Return on investments in artificial intelligence remains limited. However, the gap between revenue and capital expenditures is closing with monetization potentially occurring in 2026, according to UBS’ Sundeep Gantori.

Artificial intelligence (AI) has been all the hype and repeatedly appearing in headlines. However, the market is still some way from translating its promising technological capabilities into profits.

According to data compiled by UBS, AI capital expenditures (capex) by Microsoft, Alphabet, Amazon, and Meta – which account for half of AI spending in the space – grew from $148 billion last year to $222 billion in 2024. And this is expected to increase further to $267 billion in 2025.

«The return on investment is very limited. Forget about profits, even the revenues are very limited compared to the big capex investments that are made,» said Sundeep Gantori, equity Strategist at UBS Global Wealth Management, during a recent briefing attended by finews.asia

Narrowing Gap

Nontheless, the figures are expected to improve. Using cloud platforms, which act as AI data centers, as a proxy, Gantori noted that the three leading players – AWS, Azure and GCP – have seen combined revenue growth exceed 25 percent, compared to around 20 percent in the fourth quarter of 2023.

«So clearly, the investments in AI for them are translating into increased monetization and in the next 12 to 18 months, this growth should basically lead to better ROA (return on assets),» he explained. «Again, we're not talking about profits yet in AI. Revenues are improving but not at a stage where that matches capex. But the good news is the gap is narrowing. Monetization […] would be more of a 2026-and-beyond story.»

Capabilities and Adoption

Aside from revenue, AI is also seeing improvements in capabilities. Large language models are advancing to use the outputs from generative AI prompts as input, also known as synthetic data. NVIDIA is leading in hardware through chips that are powering AI at increasingly impressive rates.

Usage is also on the rise. According to business surveys and public data, adoption by companies in the US has grown from 3.7 percent in September 2023 to 5.9 percent in September 2024 and this is expected to reach 8.7 percent in the next six months. By sectors, leading adopters include IT, professional services, educational services, finance and insurance, real estate and rental, and healthcare.

«In the future, eventually, I don't think any industry will be left out,» Gantori said.

Investment Strategy

In terms of investments, UBS’ AI strategy focuses mainly on big tech companies. Its preference is due to such firms’ deep pockets and customer advantage when investing in AI as well as strong pricing power and cash generation. While valuations are high, the bank believes this is justified by the positive earnings growth outlook which could drive returns.

The strategy divides allocations into three areas: the enabling layer (infrastructure like GPUs, data centers and power generators), the intelligence layer (i.e. large language models) and the application layer for monetization. 50 to 60 percent of the portfolio is invested in the enabling layer followed by the intelligence (15 to 20 percent) and the application layer (30 percent).

In terms of risks, the bank highlighted areas such as AI regulations, cybersecurity concerns, seasonal volatility and potential export controls on goods like chips.