Goldman’s CEO Solomon says to watch political volatility around data centers

Goldman Sachs CEO David Solomon said client demand for AI compute significantly exceeds supply, while larger and more complex AI financing needs, pressure on rates and political volatility could challenge some projects’ returns and viability. He described the U.S. economy as solid and the bank’s deals backlog as robust, but said geopolitical tensions and energy-flow disruptions add uncertainty; the CFO noted that greater interest-rate volatility could create headwinds. Solomon said a 5% 10-year Treasury yield would not deter him operationally, and confirmed there is no succession transition or timeline.

Analysis

The investable distinction is between AI demand and financeable AI returns. Compute scarcity supports continued infrastructure spending, but it does not establish that marginal data-center projects can earn adequate returns once debt costs, power and utilization are included. If financing becomes more selective, the pressure should fall first on highly levered or weakly contracted projects—not necessarily on established compute suppliers. That creates a potential relative-value opportunity in power and electrical-equipment providers versus speculative data-center developers, subject to checking project pipelines and valuations.

For Goldman Sachs (GS), the comments are two-sided: financing activity can support markets revenue, while larger, more complex client needs raise underwriting and concentration questions. Rate volatility may also help trading activity while making project finance less viable; the net effect on earnings is not established by this call. Near term, the remarks alone do not justify a directional GS trade. Over 1–3 months, watch financing-related revenue, disclosed commitments and credit metrics, alongside Treasury yields and rate volatility. Over 6–18 months, political resistance or tighter permitting could delay capacity buildout even if compute demand remains high.

Contrarian risk: the market may treat compute scarcity as proof of durable pricing power, when it could instead reflect a temporary buildout bottleneck. Conversely, if contracted demand and project economics remain robust through higher funding costs, financing concerns may be over-discounted. Falsifiers include evidence of worsening credit quality or reduced financing activity at GS, project cancellations, or sustained evidence that customers will fund capacity despite higher rates.

Key Decisions for Investors

• Keep GS at neutral on this disclosure; do not infer an earnings upgrade without confirmation in financing-related revenue, underwriting exposure and credit-quality disclosures.
• Watch for a relative-value entry: long established power/electrical-equipment exposure versus speculative, debt-dependent data-center developers if funding costs rise or project delays broaden. Validate order backlogs, customer contracts and valuations before sizing.
• Use a sustained rise in Treasury yields or rate volatility as a catalyst to reassess marginal AI-project economics—not as an automatic short signal. Project cancellations, weaker contracted utilization or tightening credit terms would strengthen the bearish case.
• Revisit the cautious view if AI infrastructure operators demonstrate durable contracted cash flows and continued project funding despite higher rates; that would weaken the thesis that financing constraints will separate winners from marginal projects.