Technology
IBM's earnings miss exposes the infrastructure-first shift remaking enterprise IT budgets
As companies rush to secure AI hardware, traditional software vendors like IBM are losing ground—and engineering teams are paying the price.
The New Stack
IBM's stock plummeted after the company released a preliminary second-quarter earnings update revealing a miss on revenue and profit forecasts. The underlying reason: customers are aggressively reallocating IT budgets away from software toward the AI infrastructure buildout.
CEO Arvind Krishna warned investors that second-quarter revenue would land at $17.2 billion (1% year-over-year growth) with earnings per share of $2.93 (5% growth)—both falling short of Wall Street targets of $17.86 billion and $3.01 per share. But the numbers tell only part of the story. The real signal comes from what triggered the shortfall.
The capex reallocation taking hold#
Enterprise customers are making a calculated bet: they're diverting spending away from expensive middleware and software services toward purchasing servers, storage, and memory chips needed to power large-scale AI systems. In the final weeks of June, companies accelerated hardware purchases specifically to secure supply-constrained components before anticipated price increases. This spending shift caught IBM—and likely other established software vendors—off guard.
Krishna acknowledged the company's own misstep, saying, "These conditions require our teams to execute perfectly, and this quarter we faltered. We did not adapt and move quickly enough, and numerous large deals failed to close on the timelines we expected."
What this means for developers#
When enterprises freeze software budgets, the responsibility for integrating legacy systems with new AI infrastructure falls squarely on engineering teams. Rather than licensing vendor-provided solutions to connect disparate databases and platforms, developers are forced to build these connections themselves.
The practical consequence is immediate: more custom API development, maintenance of fragile integrations using open-source alternatives like Apache Kafka and Envoy, and manual system stitching. If a company opts out of expensive middleware to save cash, engineering teams inherit the integration burden—building ETL pipelines, feeding legacy mainframe data into vector databases for Retrieval-Augmented Generation (RAG), and maintaining bridges between old and new infrastructure with limited vendor support.
Looking ahead: the follow-up phase#
IBM's warning signals that many organizations are still in the hardware acquisition phase of their AI journey. Once that infrastructure is deployed and operational, executive expectations will shift: the expensive servers and storage will need to justify their cost through business value.
