Hunter Point's CEO flagged data center overbuilding risk amid aggressive AI capex cycles.

The artificial intelligence boom is driving unprecedented data center expansion, but Hunter Point Capital CEO Avi Kalichstein warns that investors may be getting ahead of themselves. While he does not see AI infrastructure spending as a systemic financial risk, he expects some investors will ultimately discover they have overcommitted to data centers that fail to generate attractive returns.

"It feels to me like somebody somewhere is overinvesting in data centers that will turn out not to be great investments," Kalichstein said in an interview with CNBC's Alts Investing.

AI companies and infrastructure partners are committing hundreds of billions of dollars to computing capacity. Global AI investment is expected to approach $1 trillion this year, driving demand for data centers, power and related infrastructure needed for increasingly compute-intensive models, according to a Goldman Sachs report.

For private-market investors, the rapid expansion poses a difficult question: How much of that capacity will ultimately be needed? Hunter Point has adopted a cautious approach, avoiding investments where it cannot fully understand the underlying economics or identify likely winners. "If we can't really understand and we can't see the future, we do think that warrants caution," Kalichstein said, contrasting data centers with assets like airports, where long-term demand is more predictable.

The uncertainty carries particular weight in private credit, where lenders face growing exposure to companies and infrastructure tied to the AI investment cycle. Kalichstein said that while private credit remains fundamentally sound—the underlying businesses continue performing well—AI is reshaping how investors assess borrowers. Managers are increasingly segmenting portfolios into three categories: businesses largely immune to AI, companies facing significant disruption or margin pressure, and businesses positioned to benefit from AI while navigating new competitive threats.

This reassessment is directly affecting pricing and decisions about which companies merit financing. Software faces particular vulnerability; businesses once viewed as durable credit risks may face uncertain futures if AI reduces demand for their products or compresses margins.

The data center question differs fundamentally. Rather than assessing whether AI will generate demand, investors must estimate how much demand will exist and where it will ultimately concentrate.

On the broader private credit market, Kalichstein noted that redemption pressure from individual investors persists, even as institutional appetite remains strong. He expects "another few quarters of performance" before that pressure fully resolves. "We're not done yet," he said.