IBM’s biggest one-day plunge in history, driven by results falling short of expectations—how long can its old cash flow hold up?

TL;DR

· IBM discloses preliminary second-quarter results that fall short of expectations, and the stock drops by about a quarter on a single day.

· Customers shift budgets toward servers, storage, and memory, squeezing demand for mainframes and transaction-processing software.

· Related plays: IBM, the AI infrastructure hardware chain, memory makers, traditional software-services stocks, network security companies such as CrowdStrike.

After IBM disclosed its preliminary second-quarter results on July 14, the stock fell about 24%-26% in the day. It closed at $217.07, down about 25.23% from the previous close, marking a decline rarely seen in decades.

What makes this selloff unusual is that IBM is not a company unrelated to AI. Over the past few years, it has consistently pitched hybrid cloud, enterprise AI, consulting, and automation as growth stories. What the market is pricing this time isn’t “whether IBM has AI,” but “whether AI spending is first squeezing out IBM’s older revenue.”

IBM’s preliminary second-quarter revenue was $17.2 billion, up 1% year over year. On a non-GAAP per-share basis under an operating metric, EPS was $2.93. Media aggregations including FactSet show Wall Street expected revenue of about $17.86 billion and EPS of about $3.01. By business segment, software revenue grew 5%, consulting was flat, and infrastructure revenue fell 7%.

For ordinary investors, this can be understood as a change in enterprise IT budget prioritization. When customers worry that AI servers, storage, and memory will get more expensive or be in short supply, they spend money first on these hardware items—“without which AI deployment can’t happen.” The money that was originally earmarked for mainframes and related software may then be postponed.

AI hardware takes the budget first

A mainframe isn’t just a simple synonym for obsolete computers. For banks, insurance companies, governments, and large enterprises, it is more like the foundation of core transaction systems, handling critical business needs with high concurrency and high reliability. Transaction-processing software is the software layer that supports these core transactions.

These businesses are typically “sticky,” with high migration costs for customers—so IBM has long enjoyed relatively stable cash flow. But stability doesn’t mean every quarter will be immune to budget squeezing. When AI infrastructure suddenly becomes a priority resource, upgrades and procurement for traditional core systems may be displaced.

This is what “capex front-loading” looks like. Customers aren’t completely not buying IBM products; instead, to secure supply of servers, storage, and memory, they put near-term budgets into AI hardware. For IBM, the timing of revenue recognition gets disrupted, and large deals are more likely to be delayed.

In a letter to investors, IBM CEO Arvind Krishna admitted that the company missed the mark this quarter and didn’t adapt quickly enough as customers shifted capital expenditures toward AI-related spend. He said that in the final weeks of June, customers shifted quarterly capital expenditures toward servers, storage, and memory to secure supply-constrained infrastructure and avoid expected price increases. Rapidly changing network security issues also distracted customers and management.

The importance of this letter is that it is not external speculation—it is a direct confirmation from the company’s top management regarding budget migration and execution mistakes. AI spending is indeed growing, but that growth has not been evenly distributed across all technology companies.

Execution blamed on the CEO; the market re-prices stability

Krishna’s explanation is relatively mild. This time, IBM didn’t keep up with changes in customer budgets, and several large deals failed to close within the expected timeframe. Still, the company remains confident in its asset mix and strategy, and will roll out new response measures.

If this explanation holds, IBM’s problem looks more like a short-term budget misalignment. When customers spend first on AI hardware, it doesn’t mean they will never buy mainframes and software. The delayed deals may also be made up for in the second half. For long-term investors, this would be an execution blemish rather than a failure of the business model.

But the market reaction is harsher. The stock fell about a quarter in a single day, indicating investors aren’t just punishing a one-quarter shortfall in revenue—they’re re-evaluating IBM’s growth visibility. Previously, the market was willing to anchor IBM to a stable valuation: traditional core systems provided cash flow, and new business provided an option value during the transformation period.

The problem now is that if the AI infrastructure cycle continues to absorb enterprise budgets, the stability of traditional cash flow will be discounted. IBM still has an AI narrative, but it must first prove that new business can make up for the timing gap created by squeezing out the old business.

That’s why the 7% decline in the infrastructure segment is more striking than the headline number suggests. It isn’t isolated segment volatility; it occurs in the business line where IBM most clearly reflects traditional customers’ “stickiness.” The more “stable” the revenue was considered in the past, the more likely it is to trigger a valuation re-pricing once there is a clear deviation.

HSBC downgrades and amplifies concerns about transition pacing

According to reports by Investing.com and StreetInsider, analysts including HSBC’s Abhishek Shukla downgraded IBM’s rating from Hold to Reduce, and cut the target price from $231 to $191.

Reduce can be simply understood as reducing a position. When a traditional tech giant is downgraded into this category after reporting results below expectations, the market often treats it as a signal for fund position adjustments—especially when the stock price had already priced in expectations that the transition would succeed.

This doesn’t necessarily mean HSBC’s view is the final outcome, but it explains why IBM’s downside has exceeded what the earnings miss alone would mechanically imply. Revenue was about 4% below expectations—far from enough to mechanically derive a ~$25% market-cap evaporation. What’s truly being cut is the certainty premium over the next several quarters.

The market also discussed Krishna’s comments in his letter about network security issues within a framework of “relative resilience” for pure-play network security companies. Targets such as CrowdStrike were thus re-focused. But this kind of linkage is more like a supporting piece of evidence for sector-level re-pricing; it can’t simply be written as an IBM event directly pushing up network security stocks.

A clearer main line remains budget migration. In the order of customers’ spending priorities, AI infrastructure hardware, memory, and storage move up, while traditional software and mainframe businesses take the near-term squeeze. The split between winners and losers doesn’t happen between “whether there is AI,” but between “what customers must buy first right now.”

Earnings on July 22 will need to show a path to catch up

IBM’s drop this time still can’t be directly equated to traditional IT giants failing in the AI era. A more prudent view is that AI infrastructure spending is disrupting customers’ budget pacing and exposing pressure on IBM’s traditional business and execution cadence.

The complete earnings conference call is scheduled for 17:00 U.S. Eastern Time on July 22. Since July 14 only covered preliminary second-quarter results, the final figures could differ slightly. What the market will be looking for next isn’t management continuing to emphasize its AI strategy, but whether the postponed large deals have a clear closing timeline and whether the infrastructure segment’s decline can narrow.

If these deals are made up in the second half, the market may reclassify this event as a short-term misalignment. At that point, IBM will need to prove that customers are merely changing the procurement order—not permanently reducing spending on Z mainframes and transaction-processing software.

A more dangerous scenario is if the full earnings report and subsequent guidance still can’t provide a recovery path. If management can’t explain when large deals will be made up, or if AI-related revenue can’t cover the time gap created by the squeeze on traditional revenue, the market will continue pricing “execution mistakes” as “a transition pace that’s behind.” For investors, IBM’s core problem is no longer whether it has an AI story, but how much time its old cash flows can buy for the new story.

Click to learn about positions at Lyding BlockBeats

Welcome to join Lyding BlockBeats’ official community:

Telegram subscription group: https://t.me/theblockbeats

Telegram discussion group: https://t.me/BlockBeats_App

Twitter official account: https://twitter.com/BlockBeatsAsia