Morgan Stanley CEO Ted Pick Talks AI Risks, Appetite for AI Deals

Ted Pick · Morgan Stanley

AI capex could reach roughly $10-15 trillion over a decade; the buildout is only in the second or third inning.

Pick argues the AI investment boom is real and still early. He says 'the technology is incontrovertible' and that enterprise democratization of AI is already here, with consumers already using it. He frames the capital deployment timeline as uncertain, from seven or eight years to ten or twelve years, against a 10-15 trillion dollar range he treats as digestible over time. He cites an annual capex figure of about 11 trillion dollars rising to about 16 trillion by 2030, and says the current ranges imply roughly 1 trillion dollars a year for several years, which puts the cycle in the second or third inning.

“the kind of ranges we're seeing right now are about 1,000,000,000,000 a year for the next several years, which gets us thinking we're somewhere in the second or third inning.”at 5:05

Ted Pick · Morgan Stanley

Local backlash will slow some data center projects, but hyperscaler-backed projects will still get approved.

Pick acknowledges a NIMBY problem: communities want the cash from a data center but resist having one nearby. He expects this local resistance to become part of the broader economic debate, and expects populist politics in some locations to slow some projects. He still expects ongoing competition to host data centers, provided consumers keep access to their local grid and projects are done in a thoughtful, fair way. Projects funded by the biggest hyperscalers, built in partnership with public bodies and bringing jobs, should be approved, which he calls a public-private partnership.

“some projects might find that they're finding some opposition. But the ones that sort of are need to do and are being, funded by the biggest hyperscalers, I think there's a thoughtful consideration being done in partnership with the public players.”at 6:02

Ted Pick · Morgan Stanley

Hyperscalers' free cash flow can absorb higher rates, and projects will use tailored debt-equity structures.

Pick acknowledges that the debt and deficit problem across Western countries is real, and that AI capex may crowd out non-AI companies that need capital as older, demographically advanced sovereigns refinance. He argues the higher financing costs can be borne by hyperscalers with extraordinary free cash flow, whose returns exceed the cost of capital by a wide margin. He sees the shift as market clearing after 20-25 years of financial repression. Projects can use more debt, investment-grade or subordinated, or more equity at the bottom of the stack, and private structuring tools such as Morgan Stanley's EquityZen acquisition can fund them.

“a lot of the higher interest rate considerations can be borne by these extraordinary free cash flow hyperscalers that effectively see ROIs that aren't 25, 50, a 100 basis points too wide.”at 8:34

Ted Pick · Morgan Stanley

Not every AI project will earn its return, but aggregate efficiency gains should justify the spending.

Pick concedes that not every AI project will generate a return, and says the comparison to the 1999-2000 dot-com bust and the 2001 and 2008 cycles is helpful for that reason. He argues the economy-wide return is what matters. Higher intelligence and productivity should add 50 to 70 basis points to real global GDP growth, and the efficiency and effectiveness gains should justify the 10-15 trillion dollars of investment over 15 to 30 years. He says he is highly confident of that, while acknowledging the final total could range anywhere from 8 to 20 trillion.

“will there be ROI in terms of efficiency and effectiveness on the 10 to 15,000,000,000,000 that justifies the investment now for fifteen, twenty, thirty years to come? I'm highly confident of that.”at 3:47

Ted Pick · Morgan Stanley

Enterprise AI demand outstrips supply, so companies must pay up now or risk falling behind.

Pick says enterprises cannot wait too long to adopt AI. Right now demand is outstripping supply, so companies have to pay up for capacity that will let them compete with what they expect to need two or three years out. He argues that hesitating is a poor strategy because the technology moves ahead quickly, citing Chinese models and developments in the Americas and the rest of the world. Hedging on demand-side and supply-side timing is not an option in his view.

“But you can't wait too long if you're an enterprise... right now, the demand is outstripping the supply. So you got to pay up to be able to get your enterprise in a place where you can compete.”at 3:47