The Artificial Intelligence gold rush

Fascinated by the prospect of a new industrial revolution, billions of pounds of investments have flowed from the coffers of companies and investors eager to secure a front-row seat. This optimism has been reflected in the markets, which have seen one of the best years in recent memory. It’s one of the greatest gold rushes in the history of capitalism. Estimating the total investment already made – and yet to come – is difficult. The final tally will be measured in trillions of dollars.

From an investment standpoint, the key question for the future of market performance lies here: to justify such substantial investment, AI must fulfill two promises. The first is that progressive integration into business systems will generate increased productivity. The second is that the infrastructures being built will facilitate the development of innovative products and services that find applications across diverse industries.

The fulfillment of these promises will be measured in corporate profits. Investors will increasingly focus on financial reports, hoping to see the start of a long wave of innovation that could last for several decades. For those who have already invested, it will be important to see positive signs within relatively short periods – a confirmation that AI is the revolutionary technology it’s perceived to be, and not one of the greatest bubbles in the history of financial markets.

The AI boom – which has driven stock markets – has so far been led by specialized companies – like Nvidia – and tech giants developing language models. To see valuations confirmed, we need to see the benefits extend to other companies and industries capable of leveraging this technology to create innovative and market-dominating products.

Investment trends

But which sectors are driving AI investments? And what are the goals of these investments?

For now, the big numbers primarily come from tech giants, who are heavily betting on the future of AI, supported by the vast cash reserves they’ve accumulated over the years. Forty percent of R&D investments – which reflect an organization’s willingness to invest in discovery and commercialization of new technologies – in the S&P 500 come from the 10 largest companies of the index.

AI functions through large language models (LLMs), which are at the core of various technological applications across different sectors. These models are extremely expensive and complex to develop. The major tech companies are locked in a race to release next-generation models to establish leadership in this emerging market. The problem is that next-gen systems require exponentially higher investments to train, potentially limiting returns.

It’s estimated that the electricity used to train GPT-4 was 50 times higher than that used to train GPT-3. This particularly draws out the challenge involving energy costs, which could become increasingly burdensome as the technology evolves and its use becomes more widespread. Another issue is data availability. Paradoxically, while data is abundant, continued improvement in model performance depends on high-quality data, which could become increasingly difficult to acquire. The ability to access quality sources will become a critical factor and could negatively impact return on investment. 

Currently, the race to develop advanced models seems limited – excluding Chinese models – to four companies: OpenAI (supported by Microsoft), Meta, Google, and Anthropic (supported by Amazon). Among the tech giants, Apple appears to be lagging, evidenced by its recent announcement of a partnership to integrate ChatGPT into the latest version of its iPhone operating system, iOS 18.1. Such a move highlights the level of concern, even for a company as large as Apple, about being left out of this wave of innovation. 

The race to build large AI models is increasing the concentration of market capitalization in stock indices. The seven largest American tech companies now represent nearly 35% of the S&P 500’s market capitalization and have contributed to over 70% of its returns since the beginning of 2023. 

This outperformance has also led to an expansion in valuations. While the rest of the S&P 500 traded at a 12-month forward price-to-earnings (P/E) ratio of 19x at the end of November, the top 10 stocks in the index were trading at 29x,according to estimates from Goldman Sachs. If skepticism about the future applications of AI were to prevail, there is a risk that we could see these valuations normalize in the medium term.

To justify these valuations, it will be crucial to begin observing profit growth. Unlike past technology races, the tech giants are in a much stronger position to win this bet. The earnings growth trend observed in recent years makes the investment more sustainable, as reflected in the high but not unmanageable P/E ratios.

The second investment front: infrastructure

Another key area of investment is acquiring the technology needed to power these systems. Running this technology requires data centers and cloud connectivity. The demand generated by AI is rapidly consuming existing capacity, pushing companies to construct new facilities, which also creates potential investment opportunities.

The development of LLMs is extremely computationally intensive. These processes require semiconductors, known as graphics processing units (GPUs). A decade of significant progress in GPUs has resulted in faster and more efficient performance. This advancement has rendered most computational electronics built before 2020 obsolete, necessitating widespread upgrades.

In this context, Nvidia has been the immediate winner, and its financial results demonstrate the scale of investments in AI. Nvidia’s annual revenues have grown from $4 billion in 2014 to a projected $61 billion in 2024. The extraordinary demand for Nvidia’s AI processors and related products and services highlights the robust health of the AI sector and its growth potential. Nvidia estimates the total demand for GPUs could reach $2 trillion, including $1 trillion from data centers and $1 trillion from AI-related tasks such as training new LLMs, machine learning, and scientific simulations.

The third investment front: the “killer application”

The third critical area of investment is the development of the so-called “killer application” – a practical use of AI technology capable of driving the market. So far, there have been no significant cases of startups or companies delivering AI products or services that revolutionize their respective markets. As of now, only 5% of companies report having integrated AI into their products or services.

However, AI is driving investment in the startup sector as well. According to data from Crunchbase, 35% of startup funding this year has gone to companies operating in artificial intelligence. While it’s too early to determine which of these companies will succeed, it’s likely that the next cycle of AI-driven innovation will come from industries such as pharmaceuticals, telecommunications, or robotics.

By 2025, we may begin to see AI everywhere. If that happens, the technology will have begun to deliver on its promise.

A revolution in productivity

The success of AI will also be tied to its ability to revolutionize production processes. Currently, companies are still figuring out how to properly integrate artificial intelligence into their operations. Only slightly over 5% of companies claim to use AI to deliver their products or services, although this figure is rapidly increasing.

Meanwhile, many workers already incorporate AI into their daily routines. A recent study measured the productivity gains AI could bring across three tasks: customer support, drafting simple business documents, and coding. Assigning these tasks to workers using AI resulted in an average efficiency gain of 66%, sometimes accompanied by improvements in quality.

Interestingly, productivity gains were lower in customer support activities but significantly higher in more creative tasks: 59% for drafting business documents and an impressive 126% for coding.

Considering the potential evolution of AI models, particularly with the introduction of agent-based systems, these figures are incredibly promising. To provide some perspective, annual productivity growth in Europe averaged just 0.8% in the years leading up to the pandemic. While it’s not entirely appropriate to compare experimental results to general averages, it’s hard to imagine that a technology capable of boosting worker capacity to this degree wouldn’t have a significant impact on overall productivity.

The downside, however, could be the displacement of human labor. Increased productivity might lead to higher corporate profits, either by reducing costs – hiring fewer workers – or enabling the production of more or better products.

Long-term market impact

The impact of productivity growth on corporate earnings will likely become a key market trend, potentially driving profits higher in the coming years. For now, it’s still too early to detect this effect in business processes. However, this shift is underway beneath the surface, and we’ll likely begin seeing its effects soon.

According to McKinsey’s Global AI Survey, the percentage of companies that have adopted AI in at least one business function rose from 55% in 2023 to 72% in 2024. This suggests that we are just beginning to see the transformative power of AI unfold.

Support for market performance

As managers, our role is to analyze economic phenomena through an analytical lens, always aiming to highlight potential contradictions and risks. While we recognize that the trajectory might be less linear than initially expected, the opportunity generated by AI appears undeniable. We believe it will be a trend that extends over the years, and we are merely at the beginning of a potentially revolutionary dynamic.

It is also important to remember that this will be a global phenomenon. In this article, we have focused primarily on the United States because it is at the forefront of developing this technology. However, China is also advancing its AI models, and the applied use of artificial intelligence, along with productivity gains, will not be confined within national borders.

As technology advances – with the anticipated arrival of “agent” systems in 2025 capable of planning and executing more complex tasks – adoption could accelerate. We believe that artificial intelligence has the potential to become one of the greatest growth accelerators in economic history, representing a significant opportunity for investors in the coming years. It will be crucial to assess its dynamics to understand how to help investors capitalize on returns while mitigating risks.

Our Strategic Asset Allocation for 2025

We think that AI represents a significant opportunity to improve global productivity, even if there are challenges in a range of areas like energy consumption and regulation. It will probably take longer to feel the true impact of AI than the most optimistic forecasts, as has often been the way with new technologies, but it could have a meaningful impact on growth. In that case, we think that the baseline scenario that we’ve used for the SAA could prove conservative. 

As we do every year, we’ve been working on ­our Strategic Asset Allocation, ­in which we­ detailed our comprehensive, long-term view across a wide spectrum of asset classes and the strategic positioning of our portfolios. The Strategic Asset Allocation process provides an important reminder of some of the core tenets of Moneyfarm – focus on the long term, keep your costs and turnover low, and don’t get too distracted by the daily news flow.