AI investments surge in 2025, driving market gains, fund flows

Several announcements this month highlight the continued surge in artificial intelligence investment, with a particular emphasis on the infrastructure required to support the technology.

On Sept. 10, the Wall Street Journal reported that OpenAI has signed a contract with Oracle to purchase USD 300 billion worth of computing power over five years.^[1] Two days earlier, Nebius Group announced a USD 17.4-billion deal to provide Microsoft with graphic processing units (GPU) capacity over five years.^[2] Last week, a group of US tech firms including Microsoft and Nvidia announced a multi-billion-dollar UK AI infrastructure investment plan.^[3]

These deals suggest that this year's AI infrastructure investment boom - which kicked off with pledges from OpenAI and Meta worth more than USD 560 billion in January^[4],[5] - is far from slowing down. They are also markers in what the BlackRock Investment Institute has described as the first phase in the AI evolution: the buildup.^[6]

Equity investors have taken notice. The STOXX® Global AI Infrastructure index, made up of companies providing the fundamental 'picks and shovels' of AI (such as semiconductors, cloud computing and big data technologies), has risen 32% in 2025.^[7] The STOXX® Global AI Adopters and Applications, which tracks companies implementing AI to enhance products and efficiency, has advanced 14%. Both indices underlie respective iShares ETFs launched at the end of 2024. The iShares AI Infrastructure UCITS ETF has amassed USD 126 million in assets under management, while the iShares AI Adopters & Applications UCITS ETF has USD 200 million in assets.^[8]

"Recent corporate results continue to highlight the strength of AI infrastructure investment, fueling increasing adoption across industries," said Omar Moufti, Thematics & Sectors Product Strategist at BlackRock. "Our investors continue to orient their portfolios towards this mega trend, seeking to be deliberate in their approach to the AI opportunity."

STOXX AI thematic indices

The two indices are part of the STOXX Artificial Intelligence suite, a growing family of thematic indices that track stocks across the full AI value chain and allow investors fine-tuned exposure to companies in different stages and involvement in the disruptive technology. Index construction leverages the STOXX Thematics Framework, which uses revenue and/or patent-based approaches to capture a diverse range of innovators at various stages of AI adoption.

The STOXX Global AI Infrastructure and STOXX Global AI Adopters and Applications indices employ a dual approach that selects companies based on their revenues and patents from technologies and economic sectors associated to each targeted theme. This allows the indices to spot innovators and market leaders across the spectrum and life cycle of AI.

The two indices' different objectives are reflected in their industry composition. While 93% of the STOXX Global AI Infrastructure is assigned to stocks in the Technology and Telecommunications industries, the STOXX Global AI Adopters and Applications has Healthcare, Technology, Industrial Goods and Services, and Banks, as its highest allocations.

Software maker Oracle is now the largest constituent in the STOXX Global AI Infrastructure index following a 43% jump in its shares on Sept. 10.

Figure 1: STOXX Global AI Infrastructure index - Top 10 constituents

Figure 2: STOXX Global AI Adopters and Applications index - Top 10 constituents

Three phases

Current AI investments are expected to drive operational efficiencies and higher profitability in the years ahead. According to the BlackRock Investment Institute, the AI evolution will anchor around three key phases^[9]:

• Buildout: The first phase is the race to build the infrastructure AI needs.
• Adoption: As infrastructure grows and AI applications mature, adoption is likely to accelerate - packaged into different apps and software.
• Transformation: This phase is where companies could unlock the full value of AI adoption, as broad productivity gains and new business models and industries emerge.

The cost of querying an AI model fell 280-fold between November 2022 and October 2024, according to a Stanford University study presented this year.^[10] This is both accelerating adoption and encouraging further development of AI technologies. The same study estimated that private investments in AI in the US, Europe and China amounted to USD 138 billion in 2024. That trend appears poised to reach a new high in 2025, judging by recent corporate announcements.

[1] WSJ, 'Oracle, OpenAI Sign $300 Billion Cloud Deal,' Sept. 10, 2025.
[2] Reuters, 'Nebius signs $17.4 billion AI infrastructure deal with Microsoft, shares jump,' Sept. 8, 2025.
[3] BBC, 'Nvidia boss says UK will be 'AI superpower' as tech firms invest billions,' Sept. 16, 2025.
[4] OpenAI, 'Announcing The Stargate Project,' Jan. 21, 2025.
[5] Bloomberg, 'Zuckerberg Says Meta to Spend Up to $65 Billion on AI in '25,' Jan. 24, 2025.
[6] Source: BlackRock Investment Institute, 'AI's big questions,' November 2024.
[7] Gross returns in USD through Sept. 12, 2025.
[8] iShares data through Sept. 19, 2025.
[9] BlackRock Investment Institute, 'AI's big questions,' November 2024.
[10] 'The AI Index 2025 Annual Report,' AI Index Steering Committee, Institute for Human-Centered AI, Stanford University, April 2025.