Goldman: Investors will shift focus from AI capex to monetization

Christian Mueller-Glissmann, head of asset allocation research at Goldman Sachs, spoke on Bloomberg Television about the market's growing skepticism toward artificial intelligence. According to him, investors are asking whether current capital expenditure will turn into products that can be monetized. The published announcement of the interview contains no figures. What it does show is the shift itself: the conversation is moving from "how much companies are investing in chips and data centers" to "how much they are earning from it."

Where the skepticism comes from

For the past couple of years, the market has rewarded almost any AI infrastructure story. GPU orders, data center construction and power contracts were treated as a leading indicator of future revenue. That logic holds as long as spending grows in step with visible income. When spending runs ahead, a gap opens up, and investors start asking how large it is and when it will close.

According to the announcement, this gap is what the Goldman strategist is talking about. He is not denying the technology. The question is one of economics: if infrastructure is being built for demand that has not yet turned into paid products, the risk falls on shareholders.

Which metrics will now take center stage

The interview announcement does not name these metrics. But if the market demands proof of monetization, it is logical to expect questions about them:

• AI product revenue as a separate line item, rather than buried inside cloud segments;
• the share of paying users in enterprise and consumer AI services;
• inference margins: how much it costs to serve a request and how much customers pay for it;
• payback periods for new data centers and utilization of capacity already brought online;
• customer retention after pilot projects with agents and assistants.

For now, companies are more willing to talk about the volume of their investments than about these metrics. The disclosure gap itself fuels distrust.

What could change for the market

If investor attention does shift to monetization, different segments will be hit in different ways. Chip and hardware suppliers depend on whether customers keep increasing their orders. Cloud providers will have to explain how infrastructure turns into paid services. Startups that live on funding rounds backed by the promise of future revenue will have to show real contracts.

That said, one strategist's comment does not amount to a broken trend. The skepticism may be temporary, and returns from AI may lag because of adoption cycles at large companies, where projects take months to approve. But the market usually does not wait for perfect data: doubts are enough for it to revise valuations.

How to read the signals going forward

For those who follow the industry or make decisions about AI budgets, there are three practical markers. First, in upcoming earnings reports, check whether companies start reporting AI revenue separately: its absence is also a signal. Second, compare capex plans with comments on capacity utilization. Third, when choosing an AI service provider, ask about its unit economics: higher subscription prices and tighter usage limits will be among the first consequences of margin pressure.

Source: Bloomberg.