IBM dropped its full-year revenue growth target on Wednesday to between 4% and 5% from a prior forecast of over 5%, after sales of its mainframe hardware fell 42% in the second quarter. The cut confirms to shareholders who had watched the stock recover over the last few years that IBM’s most profitable legacy business is seeing some hindrance as the company leans harder on software and AI.

Forecast cut due to mainframe sales collapse

The sales of the business’s Z system mainframes dipped by 42% in the quarter ending June 30, wiping out the growth streak IBM had built since launching its newest Z hardware last year. CFO Jim Kavanaugh told Bloomberg the change was tied to weakness in the infrastructure unit and its associated software, adding that the rest of the company was performing well.

IBM also lowered the projections for its software unit, which Kavanaugh said would now grow only 6 to 8% for the year.

Infrastructure sales fell 7%, consulting remained flat, and software grew by just 5%, down from a pace of 11% in the first quarter. Total revenue saw an increase by about 1% to $17.2 billion, matching analyst estimates, while diluted earnings rose 5% to $2.93 a share.

Investors were prepared for the plunge

None of this was a surprise by Wednesday. IBM had preannounced having weak numbers for the quarter on July 14, a warning that sent the stock down 25% in a single session, the steepest one-day fall in the company’s history.

The July 14 numbers came in at $17.2 billion against the $17.85 billion Wall Street had expected at the time, with non-GAAP earnings of $2.93 versus a $3.02 estimate.

Due to these already low expectations, the full report saw a muted, positive reaction. Shares rose by about 2% to 3% in extended trading after the report was published, a sign that most of the damage was already reflected in the price pre-release.

IBM CFO focuses on hardware prices, not AI

Kavanaugh rejected the idea that AI tools were eating into IBM’s business. He said that customers were pouring money into expanding infrastructure to support future AI work, and that a spike in hardware and memory prices over the last 30 days had affected the extent of the shortfall.

IBM had responded by stocking up, increasing inventory by $600 million year-over-year as the company made early purchases of server and storage parts to hedge against further price increases.

“I don’t think that has anything to do with AI eating software or anything else,” Kavanaugh said, describing the trend as customers simply filling out their infrastructure layer.

Bloomberg reported earlier this month that Starbucks wanted to swap out software from IBM and other vendors for AI tools built in-house. Kavanaugh acknowledged that Starbucks pays IBM about $2 million a year for an application he agrees is “prime to be disrupted by AI.”

The CFO argues, however, that most of IBM’s software sits close to enterprise infrastructure and data, where it is way harder to rip out than the surface-level apps most exposed to AI. IBM has spent tens of billions acquiring Red Hat, HashiCorp, and Confluent to reposition as a software company, and it has partnered with Arm to keep the company’s mainframe relevant for modern AI workloads.

IBM plans cuts to cost amid defense of structure

To protect earnings, IBM said it would raise its cost savings above a $5.5 billion annual exit run rate. CFO Kavanaugh said this was generating about 300 basis points of operating leverage. This discipline will let the company grow profit by 5% and lift operating margins by 30 basis points despite the only 1% revenue gain.

IBM still expects an extra $1 billion in free cash flow this year from trimming third-party technology spending, tightening its supply chain, and cutting administrative costs, with employee headcount staying basically the same.

CEO Arvind Krishna opened the earnings call by owning the miss. “Our AI strategy is the right one for IBM,” he said, adding that the deficit was due to execution. Asked about breakup rumors, Kavanaugh said IBM “is strongest as an integrated company,” and pointed to hybrid cloud, AI, and quantum as the company’s three pillars.

The smartest crypto minds already read our newsletter. Want in? Join them.