IBM‘s board faced an uncomfortable decision after learning that the company’s second quarter had fallen well short of expectations.
Directors can now take a breather and wait until the scheduled earnings release and let executives explain the results in detail. Or they could warn investors immediately and risk a violent market reaction.
They opted for disclosure.
IBM (IBM) shares fell more than 25% on July 14, wiping off approximately $69 billion in market value in the company’s worst one-day drop ever. IBM’s market valuation is around $202.5 billion, after the stock ended July 17 at $212.67.
The early data were disappointing but not bad enough on their own to justify the magnitude of the selloff.
IBM estimates revenue for the second quarter to be $17.2 billion, up 1% from a year earlier, and operating earnings of $2.93 a share. Wall Street was looking for about $17.86 billion in sales and $3.02 per share, according to LSEG statistics as cited by Reuters.
What spooked investors was the cause for the shortfall.
Customers shifted spending to servers, storage, and memory that are needed for artificial intelligence infrastructure. And little hardware went to the front of corporate purchase queues, slowing large IBM software and mainframe-related deals.
That suggests a risk that extends well beyond one particularly poor quarter.
Artificial intelligence can weaken IBM without replacing it. It can affect IBM by reducing the technological budgets its clients once spent on traditional software, consulting, and mainframe systems.
First-quarter revenue rose 9% to $15.9 billion. Software revenue increased 11%, infrastructure advanced 15%, and IBM Z revenue jumped 51%. Management maintained its expectation for more than 5% constant-currency revenue growth in 2026 and approximately $1 billion of additional annual free cash flow.
But the preliminary second-quarter data altered that tale quickly.
Software grew at a slower rate of 5%. Consulting revenue was flat. Infrastructure revenue decreased 7%, below management’s prior guidance for a low single-digit decline as the first IBM z17 mainframe launch cycle developed.
IBM claimed the infrastructure drop was worse than expected due to inadequate Z system performance and related transaction processing software.
That counts.
IBM did not say large acquisitions had failed or businesses newly acquired were crumbling. Krishna said HashiCorp and Confluent performed well, while Red Hat revenue growth accelerated to 11%.
There was significant wholesale infrastructure demand for IBM.
Its distributed infrastructure business, including power systems and storage, gained 37% and ended the quarter with a backlog of over $500 million. IBM also claimed that the z17 program was still approximately 130% of the corresponding z16 cycle, despite the quarterly setback.
The trouble was that IBM didn’t translate enough of that demand into the sales mix and timing that investors wanted to see.
Customers moved late-quarter capital investment to supply-constrained servers, storage, and memory ahead of expected price hikes. Many significant transactions did not close on time because the corporation did not respond fast enough, said IBM.
Related: Oppenheimer sends warning on IBM after shares crash
Some deferred deals may still close in later quarters.
That would imply the second-quarter failure was mainly a timing issue, not a sign of lasting demand destruction.
But a corporation that portrays itself as a trusted navigator through complicated technology shifts should know how its greatest customers are allocating their finances.
The board of IBM reportedly grilled Krishna before choosing to send out the early warning. The move might ultimately help the company regain confidence through transparency, but for investors, the decision initially was considered a sign that management had lost visibility into its own sales funnel.
That’s why a small revenue miss set up a massive stock-market reaction.
The quarter didn’t just test IBM’s financial guidance. It did not make the case that the corporation could navigate a shift from traditional enterprise computing to artificial intelligence with any reliability.
AI is squeezing IBM from two directions
So the first challenge is budget competitiveness.
To develop artificial intelligence systems, companies require processors, memory, storage, networking equipment, and data center capacity. If these components are in limited supply, or if their price is likely to rise, consumers may buy them before accepting less urgent software or consultancy projects.
That seems to have been the case in the closing weeks of IBM’s second quarter.
Banks and other big companies focused on infrastructure buys, throwing off the timing of software and mainframe deals IBM anticipated to close. IBM’s warning was among the clearest signals thus far that spending on artificial intelligence may squeeze other business technology expenditures, Reuters said.
So IBM can get a piece of that spend.
The 37% rise in distributed infrastructure shows that consumers were buying some IBM servers and storage solutions. But its strength was not enough to offset weakness in the higher-end mix of mainframes and transaction-processing software.
That points to a big weakness in IBM’s broad portfolio.
The company supplies gear, software, and consulting services that work together. That mix can improve customer ties as tech budgets broaden.
Things become trickier when clients have strong preferences for one group or the other.
IBM could win a storage sale but lose or delay the software and consultancy revenue that was to accompany it. Therefore, spending on artificial intelligence infrastructure can stimulate demand within one IBM division while at the same time hurting another.
The second challenge is product replacement.
AI-powered coding bots and automation technologies could one day lower the costs firms spend for traditional software development, application maintenance, and consultancy services. They could also facilitate the modernization of legacy systems without such heavy reliance on the vendors who built and maintain them.
IBM is especially vulnerable since its hybrid-cloud strategy straddles two computer eras.
Red Hat helps users run applications on private systems and public clouds. IBM consultants help major enterprises modernize their technology while keeping vital software and data that resides on mainframes.
The plan relies on clients continuing to pay IBM to bring their legacy operations together with modern technologies.
If companies require help with securely implementing complex systems, artificial intelligence could make that bridge more useful.
It also could lower the value of the bridge if automation makes it easier to move or if consumers spend their available money elsewhere, such as on infrastructure vendors, cloud platforms, and independent AI developers.
That’s the secret problem underlying IBM’s warning.
The corporation isn’t merely trying to market artificial intelligence. It is vying for its customers’ dollars with artificial intelligence.
IBM has proven to be able to create demand for AI.
The corporation announced a generative AI book of business exceeding $12.5 billion as of the close of 2025. IBM’s definition, however, mixes software transaction revenue and new annual contract value from software subscriptions and consultancy contracts. That does not mean $12.5 billion of reported quarterly or annual AI revenue, however.
IBM still has to translate those contracts and signings into sustainable growth that can offset challenges elsewhere.
More AI:
IBM is increasingly looking to software businesses and technology it has bought, which still hold the most significant money-making potential in the future.
Red Hat has continued to post double-digit growth, while Krishna claimed HashiCorp and Confluent did well throughout the quarter. Those businesses improve IBM’s hybrid cloud, infrastructure automation, and data management position.
IBM is also making a big investment into quantum computing.
The firm says it expects to invest more than $10 billion over five years and is on course to deploy a large-scale fault-tolerant quantum computer in 2029. IBM also revealed plans for a quantum wafer foundry, financed by $1 billion in proposed federal incentives and $1 billion in business contributions.
Those projects could gain strategic importance.
They can’t fix the next couple of quarters.
Investors looking to buy IBM after the selloff face the question of whether the company’s existing software business can pay for long-term objectives as artificial intelligence impacts client spending in real time.
IBM’s latest warning puts its AI transition under pressureAlex Wong / Getty Images
What IBM investors should watch before buying the dip
The first issue is full-year guidance.
It did not revise its annual outlook in the preliminary-results letter. Management said it would disclose full-year estimates when it announces final second-quarter earnings on July 22.
IBM had previously projected more than 5% constant currency revenue growth and $1 billion of incremental free cash flow for 2026 ahead of the warning. A big cut would suggest management believes the expenditure interruption is more than a one-quarter timing issue.
The second question is whether delayed deals pay off.
Many significant transactions did not close within IBM’s projected schedule, said Krishna. Investors will have to figure out whether those clients merely delayed their selections or whether they shifted the money permanently.
A delayed sale can increase revenue over time.
If a project is cancelled, that would mean that artificial intelligence has shifted the customer’s priorities in a more fundamental way.
Third is IBM Z.
The z17 was still ahead of the identical z16 program, but the division had not yet delivered the quarterly results management had expected. Investors should gain more clarity on shipments, capacity expansion, and revenue from transaction-processing software that comes with those systems.
The fourth problem is the quality of software growth.
The 11% growth at Red Hat, plus HashiCorp and Confluent’s showing, is proof that IBM’s modernization portfolio is solid. But shareholders need to take organic performance apart from acquired growth and see if new products can make up for decline in mature software sectors.
In the first half, IBM had $4.8 billion in free cash flow. IBM’s debt was $66.4 billion at the end of the first quarter, an increase of $5.1 billion from the end of the year as it invested in the Confluent purchase.
That balance sheet does not mean an imminent crisis.
That means IBM has less space for repeated execution errors as it integrates acquisitions, maintains its dividend, and funds costly artificial intelligence and quantum computing programs.
Artificial intelligence could pressure IBM both by consuming technology budgets and automating work performed by traditional software and consulting vendors.
IBM’s July 22 outlook will help determine whether the miss was temporary or evidence of a deeper turnaround problem.
IBM’s historic decline does not prove the company has become irrelevant.
Its biggest customers still rely on IBM systems for important financial, governmental, and industrial workloads. Red Hat continues to grow, demand for distributed infrastructure remains high, and the z17 program continues to outpace its predecessor on a comparable basis.
But the warning changes the burden of proof.
Krishna’s goal has been to position IBM as the link between old corporate systems and the future generation of cloud, artificial intelligence, and eventually quantum computing.
That middle position looked promising as customers were growing multiple technology expenditures at the same time.
The danger increases when artificial intelligence pressures customers to decide which investments to fund first.
IBM may find that the infrastructure needed for AI eats up the budget before IBM’s software and consultancy businesses can monetize the transition. At the same time, more advanced AI technologies might put longer-term pressure on the older software and services companies that are funding the company’s shift.
A race that IBM can’t win by just sitting around.
There’s a recovery case; if the delayed transactions close, Red Hat will keep double-digit growth, and the firm will preserve its full-year cash-flow objective. Then the size of the selloff could seem extreme relative to a transient disruption to buying.
If IBM decreases yearly projections, demand for transaction processing continues to weaken, or significant customers build up artificial intelligence capabilities themselves without buying the broader IBM portfolio, the bearish case is strengthened.
The question is not whether IBM is in artificial intelligence.
The question is whether the corporation can generate enough AI-related revenue before the technology cannibalizes the companies IBM is utilizing to fund its future.
Related: IBM’s historic crash exposes AI spending trap
