IBM says Q2 software deal delays are already unwinding

IBM told investors that its calendar second-quarter software shortfall reflected delayed enterprise deals rather than lost demand, saying about one-third of the affected transactions had closed in the first three weeks of the new quarter. The company also used its earnings call to pitch Project Lightwell, a new AI-assisted service for maintaining aging open source software, priced at $1 million a year per customer.

The message followed a sharp selloff after IBM’s preliminary results raised concerns about its software business. IBM did not say that the full backlog had returned, and CEO Arvind Krishna described the early closures as evidence of deferral, while stopping short of calling the issue resolved.

Krishna said the missed activity in Q2 was concentrated in large capital expenditure deals with major customers. According to IBM, those customers were directing budgets first toward AI infrastructure, including servers, storage and memory, rather than cancelling software projects outright.

“The majority of what didn’t happen in the second quarter was large capex deals at large clients,” Krishna said on the call. “We have been very pleased to see that about a third of those have already closed.” He added that the early-quarter activity was “not yet full evidence” but a sign that the company was seeing delays rather than demand destruction.

Investors are still testing the deferral argument

Wall Street’s questions centered on whether IBM’s software weakness was a timing issue or a change in customer priorities. Evercore analyst Amit Daryanani framed the debate directly, asking whether demand had been “deferred or destroyed” after the pre-announcement.

IBM’s answer was consistent: enterprise buyers spent heavily on AI infrastructure in the period, and software spending should follow as customers try to make those investments useful. The company did not disclose the dollar value of the delayed deals, the total number of customers involved or how much revenue from those contracts is expected to land in the current quarter.

That distinction matters for IBM because software is central to the company’s margin story and to its positioning around enterprise AI. A temporary shift in budget timing is manageable. A broader reprioritization away from IBM software would be a different problem.

Lightwell targets legacy open source risk

IBM also argued that AI could expand its software opportunity, particularly in security and software maintenance. Krishna said heavy spending on AI infrastructure and models would put pressure on enterprises to show returns, with more value moving to orchestration and data layers.

Project Lightwell is IBM’s clearest example from the call. The service is aimed at enterprises that still depend on older open source packages after community maintainers have moved on. IBM says Lightwell uses AI to remediate and validate those packages.

Krishna linked the opportunity to Anthropic’s Mythos release in April, saying it had accelerated the discovery of security vulnerabilities in legacy codebases. IBM described the addressable market as multibillion-dollar, though it did not provide a current revenue figure for Lightwell or say how many paid subscriptions are live.

The company said early adopters include Bank of America, Citi, Goldman Sachs, JPMorgan Chase, Mastercard, Morgan Stanley, Visa and Wells Fargo. At $1 million annually per customer, Lightwell is being positioned as enterprise software for regulated buyers with large dependency footprints, not as a broad developer tool.

IBM’s call therefore carried two related claims: AI infrastructure spending temporarily pushed some software deals out of Q2, and AI-driven security work could create new software revenue. Investors now have to see whether the delayed contracts keep closing and whether Lightwell becomes a product line with disclosed traction rather than a large market estimate.

This story draws on original reporting from The Register.