are capable of making the complex leading-edge chips destined for use in AI models. Taiwan Semiconductor Manufacturing Company (TSMC) accounted for 28% of the overall global semiconductor market in 2023, excluding memory chips, with most revenues coming from chips of 7 nanometres (nm) or smaller.⁸ It became the first foundry to mass-produce 3nm chips in 2022 and is expected to start production of 2nm next year.⁹

2. Chip-making equipment manufacturers

Taking a step up the supply chain are the manufacturers of specialist equipment upon which the global semiconductor industry depends.

Arguably the most critical supplier is ASML, the Dutch-listed company whose equipment is needed to create the most advanced chips made by the likes of TSMC. ASML’s lithography machines use extreme ultraviolet light to etch atomic-level circuits onto fine silicon wafers, enabling ever smaller transistor designs (which in turn enable chips to become ever more powerful).

Suppliers of other critical niche components are similarly exposed to growing demand for computational power. Among them is Applied Materials, a leader in low-resistance chip wiring products. The most advanced logic chips can contain billions of transistors connected by as much as 60 miles of microscopic copper wiring.¹⁰

3. Chip designers

Enabling the extraordinary complexity of cutting-edge chips are sophisticated designs developed using software that itself increasingly leverages AI.

Leaders in the electronic design automation software that provides instructions for chip designs include US companies Cadence and Synopsys. These software providers perform a key role in the emerging AI ecosystem.

Using their products, chip producers can now apply generative AI to accelerate and optimise the design process, and to create more customised chips for bespoke applications. For instance, Cadence’s generative AI solutions have been applied in the design of innovative biomedical devices and systems.¹¹

4. Cloud computing infrastructure

AI models require enormous processing power to run the computations involved. It is estimated that a ChatGPT query uses nearly 10 times as much electricity to process, on average, as a Google web search.¹²

This is fuelling rising demand for the services provided by operators of hyperscale data centres and cloud computing providers such as Microsoft, Oracle and Alphabet (the parent of Google). It is estimated that data centre workloads, as measured by compute instances, more than doubled between 2019 and 2023.¹³ The advent of AI has been a major contributor and is expected to drive an acceleration in demand growth.

The likes of Microsoft and Alphabet are leveraging the computing power of their data centre networks to develop their own in-house AI models. Over time, we expect that their investments in general large-language models, as well as specialist ‘small-language’ models designed for specific corporate clients, could create long-term drivers of growth.

Source: Goldman Sachs, May 2024 based on data from Masenet et al. (2020), Cisco, IEA and Goldman Sachs research. Data centre power demand for 2023 is estimated. 

Navigating threats to the revolution

For the potential of AI technologies to be realised – and for the opportunities arising from them to be maximised – the environmental and social challenges that AI’s adoption creates must be addressed.

First, the implications of AI’s soaring energy usage must be managed. The International Energy Agency (IEA) forecasts that energy demand from data centres could more than double between 2022 and 2026.¹⁴

The geographical concentration of data centres is straining local grids, so planning constraints could undermine the build-out of digital infrastructure. Governments and regulators can manage this risk. In Ireland, where data centres consumed 17% of electricity in 2022 – a share that the IEA expects will double by 2026 – the utilities regulator placed conditions – but not a moratorium – on new data centres in a bid to balance sustained growth of the sector with security of electricity supply.¹⁵

We will examine the environmental implications of the AI boom – including how the energy consumption challenge can be managed and the prospective environmental solutions that applications of AI could deliver – in an upcoming article.

Second, the potential societal risks arising from AI’s applications must also be addressed. Since AI models learn from existing sources, their adoption risks entrenching existing social biases, such as racism and sexism, and supercharging the spread of disinformation. AI also threatens to lead to lower labour demand and redundancies, potentially worsening inequality.¹⁶ However, this is balanced by the fact that historically, new technology has tended to create more jobs than it destroys, and we believe AI can help address talent shortages in certain industries.^{17, 18}

As with any disruptive technology, concerns over the social impacts of rapid AI uptake could trigger heavy-handed government interventions, with the unintended consequence of undermining its prospective economic benefits. We expect that AI will continue to impact industries rapidly and believe governments can work to ensure a just transition.

Additional risks to the AI revolution come in the form of geopolitical tensions and supply chain bottlenecks. Security concerns have driven the US, Japan and the Netherlands to tighten export controls on advanced semiconductor technologies to China, including manufacturing equipment. Meanwhile, a highly concentrated supply chain for cutting-edge chips has the potential to lead to acute disruption, with cascading effects across global production lines.¹⁹ Taiwan produces over 90% of the most advanced chips.²⁰

Focusing on key enabling technologies

Awareness of these challenges and the unclear direction of technological innovations dictates that investors should be cautious about some of the more breathless predictions of AI’s potential.

Irrespective of how and where it is applied to greatest effect, however, advanced semiconductors and the technologies needed to design and manufacture them will be essential ingredients. It is our conviction that the leaders in these niches underpinning the AI ecosystem will help enable and power a revolution that could fundamentally transform whole sectors of the modern economy.

1. McKinsey, June 2023: The economic potential of generative AI: The next productivity frontier ↩︎
2. IMF, January 2024: AI Will Transform the Global Economy. Let’s Make Sure It Benefits Humanity ↩︎
3. Vattenfall, October 2023: Yes, AI will improve energy efficiency ↩︎
4. Zhang, B., Shi, H., and Wang, H., 2023: Machine Learning and AI in Cancer Prognosis, Prediction, and Treatment Selection: A Critical Approach. Journal of Multidisciplinary Healthcare
   The Institute of Cancer Research, 2023: Harnessing the power of AI to improve outcomes for people with breast cancer ↩︎
5. Stanford Institute for Human-Centered Artificial Intelligence, 2024: AI Index Annual Report ↩︎
6. Goldman Sachs, August 2023: AI investment forecast to approach $200 billion globally by 2025 ↩︎
7. Nvidia, 18 March 2024: NVIDIA Blackwell Platform Arrives to Power a New Era of Computing ↩︎
8. TSMC, 2024: TSMC 2023 Annual Report ↩︎
9. TSMC, 2024: Logic Technology ↩︎
10. Applied Materials, July 2024: Applied Materials Unveils Chip Wiring Innovations for More Energy-Efficient Computing ↩︎
11. Cadence, 2024: Unleashing the Power of Generative AI in Chip, System, and Product Design ↩︎
12. Goldman Sachs, May 2024: AI is poised to drive 160% increase in data center power demand ↩︎
13. Goldman Sachs, May 2024: AI is poised to drive 160% increase in data center power demand
    ↩︎
14. IEA, 2024: Electricity analysis and forecast to 2026 ↩︎
15. Commission for Regulation of Utilities, November 2021: CRU Direction to the System Operators related to Data Centre grid connection processing ↩︎
16. Cazzaniga, M. et al, January 2024: Gen-AI: Artificial Intelligence and the future of work, International Monetary Fund ↩︎
17. Fairbanks, A. and French, C., October 2023: Opportunities in advancing more inclusive careers ↩︎
18. Manyika, J. et al, November 2017: Jobs lost, jobs gained: What the future of work will mean for jobs, skills, and wages, McKinsey Global Institute ↩︎
19. Hille, K. & Liu, Q., 23 August 2023:  Supply chain shortages delay tech sector’s AI bonanza. Financial Times ↩︎
20. Economist, 6 March 2023: Taiwan’s dominance of the chip industry makes it more important ↩︎

References to specific securities are for illustrative purposes only and should not be considered as a recommendation to buy or sell. Nothing presented herein is intended to constitute investment advice and no investment decision should be made solely based on this information. Nothing presented should be construed as a recommendation to purchase or sell a particular type of security or follow any investment technique or strategy. Information presented herein reflects Impax Asset Management’s views at a particular time. Such views are subject to change at any point and Impax Asset Management shall not be obligated to provide any notice. Any forward-looking statements or forecasts are based on assumptions and actual results are expected to vary. While Impax Asset Management has used reasonable efforts to obtain information from reliable sources, we make no representations or warranties as to the accuracy, reliability or completeness of third-party information presented herein. No guarantee of investment performance is being provided and no inference to the contrary should be made.