Advancing technology: how family offices can position for structural shifts

Against the backdrop of a series of unprecedented IPOs, panellists examined reasons for testing AI use cases now and practical steps that family offices can take to prepare for structural shifts, in a session moderated by Salman Mahdi, Global Vice Chairman of Deutsche Bank Private Bank.

In 2026, the AI market is entering a price‑discovery phase, noted Adrian Cox, Managing Director, Thematic Strategist, Deutsche Bank Research: “The great thing about these IPOs is that we’ll finally have a chance to value pure‑play AI.”

Despite frequent comparisons with the dot-com bubble, Cox cited reasons to be somewhat more optimistic. While that era saw “heavily indebted telecoms companies” laying masses of fibre optic cables that lay unused, new data centres today “immediately generate revenue”.

The potential for expansion from current levels of AI adoption could be equally as important. “Only about 20–25 percent of companies in the US are using AI,” he said. “If you imagine that rising to 60–70 percent, there’s an enormous runway for further growth.”

The panel discussed how these dynamics sit alongside the US’s strengths in innovation and rapid funding through capital markets. Kumar Shah, a Managing Director in Brookfield’s private equity group, said it should be remembered that the growth already achieved by leading US tech companies had far surpassed expectations.

Why family offices are well‑positioned for AI adoption

Mahdi invited the panellists to explore why AI matters for family office portfolios and operating companies. Only around a fifth of family offices are using AI, said Cox, “so you have an enormous opportunity, which is particular to how family offices work”. 

Far from seeing AI as an investment proposition alone, organisational “nimbleness” enables rapid in‑house adoption. With fewer structural constraints, family offices can quickly test and scale what works. The panel was clear that experimentation cannot wait.

The speed of AI improvement means the potential for it to influence investment processes, reporting and back‑office functions is increasing, said Shah. “If you’re not playing with these tools now, by the time they mature it’ll be too late because you’ll already have given up a year of trial and error,” he added.

Incentives matter too: rewarding experimentation and early adoption within an organisation can help build internal momentum. 

How family offices could turn AI into an operational advantage

As the panel took questions from the audience, the discussion moved to concrete use cases already delivering enterprise value. Neil Serebryany, Managing Director, Hyperion Horizons Group, highlighted the shift toward AI agent‑based workflows. “For every human we hire, we try to have 100 agents,” he said. “We call the humans orchestrators and their job is to manage agents.”

This set-up allows for data to be processed at a scale that was previously difficult even for machine learning models – at a cost of billions of tokens rather than “millions of human hours”, explained Serebryany. This shift is beginning to influence workforce structures, and the discussion also touched upon concerns that entry level jobs could be displaced across industries. 

Shah explained how some companies he works with – which send out thousands of proposals per year – have harnessed AI tools to create more consistent pricing. “When we buy a company we run all their pricing through our AI tools to make sure we understand how it works and then we standardise it.”

The panel highlighted the potential for further use cases that boost top-line growth as AI models improve and become more closely integrated with existing systems. While adoption is challenging, value emerges when systems mature, said Cox. 

He suggested defining a use case first and testing whether AI can address it, rather than asking what AI should be used for. The true impact of the iPhone emerged years after launch, he noted, once connectivity, mapping and payments software unlocked its best uses. 

What are the investment themes across the AI value chain?

Turning to portfolio construction, the panellists discussed where value may accrue across the AI stack. They highlighted the diversity of potential opportunities across infrastructure, models and business services. 

“We’ve explored opportunities and made investments across the value chain, and I think the risk and returns across those opportunities are very different,” said Shah. “The data‑centre capacity issue is real. Leading AI developers cannot get enough compute.” He also reflected on the importance of the broader “picks and shovels” ecosystem, including the hardware, components and supporting services needed to deploy AI at scale within large businesses.

AI investments may need to be directed more precisely than in the past, said Serebryany. He identified three structural principles: “There will be an ever‑increasing amount of token usage, so what opportunities does that create? Second, there will be more focus on AI sovereignty. Third, the deeper you go into the technical stack, the longer it takes for a competitor to catch up.” 

Adjacent technologies – from quantum computing to early work on space‑based data centres – were also highlighted for their potential to broaden the arc of AI‑enabled innovation, influencing how future compute and data‑processing capacity is built.

Key takeaways:

• A series of unprecedented IPOs in 2026 mark the first real opportunity to value “pure‑play” AI businesses rather than proxies.
• With only around 20–25 percent of US companies using AI today, there is substantial long‑term growth potential for the technology in the US, where innovation and capital markets are also strong.
• Family offices’ nimble structures allow for faster experimentation with AI than in large institutions and this could represent a significant opportunity for adoption.
• Use cases that boost top-line growth are emerging and are likely to continue to do so as AI models mature.
• Growth in token usage, AI sovereignty and deep‑stack differentiation are three potential value drivers for investors to keep in mind, in addition to high demand for data-centre capacity.