• Goldman Sachs (NYSE:GS) CEO David Solomon expressed optimism that the U.S. will see a 'significant amount' of continued investment in data centers and compute capacity, according to a CNBC interview.
  • The outlook comes as permitting, power supply, and financing challenges threaten to slow some projects, but industry data still points to robust spending.
  • Analysts say the divide is widening between well-funded, permitted developments and speculative builds that may never break ground.

Goldman Sachs CEO David Solomon said he remains optimistic that the U.S. will see a "significant amount" of investment in data centers and computing capacity, according to an interview with CNBC, even as the industry confronts mounting obstacles from permitting delays, electricity constraints, and tighter financing conditions.

The remark, which signals confidence in the durability of the AI infrastructure boom, comes at a delicate moment for the sector. Demand for computing power remains intense—IDC reported $121 billion in global AI-infrastructure spending in the second quarter of 2026 and raised its full-year forecast to $521 billion—but the path from announced project to operational facility has grown increasingly fraught.

Political Headwinds and Power Constraints

Recent state and local elections have injected new uncertainty into the approval process. Democratic gubernatorial victories in several states have raised the prospect of stricter permitting rules and greater scrutiny of large-scale data center developments, particularly around water usage and strain on local power grids. Nomura (NYSE:NMR) analysts expect projects slated for 2028 to face greater exposure to these regulatory shifts than the 2027 capacity already under construction.

Electricity supply looms as perhaps the most concrete bottleneck. The International Energy Agency has warned of a mismatch between fast-moving data-center development and slower investment in generation and grid infrastructure, creating risks of delayed connections and higher costs. Developers are increasingly exploring on-site gas generation as a stopgap; the IEA estimates that 15 to 27 gigawatts of on-site capacity could power data centers by 2030, mostly in the United States.

Financing conditions are also becoming more selective. JPMorgan (NYSE:JPM)'s James Sullivan noted that borrowing spreads have widened for so-called neoclouds—specialist AI-computing providers—relative to large, diversified cloud companies. That dynamic is pushing lenders to favor operators with secured power, creditworthy customers, and durable balance sheets.

A Widening Gap Between Winners and Losers

Solomon's optimism, as characterized in the headline, is not an announcement of a new Goldman Sachs investment commitment. Rather, it reflects a view that aggregate spending can remain substantial even as fewer projects reach completion. Large hyperscalers with diversified revenue and internally generated cash are well positioned to push forward, while smaller, debt-dependent developers may struggle to secure financing or navigate permitting.

"What institutional investors like us are really focused on is regulatory stability," said Andrea Valeri, Blackstone (NYSE:BX)'s country chairman for Italy and chief investment officer for private credit in Europe and APAC, speaking at a separate industry event. His comment, while focused on European markets, underscores a broader theme: capital flows toward jurisdictions and projects that offer predictability.

Goldman itself stands to benefit from a continued buildout as a financial intermediary, providing advice, financing, and capital-markets services. The bank reported third-quarter net revenue of $18.68 billion and diluted earnings per share of $17.01, with annualized return on common equity of 18.7%. For the first nine months of 2026, net revenue reached $56.24 billion, a record for the period. A sustained infrastructure cycle could generate further demand for Goldman's services, though it also ties the firm's fortunes more closely to project economics and credit conditions.

What to Watch

Solomon's comments echo a broader consensus that the AI infrastructure buildout is far from over, but the next phase will be defined by execution rather than ambition. Key variables include the pace of AI monetization—whether customers generate enough value to justify the spending—and the ability of utilities and developers to bring new power capacity online in time.

Goldman COO John Waldron said in a company podcast earlier this month that he expects 2027 to be a substantially bigger year for Goldman's own AI deployment, a signal that the bank sees internal efficiencies from the technology as well.

This article was updated to clarify that Solomon's remarks were characterized in a CNBC interview headline and that the full interview was not independently reviewed. An earlier version misstated the timing of the IEA's estimate for on-site gas generation.