AI Returns, Data Center Buildout, the Inflation-Rate Overhang, and the Nvidia/Salesforce.com Narratives

September 2026

Since mid-August, the technology-heavy portion of the equity market has come under pressure from a combination of fundamental and macroeconomic concerns. Investors have grown more skeptical about the scale of capital expenditures required to support the AI buildout, while local resistance to data centers has slowed or complicated some projects. At the same time, renewed inflation concerns and the possibility of higher interest rates have dampened risk appetite. Together, these forces have compressed valuation multiples and weighed on many of the stocks that had previously led the market.

That said, the bearish case is not the only plausible path. Several developments could help revive the favored AI-related trade, particularly if investors begin to see clearer evidence that today’s spending can translate into durable cash flow, manageable infrastructure growth, and a more benign rate environment.

First, forthcoming AI listings could provide an important test of public-market confidence. If Anthropic, widely expected by market observers to pursue a public offering, can present a credible path to positive cash flow within a reasonable timeframe, it could shift the debate from capital intensity alone to return on invested capital. Such a presentation would also increase pressure on OpenAI, should it move toward the public markets, to articulate a comparable path. If investors become more comfortable that large AI infrastructure spending can produce substantial returns, the current skepticism toward hyperscalers’ capital expenditures could ease meaningfully.

Second, the political pushback against data centers may prove more manageable than it currently appears. Developers and hyperscalers have done a poor job explaining the economic and physical footprint of many proposed projects. Concerns about water use and power costs have often gone unanswered, even though some projects are designed around closed-loop water systems, dedicated power generation, or other measures intended to limit their burden on local infrastructure. Given the political sensitivity, developers are also likely to improve community terms through tax agreements, local investment, job commitments, or infrastructure support. For that reason, data-center resistance could remain a headline risk without necessarily becoming a lasting constraint on the broader buildout.

Third, inflation and interest-rate fears remain central to the market’s current caution. Inflation had already proven sticky before the latest energy shock, with earlier tariff effects filtering through parts of the supply chain and keeping pressure on goods-heavy categories such as manufacturing, autos, and housing. The AI buildout itself has also been somewhat inflationary in the near term, as demand for specialized components, power, construction labor, and equipment has outpaced supply. Yet a near-term Fed rate hike, if interpreted as a credible effort to re-anchor inflation expectations, could paradoxically lower longer-term yields. In that scenario, the market might begin to look through today’s inflation pressure and re-rate long-duration growth assets as bond-market confidence improves.

Fourth, June quarter results and outlooks from leading AI proponents Nvidia and Salesforce.com have eased concerns about returns on investment, growth prospects, and customer concentration. Nvidia says roughly half of its customers are outside the hyperscale category—evidence, in its view, that adoption extends well beyond the largest cloud providers. The company also surprised Wall Street by projecting 70% growth for its next fiscal year, up from 40%. Salesforce.com is likewise seeing faster AI adoption across its customer base, challenging skeptics who argued that AI would undermine the software opportunity.

The key point is that the recent selloff reflects a real shift in investor questions, not necessarily a permanent break in the AI thesis. Markets are asking whether the spending is too large, whether the infrastructure can be built, and whether the rate environment will remain hostile to growth equities. If the answer to any of those questions improves, the same factors that have pressured the trade could become catalysts for its recovery.

In short, skepticism toward AI returns, datacenter permitting, and inflation is understandable. But each concern also has a plausible path toward resolution. A credible profitability roadmap from leading AI firms, better communication and community engagement around data centers, and a Fed willing to defend its inflation-fighting credibility could all help restore confidence. If those pieces fall into place, the market may again reward the companies best positioned to monetize the AI infrastructure cycle.

Michael P. DeSantis
Managing Member

Bruce M. Lupatkin
Managing Member