On January 23, just three days after being sworn in as the 47th president, Donald Trump signed an executive order setting out his administration’s stance on artificial intelligence. Fraught with political language, the order aimed to revoke former President Joe Biden’s “dangerous” approach that Trump argued hindered AI innovation with “onerous and unnecessary” government control.
This order – which covered how the US would “maintain” global AI leadership – followed the announcement of the Stargate Project, a $500 billion investment deal between various private firms to prioritize AI development in the US.
Overall, such strong state support can be seen as a positive according to Mike Minevich – founding partner and chair of Going Global Ventures (GGV) and chair of AI policy for INESCO – but he argues it is not as simple as political rhetoric makes it seem.
“Trump’s emphasis on deregulation, tax incentives and reshoring AI infrastructure could provide a more predictable investment climate,” says Minevich, pointing out this could favor private capital and accelerate US chip production. “But there are significant risks such as trade tensions with China that could escalate, leading to supply chain disruptions. Policy execution remains uncertain, and investor confidence depends on sustained follow-through.”
Identifying the right AI opportunities
Trump’s focus on the sector follows a year when several AI breakthroughs created a frenzy of interest in the technology. Some investors were already wary of a bubble being created and are now taking pause to determine the best long-term opportunities not caught up in the hype. Jonny Bethell, corporate partner in the private equity group at Taylor Wessing, says the main difference now is the sheer scale of opportunity and demand in the space.
“A huge number of businesses and most governments are expecting and planning for [generative] AI to be the driver of game-changing improvements in efficiency, innovation and productivity,” says Bethell. “Investors in AI infra are looking to take a piece of something with potentially unprecedented demand.”
This is translating into different opportunities, depending on the investor. Ankur Dubey, investment director (private equity) at Schroders Capital, says the focus is now on AI-first companies. Here, he expects significant investments to be made, especially for data platforms and software as traditional infrastructure has limitations.
“There are already a few companies emerging as category leaders in this part of the AI stack and [have] achieved significant scale in terms of revenue and value creation, of which several are our portfolio companies through our GPs and direct investments,” adds Dubey. “[And there is] less emphasis on hardware, which is for now dominated by one large publicly traded company.”
This pivot is also being seen by Minevich, with PE funds looking past traditional tech infrastructure opportunities.
He adds: “AI workloads demand high-density GPUs, custom chips, and energy-intensive infrastructure. Rising AI power needs make smart grids, liquid cooling, and nuclear-powered AI infra top investment themes and government-driven AI infra projects (sovereign AI cloud, AI defense systems) create new investment angles.”
“[AI] is not a magic wand! You don’t want to get enamored with shiny objects and investing in AI with vague or immeasurable impacts”
Jason McDannold, AlixPartners
Separating the wheat from the chaff in AI furthers the need for thorough investment research in this area. Jason McDannold, partner and managing director at AlixPartners – and co-head of the firm’s private equity practice in the Americas – likens the recent explosion of AI firms to the period preceding the dot-com bubble.
“I’ve seen PE firms fall prey to not really understanding the nuances, the use cases and the real needs that need to be deployed to pick the right solution,” says McDannold, who argues the opportunity should be clear and obvious to stand out amidst the hype.
“[AI] is not a magic wand!” he adds. “You don’t want to get enamored with shiny objects and investing in AI with vague or immeasurable impacts.”
Getting to grips with Stargate
The $500 billion Stargate deal has already generated a lot of discussion. In particular, headlines were created when Trump’s own adviser Elon Musk – and rival to signatory OpenAI CEO Sam Altman – publicly criticized the deal as not having required funds. The ambition is undeniable though and Tom Whelan, a partner at Reed Smith focusing on private equity, sees logic in how it is structured.
“The investment in new data centers, which appears to be where the lion’s share of the money is going, makes a lot of sense,” says Whelan. “I’d expect future developments to make these data centers more energy efficient than many existing older data centers.”
Less convinced is GGV’s Minevich, who sees this as a “high stakes bet” on AI dominance from the US. In a similar vein to Trump’s use of trade tariffs, investment in AI is a play on enhancing US leadership and Minevich questions the risks of how much the US government is aligning with these private sector firms.
“There are lots of risks here – SoftBank’s debt burden, OpenAI’s evolving business model, and unclear ROI timelines,” explains Minevich. “Execution risks also loom, as similar large-scale initiatives often face delays.”
A key risk has emerged in DeepSeek, a recently launched Chinese AI firm that has already overtaken ChatGPT downloads in the US and has led to Nvidia losing around $600 billion in market value. This disruption, of a technology already threatening to massively disrupt, cannot be ignored, says Minevich: “If AI models like DeepSeek drastically reduce compute needs, infra demand projections could weaken.”
In agreement is Whelan, who points out this is the nature of technology, where dominance is only safe until the next innovation. “A gamechanger can appear at any moment, and US competitiveness is by no means guaranteed in this space,” he warns. “DeepSeek is a good example of a potential gamechanger (despite the political controls applied to it by China) in terms of the cost of AI and the impact it has already had on the share prices of the biggest tech companies when news broke.”
A long-term view
With AI dominating the headlines, in part thanks to continued focus from high-profile figures like Trump and Musk, it can be easy to get carried away with what AI advancements could occur in 2025. Instead, some investors are adopting a more long-term approach about what these new developments could mean in confluence with megatrends like energy transition and infrastructure modernization.
“The best opportunities for me include data centers and the distribution networks that serve them,” says Reed Smith’s Whelan. “In addition, [there is] the back-up battery and generator technology that these data centers need in the event of power failure, so that they don’t overheat, and making (where practicable) the best use of renewable energy technology to help keep these assets as environmentally clean as possible.”
Likewise, Schroder Capital’s Dubey objectively points out how early AI is in development as an investment theme – currently only representing around 10 percent of global venture investment volume.
“We continue to focus our AI investments on the infrastructure layer and application layer of AI stack primarily in software companies generating strong top line growth with high recurring revenue and margins,” says Dubey. “As many companies are competing for AI leadership across the AI stack, something to pay attention to is further de-globalization of AI driven by national competitiveness.”