Goldman's top banker says you're asking the wrong AI question

If you've been trying to figure out whether AI is heading for a crash, Goldman Sachs' top dealmaker thinks you're asking the wrong thing entirely.

You might remember Kim Posnett telling CNBC back in November that it was too early to tell whether AI had become a bubble.

That caution is gone now. Goldman's co-head of investment banking told Business Insider the real question isn't whether a bubble exists, but whether AI will fundamentally change how companies compete and create value, and she says the answer to that is clearly yes.

Her framing: even if pockets of the AI ecosystem inflate and pop, that doesn't make the underlying shift a bubble itself, since the productivity opportunity is too large and too broadly applicable across the economy.

No need to take her word for it; there is solid deal data to substantiate the confidence. The global IPO activity was at $206 billion through September 10, representing a growth of 183% year-over-year. At the same time, M&A activity had amounted to $3.73 trillion, growing by 33% year-over-year.

Goldman Sachs had become the first investment bank with $1 trillion in announced M&A deal activity in six months this year, according to Goldman Sachs' CEO David Solomon.

When you ask how Goldman Sachs keeps doing these major IPOs such as that of SpaceX, Posnett cites an ingredient of the bank's success which is not necessarily technical: long-term client relations.

According to the CEO himself, "Our relationship with Elon Musk goes back two decades. We have been a long-term partner to Elon, his companies and his teams, the kind of relations which companies rely on when they do big deals.

You're likely to hear Posnett's framing tested soon. Both Anthropic and OpenAI have filed initial paperwork to go public, and how those listings land will say a lot about whether "supercycle" or "bubble" turns out to be the more accurate word.