# AI investment faces a supply shortage, with corporate earnings the next test

By Riza Dagoc

Canonical URL: https://www.tokenpost.com/news/insights/37861
Published: 2026-10-22T12:43:13.000Z
Updated: 2026-10-22T12:43:13.000Z
Section: Insights & Views

> Daniel Newman says scarce computing capacity remains the immediate constraint on AI investment. Broader confidence will depend on revenue and profit gains at companies adopting the technology.

AI infrastructure investment faces a more immediate constraint from scarce computing capacity than from overbuilding, according to Daniel Newman of the Futurum Group. The next test is whether businesses using AI can demonstrate stronger revenue and profits.

In an October 22 interview with Yahoo Finance, Newman said cloud providers had no spare capacity and that the industry’s immediate challenge was building quickly enough to meet demand. He argued that current supply conditions did not support concerns about excess construction.

Newman said buyers were available whenever suppliers could bring computing capacity and power online. Premiums in spot pricing and rising prices per megawatt pointed to strong demand, he said, while customers would continue to require more computing resources.

For Meta, Newman identified user retention and monetization as the next tests for its AI agents. He said investment in its Super Intelligence Lab and recruitment had produced competitive models, but cited concerns that some users signed up for the agent and stopped returning after two weeks. The company needed to explain how quickly adoption was translating into revenue.

He described two potential revenue channels: payments for agent services and integration with Meta’s advertising business. Personalized agents carrying out tasks for users could interact with advertising and generate additional revenue, he said. The question was how effectively those services could strengthen the company’s advertising advantage.

Newman said investors were uneasy both about rising AI spending and about the possibility that major cloud providers might cut investment. He viewed higher spending as a signal of confidence in future returns, while stressing that companies needed to explain how quickly their investments could pay off.

Returns were already appearing among semiconductor, equipment, packaging, foundry and infrastructure companies, Newman said, with some benefits also reaching software companies and major cloud providers. Evidence was less clear across businesses spending heavily to adopt AI. He said meaningful revenue and earnings gains at insurers and manufacturers would be an important measure of progress.

Cost optimization and greater productivity among knowledge workers were already contributing to corporate earnings, he said. Newman added that early cost reductions from technology adoption could eventually be followed by renewed hiring as businesses expanded. The debate, in his view, had shifted toward which companies would capture AI’s economic value.

Newman expressed a preference for infrastructure, semiconductors, equipment manufacturers and foundries, highlighting the market positions of ASML and TSMC. He projected that NVIDIA would generate $1 trillion in profits over the next six to eight quarters. Broader confidence in AI investment, however, would depend on evidence that adoption was also improving profitability in industries such as healthcare and financial services, he said.