Brookfield Asset Management CEO Connor Teskey has a simple read on the AI boom. The world wants far more compute and electricity than it can build, and that gap is not closing anytime soon.
Teskey expects demand for compute and energy infrastructure to exceed supply through the end of the decade and beyond.
A shortage measured in multiples
Teskey is not talking about a modest shortfall. He described the imbalance in unusually blunt terms.
“Not by percentages, but by three, four, five, six, eight, 10 times.”
That is how much he expects demand to outstrip supply over the rest of the decade.
He also argued the problem is structural rather than temporary. Small efficiency gains or supply chain tweaks will not solve it.
Brookfield puts a number on the scale. The firm estimates the global AI buildout could reach $7 trillion over the next ten years.
The power side may be the tightest constraint of all. The US needs approximately 100 GW of additional power to support AI over the next decade, while the current grid can supply only about 30 GW.
Brookfield’s $100 billion bet
Brookfield has already moved to position itself. In November 2025, the firm launched a $100 billion global AI infrastructure program.
At the center sits the flagship Brookfield Artificial Intelligence Infrastructure Fund. It targets $10 billion in equity and counts Nvidia and the Kuwait Investment Authority among its backers.
The program spans three broad areas: data centers, power generation and the wider compute landscape. On the power side, Brookfield favors technologies that can be deployed quickly.
Those include fuel cells, renewable energy, batteries and nuclear.
That logic explains Brookfield’s expanded deal with Bloom Energy. The partnership has been scaled up, with a revised target of $25 billion for quick-deploy power solutions.
Bloom makes fuel cells that can generate electricity on site. That lets a data center sidestep some of the wait for grid connections, which can stretch for years.
A firm reshaping around digital infrastructure
Teskey took over as CEO of Brookfield Asset Management in February 2026. Under his watch, the firm’s center of gravity has visibly shifted.
AI and digital infrastructure now make up 25-40% of Brookfield’s activities.
Teskey has said he does not fear immediate overbuilding in the sector. His confidence rests on strong backing from hyperscalers and sovereign entities, much of it locked in through long-term contracts.
Rivals turned partners
One of Teskey’s more interesting claims is that the scale of the buildout will push competitors into cooperation. No single firm can fund $7 trillion on its own.
Brookfield’s partner list reflects that. Beyond Nvidia and Bloom Energy, its AI infrastructure collaborations include Apollo, Blackstone and KKR.
What this means for investors and the AI trade
The roughly 70 GW gap between the approximately 100 GW of new power the US needs and the about 30 GW the grid can supply is the clearest stress point.
Teskey’s thesis depends heavily on hyperscalers and sovereigns continuing to sign long contracts. Regulation is another variable, as permitting rules, grid interconnection policy and nuclear licensing will shape how quickly new capacity can come online.
The next signals to track are fundraising progress on the flagship fund, new power partnerships, and whether hyperscaler capital spending keeps pace with Teskey’s forecast.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.