AI at Trillion-Scale: How Family Offices Should Rethink Exposure, Governance, and Risk

OpenAI is reportedly in talks to raise at least $50 billion at a valuation of $750 to $830 billion, with Middle Eastern state-backed funds part of the conversation. In parallel, the company is said to have lined up more than $1.4 trillion of spend on AI infrastructure over the coming years. Others are moving too. Anthropic, for example, has explored funding discussions around a roughly $350 billion valuation.

For family offices, this matters less as a headline and more as a map of where risk and influence are piling up. When private companies start posting valuations in the high hundreds of billions and planning trillion-scale capex, public-market yardsticks stop being enough on their own. You can still use them, but you are flying with one instrument.

What we are watching is the build-out of a parallel asset universe. Late-stage private platforms. Secondary deals that give employees and early investors liquidity. Then the “picks and shovels” layer that starts to look like infrastructure: data centers, power, chips, and the real estate that sits under all of it. For multi-generational capital, the better question is not “how do we chase the biggest AI winner?” It is “where can we show up in the ecosystem, get paid for time and discipline, and avoid paying any price for a single story?”

From a UAE vantage point, two implications keep coming up in conversations. First, allocation frameworks need to treat AI as a cross-asset theme, not a venture sleeve. It runs through venture, growth, public equities, infrastructure, and increasingly real assets. Second, governance has to catch up with what these exposures actually imply: longer duration, more concentration risk, and real dependency on a small number of platforms and suppliers.

So the question I would put back to you is simple: how are you updating your investment policy for the scale of AI capital formation, and where do you draw the line between strategic exposure and speculative overreach?