Goldman banker says AI investors are getting pickier

A senior Goldman Sachs banker says investors are getting pickier about AI, even though she remains very positive on the technology. Kim Posnett, co-head of investment banking at Goldman, spoke on Tuesday, Oct. 28, at The Information's WTF Summit. She said weaknesses are developing in the AI market as the cost of capital climbs. Put simply, money is getting more expensive to borrow.

What she said

Posnett said she is "hugely bullish" on AI overall. She said the shift is not a change in how people feel about AI. Instead, she described investors "repositioning" and separating the strong players from the weak ones across what she called the AI trade.

She also said AI projects will not deliver the returns people expect if interest rates stay high for a long time. Interest rates are the price of borrowing money. Her line on bubbles was careful: "While AI itself is not a bubble, I think you can have bubbles within that."

Who investors favor

Posnett said investors are leaning toward companies that do several jobs themselves. These are firms that make top-tier AI models, the software, and the AI chips. Goldman is involved in major deals announced this week, including SpaceX's GPU financing, Meta's new stock offering and Anthropic's planned initial public offering, or IPO. GPUs are the powerful chips that run AI.

The circular money question

Posnett was asked about circular financing, where big tech companies such as Nvidia fund their own suppliers and customers. She said she thinks more about demand, but admitted there is risk. "It makes sense that they want to build closer partnerships with each other," she said. "And I think it creates greater risk. I think both are true."

Other signals

Reports published around the same time show spending is still strong:

• Reuters said the AI boom drove an increase in core capital goods, and business equipment spending remains robust.
• Bloomberg, citing Census data, said AI-related spending has led business investment all through 2026.
• Citi Wealth said the AI buildout is still in its early-to-middle stages.
• Some firms told Bloomberg their plans depend on the midterm elections and Federal Reserve policy.

What it means

The message is not that AI is fading. It is that money will flow more selectively. If you invest in AI companies or rely on them, the strongest firms look better placed, while weaker ones may face more pressure if borrowing stays costly. The material does not name which companies Posnett considers weak, and it does not say how long rates might stay high.