Bloomberg Talks · Oct 9, 2026
With Peter Orszag, Chairman and CEO of Lazard
AI investment is running at maybe 3% to 3.5% of US GDP, Peter Orszag said, a bigger share of the economy than railroads, highways or electrification took.
Most of the argument over AI is about whether the technology works. Orszag, who said he probably has 70 or 80 AI agents running at once, believes in the technology. His worry is the financing: so much of the build-out runs on borrowed money that a shift in where customers spend on AI could leave lenders exposed, and he thinks cheaper models running on local machines could cause that shift.
"I think that's the part that people are not appreciating — the ability to run the models on an air-gapped computer based on local compute." — Peter Orszag
Peter Orszag, Chairman and CEO of Lazard, on Bloomberg Talks, ran the Office of Management and Budget in the Obama administration and the Congressional Budget Office before that, and his research led to automatic enrollment in 401(k) plans. In January he and Adam Posen of the Peterson Institute forecast that US inflation would rise above 4% by the end of 2026.
The full interview is covered here so you can skip it.
Here are the 7 takeaways that matter.
Key Takeaways
Inflation is coming from repeated supply shocks, so treating each one as a one-off is a mistake, he said
Washington missed the chance to lock in low rates on longer-term Treasuries and shortened its borrowing instead
AI investment is larger than railroads, highways and electrification, and much of it is borrowed
Spending that moves to open-weight models or local computers could leave lenders exposed
Auto-enrolling children in Trump Accounts is a good thing, by the same logic as automatic 401(k) enrollment
Social Security and Medicare will not be fixed before a crisis forces it, as in 1983
Lazard is hiring more managing directors but running smaller deal teams, so junior bankers take on responsibility sooner
AI should end the 2 a.m. busywork, and bankers can be trained the way pilots are, with simulators
1. Supply-Side Era 2.0
The host read from an essay Orszag wrote with Adam Posen in January, "The Risk of Higher US Inflation in 2026." It said research shows consumers judge inflation by how often they buy things like eggs, and that inflation above 4% by the end of 2026 was plausible and the most likely scenario. The host said the two had called it correctly and asked what comes next.
He said the point was the reason for the forecast, not the forecast itself
"I think the important thing about that is not necessarily the victory lap, but why did we say that?" — Peter Orszag
"And the reason we said that is we are living in a new supply side era where if you want to understand inflation today, you need to look at the supply shocks that happened a year or two ago." — Peter Orszag
New shocks will keep arriving, so inflation will not simply fade
He said he had just read research arguing that inflation will ease over the next few months as one-time effects wear off. His and Posen's other point is that the shocks will keep coming.
"So we're living in a supply-side era 2.0, in which you're going to be hit by these things on an ongoing basis. And so treating them each as a one-off thing is a mistake." — Peter Orszag
