Artificial intelligence is seeping into one sector after another, developing capabilities exceeding those of humans. What does this mean for technology spending? Will the outlays start paying off? Portfolio Manager Derek Glynn provides his assessment.
Generative AI – the technology capable of producing content in response to text or image-based prompts – has achieved impressive levels of advanced reasoning with access to more data and the right type of training. Advances continue rapidly. Recent models can even review and refine their own outputs before delivering a final response.
This year’s breakthrough by China’s DeepSeek large language model (LLM) — developed at just a fraction of the cost of a leading comparable model in the US — has highlighted the potential for AI models to become ever cheaper to train and use.
The cost of processing queries for leading AI models is now roughly 1/100th of what it was a few years ago, on average. Falling prices will likely accelerate adoption and usage to the point where AI will seem to be everywhere, embedded in ever more systems, applications and technologies.
We are now at the intersection of three important trends:
- A continued decline in the cost of processing queries
- A continued reduction in error rates and an improvement in model intelligence, including math, visual reasoning, and image classification,
- A broadening in the number and type of use cases for AI.
AI applications in areas such as education, technical assistance, and code development are well understood. But how about therapy or companionship, now emerging as some of the most impactful uses of generative AI? Here, AI has the potential to enhance the quality of life for a wide range of individuals, including those with limited access to human support. This illustrates not only the diversity of use cases, but also the broad reach and general applicability of AI.
AI adoption rates should increase over time
According to a 2024 McKinsey survey, 78% of enterprises have adopted AI in one or more functions or departments. That is up from 55% a few years ago. Only 16% of enterprises, however, have adopted it in five or more departments. Penetration rates are rising, but it is still early.
In the long run, AI technologies could be integrated into physical form factors such as robotics or augmented reality glasses. Adoption rates will rise when these systems have AI capabilities with access to real-time information about the world around us.
The AI opportunity is driving capital expenditures higher and further investment looks likely. For 2025, we estimate spending by the major US AI leaders to grow by more than 40% year-over-year to almost $290 billion. These investments will be used to build out datacentres and equip them with servers and graphics processing units (GPUs).
We expect to see continued year-over-year growth in 2026, albeit at a slower rate than in 2025.
Monetising Gen AI
There is ongoing debate among investors about whether AI companies, including cloud service providers, can earn acceptable returns on the substantial investments they are making. We believe they have many ways to monetise these outlays. These companies offer essential infrastructure such as storage and computing resources, maintain broad access to GPUs, and are developing applications that leverage the technology. They are also using AI internally to drive efficiencies.
Software developers are more productive, sometimes by up to 30 to 40%, because AI tools can help them autocomplete code as they write it.
The cloud service providers can fund the capital expenditures without compromising the strength of their balance sheets because they have strong core businesses that generate significant free cash flow.
In other areas, semiconductor and semiconductor equipment companies remain well positioned to monetise AI. They are fundamentally enabling the technology by providing the chips, equipment, networking, and hardware that are necessary to train and use the models.
Although we are optimistic about the opportunities related to AI, we also acknowledge we are in a market environment that requires careful navigation of risk factors such as tariffs, export controls, and elevated market expectations embedded in some stocks.
Underappreciated beneficiaries of generative AI
We believe high-quality businesses with proprietary data and few competitors stand to benefit from generative AI. Many investors underappreciate these companies despite widespread optimism about AI as a major driver of growth in the broader economy and as a key factor supporting recent (US) stock market gains.
Strong businesses with unique data can leverage AI to develop new products or services that lead to more revenue opportunities. They can also use AI internally to improve productivity and reduce costs. AI helps automate workflows in areas such as customer service, sales, and application development, resulting in expanded margins and higher profits.
We expect there to be many winners from generative AI, including companies outside of the IT sector.
Also listen to our Talking Heads podcast with Derek Glynn as he explains how use cases are set to expand far beyond personal assistants and chatbots. Derek argues that the proliferation of AI should help justify the billions of dollars of capital being invested in areas such as datacentres and software development.
