Key Takeaways
Wall Street is demanding evidence that artificial intelligence investments are generating measurable returns for corporate customers, not just for the companies selling the infrastructure.
CNBC's Jim Cramer said Wednesday he needs "cold hard" proof that AI is paying off for enterprise clients, as a growing divide emerges between companies supplying AI infrastructure and those trying to use it. His comments come as IBM's disastrous pre-announcement — shares plunged 25%, wiping out about $70 billion in market value — showed that the shift in corporate technology spending toward AI hardware is actively hurting demand for traditional enterprise software.
"The conditions require our teams to execute perfectly, and this quarter we faltered," IBM Chief Executive Officer Arvind Krishna wrote in a letter to investors, acknowledging the company underestimated how quickly clients would redirect capital toward servers, storage and memory chips to secure supply-constrained AI infrastructure.
The tension crystallizes a question that has hung over the technology sector for two years: When will the hundreds of billions flowing into AI show up in corporate customers' profit-and-loss statements? Microsoft, Amazon and Alphabet are committing roughly $575 billion collectively this year on AI infrastructure, on top of hundreds of billions already invested. Yet enterprise software companies are seeing the opposite — their customers are delaying purchases or building in-house alternatives.
The Enterprise Software Squeeze
IBM now expects second-quarter revenue of about $17.2 billion, representing annual growth of just 1% and falling short of analysts' estimates of $17.86 billion. Adjusted earnings per share are projected at $2.93, below Wall Street's $3.02 consensus. The guidance triggered a selloff that spread across the software sector, dragging down Microsoft, Salesforce, ServiceNow and Intuit.
Salesforce has been hit even harder than IBM, plunging 36% year to date on similar concerns that AI tools will replace rather than complement traditional software. Last week, Bloomberg reported that Starbucks is looking to reduce the $400 million it pays Microsoft and IBM annually for software tools by replacing them with in-house applications built with AI.
Citi analysts offered a contrarian view on Microsoft's Copilot, raising their adoption estimates and predicting "notable stronger Copilot adoption momentum" with M365 Copilot net adds of 8 million versus 5 million in the prior quarter. Microsoft reports fiscal fourth-quarter earnings after the close on July 29. Citi maintained its buy rating but cut its price target to $570 from $620, citing multiple compression in enterprise software.
Microsoft shares rose more than 3% on Wednesday, extending their month-to-date gains to 6%. Still, the stock is down 18% year to date and 27% below its record close of about $542 in late October 2025.
What's at Stake for Investors
The AI spending rotation creates a two-tier market. Infrastructure providers — Nvidia, data center operators, memory chipmakers — continue to benefit from the $575 billion capital expenditure wave. But enterprise software companies face a period of uncertainty as customers reassess budgets and build internal AI capabilities.
For Microsoft, the path to justifying its roughly 30x forward earnings multiple depends on showing that Copilot and Azure can convert infrastructure spending into recurring software revenue. IBM's Red Hat business, which grew 11%, and its distributed infrastructure backlog of about $500 million offer some bright spots, but the core software business is under pressure.
The broader question — whether AI will ultimately expand the software market or cannibalize it — may not be answered until enterprise customers report their own cost savings and revenue gains from AI deployments. Until then, Cramer's demand for "cold hard" proof reflects a market that has priced in the promise but has yet to see the payoff.
This article is for informational purposes only and does not constitute investment advice.
