Episode summary
AI Exchanges: Power Problems?
Exchanges (Goldman Sachs)
· · Guests: Brian Singer
Summary
Goldman Sachs Research’s Brian Singer argues that AI-driven data center demand is still in an early, tight supply/demand phase, even after hyperscalers’ 2026 and 2027 capital budgets rose by more than $300 billion combined. He says demand for compute and tokens is being underestimated, especially as enterprise adoption grows and agentic machine-to-machine traffic expands, which could swamp today’s human-led AI usage.
Singer and the hosts frame power demand through a “6 Ps” lens: pervasiveness, productivity, price, policy, parts, and people. The biggest near-term constraint, he says, is labor, not power pricing or model efficiency. Goldman now assumes about 220 percent global growth in AI and broader data center power demand from 2023 to 2030, up from 175 percent previously, with the increase driven by more servers, more energy-intensive inference, and faster deployment of latest-generation hardware.
The discussion also emphasizes practical workarounds. Because grid buildout can take three to five years, hyperscalers are increasingly using behind-the-meter solutions, often natural-gas-based, to get capacity online faster. Looking ahead to 2030, Singer expects data center power to come roughly 60 percent from thermal sources and 40 percent from renewables, with nuclear becoming more important in the 2030s as parts and reactor availability improve.
Key takeaways
- Brian Singer says hyperscaler capex for 2026 and 2027 combined is up by more than $300 billion, reinforcing a still-tight power-demand outlook.
- Goldman now models about 220 percent global growth in AI and broader data center power demand from 2023 to 2030, versus 175 percent before.
- Singer says the biggest constraint is people: the US may need 500,000 new jobs tied to generation and transmission buildout, including 45,000 current energy apprentices rising by another 20,000 to 25,000.
- Because grid upgrades can take three to five years, hyperscalers are increasingly turning to behind-the-meter natural gas solutions to bring capacity online sooner.
- Singer expects AI data center power to be sourced roughly 60 percent from thermal and 40 percent from renewables through 2030, with nuclear playing a larger role later in the 2030s.
Notable quotes
“And that is going to trickle down into power and a whole host of other uses”
“We think we're going to need to see 500,000 new jobs in the US alone”
“the major constraint that we are focused on”
“we broadly think about power being sourced for data centers as roughly 60 percent coming from thermal sources”
How the episode opens
Welcome to Goldman Sachs Exchanges. I'm Allison Nathan and I'm here with George Lee, co-head of the Goldman Sachs Global Institute. Together we're co- hosting a series of episodes exploring the rise of AI and everything it could mean for companies and investors and economies.
George, great to see you again. Good to be back. We had a little bit of a hiatus here while I was recovering from knee surgery.
But fortunately, not much happened in the world of AI over the past few months. I'm kidding, of course. It has been a cavalcade of extraordinary events.
And it'll be really fun to talk to our guests about what's shaping the movement right now. Yes. Well, great to have you back, George.
And today we're are going to take a closer look at an area that we touched on before in our conversations, you and I, but really deserve a closer look. And that is the resources that the AI build out could require. We are joined here by Brian Singer, head of GS SUSTAIN in Goldman Sachs Research.
Thank you for having me, Allison. Thank you, George. Glad to have you here, Brian.
Great to be with you all. And Brian, just to kick off the conversation, we have been talking about the tremendous amount of power that data centers will require for a while now. But the amount of spending on these data centers have just continued to grow and grow and grow.
I think that was a key feature of the last few months when you think about the AI space. So, first, put some numbers on those developments and what it could mean for power demand. Sure.
I mean, we have seen the hyperscalers’ capital budgets and R&D budgets in 2026 and 2027 combined increase by more than $300 billion.
From the episode’s captions: roughly the first two minutes, lightly cleaned. Keep watching on YouTube →
Topics
- ai data centers
- power demand growth
- power constraints
- data center expansion
- energy infrastructure
Companies and tickers mentioned
- GS
- Goldman Sachs
Watch the episode
Watch “AI Exchanges: Power Problems?” on YouTube
Summary, takeaways and quote selection by Octagon. Quotes are short excerpts; the full conversation is in the episode.
More from Exchanges (Goldman Sachs)
-
Playing for Scale: The Future of Sports Investing -
John Waldron on GDP Growth, AI, and the M&A Outlook -
Brookfield’s Connor Teskey on the Real Bottleneck in the AI Buildout -
Goldman Sachs Chairman and CEO David Solomon on AI: Implementation, Accountability, and Jobs -
Why Oil Prices Are Surging Despite Recovering Supply -
Where Chinese Exports Are Gaining the Most Market Share
Exchanges (Goldman Sachs), this episode and its artwork belong to Goldman Sachs. Octagon is not affiliated with the show; this page is an independent summary.
