Andrew Pitt: Intelligent investment in artificial intelligence?

07 Sep 2023 Expert insight

Head of charities at Rathbones Group, Andrew Pitt, looks at whether charity trustees should consider investing in artificial intelligence…

By sdecoret / Adobe


This content has been supplied by a commercial partner.


Trustees seeking prudent and long-term investment approaches for their charities do not generally rely on gold rushes – either real or metaphorical.

However, the November 2022 launch of ChatGPT has triggered a new artificial intelligence (AI) gold rush among investors, who have rapidly bidden up share prices across the technology sector. Amid such frenzy, it’s far from clear how many valuable nuggets will be left in the end in investors’ prospecting pans.

Machine learning isn’t a new phenomenon. At its heart, AI applies statistical analysis to large data sets. For instance, Amazon uses it for product recommendations. What’s new about this craze is that it’s generative AI. This subset of machine learning is different because instead of just trying to spot patterns in data, its algorithms appear to generate original content. 

Purple cats on the moon

Take pictures of cats, for example. Normal machine learning can identify a cat after scanning millions of cat images. But generative AI programmes can create entirely new images, such as a purple cat on the moon – should any animal welfare charity have a pressing need for one. 

This machine learning is typically “trained” on graphics processing unit (GPU) chips. These are almost all designed by Nvidia, a US company. They are mostly manufactured in Taiwan by TSMC, using equipment provided by ASML, a Dutch business. Large cloud services providers, including Microsoft’s Azure and Amazon’s AWS, are buying the chips and renting the resulting AI computing capacity to software developers and businesses to train and run their own AI programmes and models.

The value in the industry right now is almost entirely accruing to Nvidia. This became apparent when the company reported its blowout second-quarter results. Revenue doubled compared with the same period last year, on soaring demand for the GPU chips needed for generative AI training.

Monopolies in the cloud

The question is whether the firm can maintain this monopoly position. The cloud service providers – its biggest customers – have a huge economic incentive to develop an in-house solution or cultivate an alternative supplier.

It’s not clear yet how generative AI will generate profits for the rest of the tech sector. Practical applications are already emerging. For example, financial technology platform Intuit is using linguistic versions of generative AI, called large language models (LLMs), to help solve complex tax queries. But generative AI could turn out to be a feature that businesses need to incorporate in order to avoid obsolescence rather than make bigger profits. For instance, Adobe has launched Firefly to generate art through generative AI. If it hadn’t done this, a start-up could have offered this service and disrupted its business. 

Some companies will certainly get disrupted by generative AI. The online education assistant Chegg has already blamed poor corporate results on students using ChatGPT. But identifying the losers isn’t straightforward. 

Falling back down to earth

Markets are fickle, as charity trustees, who invest for the long term, know. We suspect many stocks that have seen their share prices surge in recent months will fall back down to earth as their earnings reports reveal a lack of boost to their revenues and profits from this technology. 

For instance, IT consultancy Accenture’s recent Q3 results showed that AI would have a negligible impact on revenues in the short term. 

All things considered, the potential for this technology is exciting, but it’s uncertain how it will develop. At this stage, we think it makes sense to evaluate businesses on their current earnings power rather than future visions of what generative AI might deliver.


More on