Investment: What’s next in artificial intelligence?

Few people now argue with the idea that artificial intelligence (AI) will change the world – but how it will evolve is still a topic of debate. The early focus has been on the build-out of AI – developing large language models, cloud migration, data assimilation, storage and analysis, but where could the next wave of beneficiaries be found?

That AI will have a long-term impact is largely beyond dispute. Bill Gates says: “The development of AI is as fundamental as the creation of the microprocessor, the personal computer, the Internet and the mobile phone. It will change the way people work, learn, travel, get healthcare, and communicate with each other. Entire industries will reorient around it. Businesses will distinguish themselves by how well they use it.”

This process has already started. In an international online survey conducted by McKinsey in early 2024, 65% of respondents said their companies were already using artificial intelligence in some areas of their business, double the proportion from a year earlier. Participants also reported that their organisations were already seeing material benefits from the use of generative AI, through lower costs and increased revenue.

Some sectors have been swifter to exploit AI than others, with the strongest adoption occurring in professional services. Across all organisations, the primary areas of use for AI has been in marketing and sales functions, and in product and service development, according to McKinsey.

In the longer term, AI will need to offer more if it is to deliver sufficient productivity gains to justify the investment.

However, in the longer term, AI will need to offer more if it is to deliver sufficient productivity gains to justify the investment. Most organisations believe these gains will eventually arrive, with three-quarters of the McKinsey survey respondents predicting that generative AI will lead to significant or disruptive change in their industries in the years ahead.

After Nvidia

A study by Goldman Sachs Research examining the implications of AI for investment themes argues that after the initial surge of money into the sector, which has principally benefitted US-based AI chipmakers, the next phase for investors to follow will involve companies that are helping to build AI-related infrastructure.

These might include companies that are part of the supply chain elsewhere in the world, such as semiconductor providers in Europe and Asia, that so far have not enjoyed the same attention as US businesses. The authors say: “Phase 3 deals with companies incorporating AI into their products to boost revenue, while phase 4 is about AI-related productivity gains that should be possible across many businesses.”

The market environment has been complicated by the abrupt arrival of Chinese challenger DeepSeek, which claims to have developed its AI model at far lower cost than US competitors such as OpenAI. If AI models can be built with fewer high-end chips, this could squeeze margins at Nvidia and potentially undermine the perceived need for large volumes of cloud storage and analytical capacity.

Some analysts believe that large language models may end up as a commodity, while the applications that sit on top of them will deliver the real value of AI.

Some analysts believe that large language models may end up as a commodity, while the applications that sit on top of them will deliver the real value of AI. This was certainly true of the Internet, where Amazon and video streaming services have far outstripped the performance of businesses such as web browser developers and internet service providers. The arrival of DeepSeek could accelerate this discovery process, making it cheaper for companies to experiment.

What use cases?

For the time being, AI lacks an overwhelming use case. Although it took Amazon many years to become profitable, for example, this delay in massive application is still a concern for some investors, given the billions that the industry has already attracted – the worry is that the return on investment might never come.

The focus is on identifying applications that will attract substantial investment and generate meaningful revenue. Despite the efficiency gains achieved so far, artificial intelligence has yet to deliver the significant productivity improvements that were initially expected.

At the margins, some use cases are emerging, even if not at the scale that investors are hoping for. For example, BlackRock says of the mining industry: “[AI] can provide a processing boost, helping companies extract more from their mines, and with greater efficiency. It can help drilling and exploration become more precise and targeted.

“It can also help with the maintenance of expensive machinery and equipment. A growing range of sensor and data networks – the Internet of things – can detect small problems, allowing them to be fixed quickly and improve the lifespan of equipment. Digitalising supply chains can help introduce early warning systems for problems, which can help minimise supply disruption.”

Agriculture and smart farms

There is also strong uptake in the agricultural sector. Smart farms are emerging where digital monitors adjust fertiliser and watering systems for maximum efficiency. They can give early indications of disease and changes in climate, helping to boost yields.

There is also strong uptake in the agricultural sector. Smart farms are emerging where digital monitors adjust fertiliser and watering systems for maximum efficiency.

A World Economic Forum paper notes: “The digitalisation of agriculture offers benefits like higher farm incomes, improved environmental outcomes, and better commercial viability when working with smallholder farmers. Research suggests that digital agriculture could boost the agricultural GDP of low- and middle-income countries by more than $450 billion, or 28% per annum.”

The World Economic Forum has launched an initiative in collaboration with the government of the Indian state of Telangana to implement AI tools in chilli farms. It has already helped farmers achieve a 21% increase in yields, a 9% reduction in pesticide use, and an $800 boost in income per acre per cycle, according to reports.

Autonomous driving is another area that is slowly gaining ground while making extensive use of AI. Driverless taxis are already operating on the streets of San Francisco. AI is also making inroads into medicine, particularly in areas such as cancer diagnosis.

Navigating the roadblocks

Many potential use cases are emerging in which AI may be transformational – in transportation, to analyse traffic patterns and optimise flows; in healthcare, in diagnostics, to analyse medical records or for drug discovery; in education, for personalised learning plans; and in business to optimise supply chains.

There are still some roadblocks. Companies have complex legacy systems with structures not necessarily capable of incorporating AI easily, and they will take time to adapt. AI still needs humans to function, meaning it can be met by reluctance to embrace new technologies. Furthermore, the future shape of regulation has yet to emerge; most governments worldwide are drawing up some form of AI regulation, which could hinder growth, depending on how it is implemented.

As with all innovations, the evolution of AI may not be linear, and we are likely to see the emergence of uses that had not been previously considered. Investors should be cautious about putting all their bets into a single element of the AI ecosystem, instead seeking to spread their investments over a range of potential opportunities.