�۶���Ƶ

Notably, CIO expects the share of AI spend beyond the Big 4 to rise to over 40% in 2025 from below 20% in 2023, thanks to the emergence of new AI players in China, the rise of neoclouds, and other developments. Hence, we believe AI industry fundamentals are more resilient than what the recent market volatility implies. In this blog, we provide a deep dive into the resilient AI spend and adoption trends and expand the scope of our AI estimates beyond the Big 4 to introduce global AI forecasts.

With 1Q25 reporting around the corner, AI investors are eagerly waiting for both the spending and adoption data given the roller-coaster ride in sentiment so far this year. The good news is recent supply chain results and industry adoption data from the US Census Bureau are reassuring, supporting our view that AI fundamentals are intact.

While we recommend investors to closely monitor the management guidance from the Big 4 (Microsoft, Amazon, Alphabet, and Meta), it is worth highlighting that other AI players are now spending at a faster pace and driving ongoing resilience These new players include those in China following the success of low-cost models like DeepSeek; neocloud providers like CoreWeave or Lambda, which run dedicated AI-only data centers; and other enterprise- and sovereign-based players like Oracle and Softbank. We expect the combined share of these companies as a percentage of AI spend to rise from less than 20% in 2023 to more than 40% in 2025. It is therefore imperative to also track the spending of these companies to assess the broader risk-reward for AI. The good news is AI spending beyond the Big 4 should grow by more than 80% in 2025 and more than 50% in 2026, based on our estimates, meaning overall AI growth momentum should continue.

So, we have expanded the scope of our AI estimates beyond the Big 4 to now introduce global AI spend estimates. We expect global AI spend to increase by 60% y/y in 2025 to reach USD 360bn and 33% in 2026 to reach USD 480bn, with the Big 4’s share of AI spend to decline from 58% in 2025 to 52% in 2026.

The broadening of AI spend is a healthy development for the AI theme, as reduced concentration should eventually lead to less market volatility. We expect AI spend beyond the Big 4 to reach a solid USD 150bn in 2025, with the spending split based on our estimates. The success of low-cost models, strong support from the government, and increased use cases in consumer applications like e-commerce, social media, and advertising should drive robust spending from China AI; we estimate China will account for 35% of the USD 150bn in spending in 2025. Neoclouds are another emerging segment, claiming 25% of the spend given their unique software integration. Other hyperscalers and enterprise and sovereign cloud providers should drive the rest.

Globally, within the AI market, we expect AI compute to remain the dominant spending item, but other segments like high-bandwidth memory (HBM) and elsewhere in tech like networking are gaining traction. Meanwhile, we expect industrial capex, which includes spend on cooling, power, and other infrastructure, to drive the rest of the spending. The strong outlook for industrial capex from the Big 4 and the other AI companies bodes well for our Power and resources theme.

While strong capex should continue, investors will likely continue to focus on AI’s potential return on investment (ROI) in the near term given the ongoing tariff-related uncertainty and mixed economic outlook. The tug of war between AI bulls and bears will therefore likely continue throughout 2025. While bears may continue to argue that AI revenues still lag significantly behind capex (which will likely remain true for this year), bulls may argue that 2025 should witness a sharp improvement in AI’s monetization.

The good news is with more supportive signs emerging, we believe AI monetization should indeed improve sharply in 2025. First, cloud growth should stay strong at the leading three platforms—potentially reaching USD 265bn in 2025. As a result, while AI revenues should continue to lag capex spend in 2025, investors who care more about “delta” will likely be pleased to see a narrowing in the gap between AI capex and revenues.

Second, while for many companies investing in AI means maximizing revenue opportunities, for some it's simply about reducing costs thanks to AI's ability to greatly boost productivity. So we believe measuring the economic value add from AI, which captures profits, is a better metric to track AI monetization than simply looking at revenues.

Recent corporate filings from industry leaders that are early AI adopters like Meta and Klarna (unlisted) are already showing strong productivity improvements. To measure employee productivity, we compared revenue per employee metrics since the end of 2022, the year generative AI applications like ChatGPT were first introduced.

Meta's revenue per employee improved from USD 1.35mn in 2022 to USD 2.22mn in 2024 (more than 64% improvement in two years). For Klarna, we see an even stronger improvement from USD 0.34mn to USD 0.82mn (more than 140% improvement in two years). Against this backdrop, we believe our 30% CAGR estimate for AI compute spending from 2024-29 is reasonable.

The broadening AI trend is also evident in adoption rates, which are expanding both in terms of industry and size. The recent quarterly report from the US Census Bureau on the results of its Business Trends and Outlook Survey, which tracks AI adoption across 1.2 million firms in the US, showed strong sequential improvement in AI adoption rates. The AI adoption rate rose from 5.7% in 3Q24 to 7.4% in 1Q25, and we expect it to reach 9.9% by 3Q25 and to cross 10% by the end of the year. To put this into perspective, it took 24 years for US e-commerce penetration to cross the 10% threshold.

Delving into the broadening trends, we see a broad-based improvement in adoption rates across industries. For this purpose, we compared the overall improvement in US AI adoption rates from 4Q24 to 1Q25 (1.3 percentage points) and then looked at which industries reported an above-average improvement in sequential adoption rates (i.e., more than 1.3 percentage points). While technology adoption is not a surprise, the inclusion of new industries like media (arts, entertainment, and recreation), manufacturing, and wholesale trade is encouraging and further supports the AI adoption story. We are also encouraged to see steady AI adoption improvement in the health care and financial services sectors, which bodes well for our Longevity theme and for fintech firms.

From a size perspective, while large enterprises continue to lead AI adoption, it is interesting to note that small enterprises are embracing AI faster than medium-sized businesses —highlighting the broadening trend.

While large enterprises' AI adoption can be explained by their deeper pockets and focus on efficiency, the adoption by smaller firms is perhaps best explained by the extreme automation advantages that computational systems engender. For instance, the latest breed of AI-based unicorns are able to drive extreme automation thanks to the suite of generative AI applications and hence able to disrupt many traditional industries.

In summary, with broadening AI trends, we recommend diversified exposure across both incumbent big tech companies and other fast-growing AI beneficiaries like Chinese AI players. We also continue to like AI semiconductors exposed to compute and memory spend and highlight our Power and resources theme due to strong AI industrial capex.

Original report -