Nvidia has been the poster child for the Artificial Intelligence (AI) revolution; however, the story goes far beyond just one company.
Hyperscale cloud providers are investing heavily into graphics processing units (GPUs) to support AI tasks and, as the market leader, Nvidia has been the biggest beneficiary. This investment is capital expenditure, which gets capitalised on the balance sheet rather than expensed on the profit and loss accounts.
For Nvidia, this is recognised as revenue and immediately flows through to earnings-per-share growth, propelling it to one of largest companies in the world.
Data centre demand
In a recent piece titled ‘The Golden Opportunity for American AI’, Microsoft noted that it believes AI “is the electricity of our age” and is on track to invest around $80 billion to build out AI-enabled data centres.
We are currently in the midst of an investment super-cycle to meet the demand for computing power. This is similar to the Cloud investment cycle, of which Amazon was a pioneer. This involved a lot of investment, and was initially met with scepticism, but the Cloud transition has been a huge success as it has allowed companies to reduce costs associated with maintaining technology hardware.
Nvidia chips are an important part of the story, but there is a substantial data centre upgrade and build-out happening. Data centres are incredibly complex buildings requiring physical infrastructure, climate control, security, and crucially a lot of reliable power. These projects involve multiple companies, particularly in the industrials and utilities sectors.
The US Department of Energy estimates that electricity demand from data centres will triple over the next three years.
According to consultancy Employ America, software, technology hardware, industrial equipment and data centre structures currently account for almost 6% of the US economy. It notes that, by the end of 2025, this share is likely to surpass the tech-industrial-telecom boom of the late 1990s, which reached its peak back in 2000. It may even surpass the 7% share that housing investment represented at its 2005 peak.
The US has led the way with the data centre buildout, but Bernstein Research estimates that the annual data centre spend growth rate in Europe through 2030 may be 20%.
The UK economy has suffered from a lack of economic growth, but the government recently unveiled an action plan focused on developing and deploying AI. This will create “AI growth zones” which include a streamlined planning process for data centres and power.
AI will impact every sector of the economy. There is an interesting bifurcation where AI is being deployed by very large companies with the budgets and resources, but also venture capital where new businesses are getting funded with the aim of creating the winners of tomorrow.
Small and medium sized businesses will follow, but are more likely to wait for proof of concept and then utilise software as a service offerings.
Walmart has over 850 million pieces of data across its product catalogue, so AI has been used to update product listings. Its CEO, Doug McMillon, CEO noted “without the use of generative AI, this work would have required nearly 100 times the current headcount to complete in the same amount of time.”
The CEO of Goldman Sachs, David Solomon, recently noted that 95% of an initial registration prospectus for an IPO can be written in minutes by AI, whereas it might have taken a six-person team two weeks to complete.
Hype cycle
New technologies are often subject to a hype cycle which goes too far and leads to a “trough of disillusionment”. This is a risk with investing in AI, but the proliferation of real-world use cases is remarkable.
Copyright is an issue for the industry and many AI businesses are currently the subject of litigation.
Mistakes are also a challenge; for example, McDonald’s has recently pulled back from AI ordering technology, with reports suggesting that it confused ‘Mountain Dew’ with ‘medium Coke’.
The AI buildout has supported US growth and the equity market. The dot-com bubble of the late-90s was a product of irrational exuberance over the internet buildout. There are some similarities to that period, but valuations are lower. The ability of the hyperscalers to convert capital expenditures to revenue is key, as depreciation expense rises.
Signs are promising, as demand for computing power is so strong. ChatGPT charges $200 per month for its pro subscription, but OpenAI CEO Sam Altman recently noted that “we are currently losing money … [as] people use it much more than we expected”.
A lot of money will be lost in venture capital, but that is the nature of the industry, and there are also likely to be some spectacular success stories.
In public markets, the broadening of the AI theme creates opportunities outside of technology and ultimately in companies across the economy with the ability to harness the benefits of technological progress.
Disclaimer: The views expressed are the opinions of the writer and whilst believed reliable may differ from the views of Butterfield Bank (Cayman) Limited. The Bank accepts no liability for errors or actions taken on the basis of this information.
Nicholas Rilley, CFA, is an Investment Manager and Strategy Analyst with Butterfield Butterfield Asset Management.
Related Videos
