When International Business Machines warned investors on a Tuesday morning in July that its second-quarter results would fall well short of expectations, the shockwave didn’t stay contained to one stock. Shares plunged as much as 26% in early trading — on pace, by Motley Fool’s reckoning, for the worst trading day in the company’s history — while cybersecurity names like CrowdStrike and Okta jumped more than 10% and memory-chip makers Micron and Sandisk rose 5%. For a Canadian audience tracking IBM’s cross-listed shares on the NEO exchange (IBM.NE) alongside its primary New York listing, the episode is a useful, if uncomfortable, window into how fragile the AI trade has become — and how one company’s stumble can simultaneously crush some sectors and supercharge others.
What Actually Happened
IBM’s preliminary numbers, released ahead of its formal July 22 earnings report, showed revenue growing just 1% to $17.2 billion with adjusted earnings per share of $2.27 — both well below Wall Street’s consensus of $17.85 billion in revenue and $3.02 in EPS, according to Motley Fool reporting that cited the company’s own shareholder letter. When the formal results landed, the final figures came in slightly different — adjusted EPS of $2.93 against a Zacks consensus of $2.93 (in line) and revenue of $17.16 billion versus an estimate of $17.17 billion, per TradingView’s account of the earnings call. CEO Arvind Krishna explained the shortfall plainly: in the final weeks of June, clients diverted their capital spending toward servers, storage and memory — essentially stockpiling physical AI infrastructure ahead of expected price increases — leaving a string of large software deals unclosed.
Krishna also pointed to a second, less expected culprit: cybersecurity anxiety. He told press that the emergence of Anthropic’s ‘Mythos Preview,’ which reportedly surfaced ‘thousands of zero-day vulnerabilities across critical infrastructure,’ had made corporate clients freeze new technology spending while they assessed exposure — ‘They’re pausing on new deals until they know,’ as Krishna put it.
Who Won, Who Lost
The reaction across markets following IBM’s warning reads like a map of where AI capital is actually flowing right now. Memory and storage companies — whose products clients were reportedly racing to buy before anticipated price hikes — were the obvious beneficiaries. Even accounting for a later pullback, Micron’s stock had gained 694% over the prior year, while Sandisk had soared a startling 3,750%, according to Motley Fool’s July 14 report. Cybersecurity firms CrowdStrike and Okta also caught a bid, as the Mythos-driven anxiety Krishna described translated into expectations of higher security spending.
On the losing side sat enterprise software broadly. ServiceNow fell 4% and Microsoft fell 1% in the immediate aftermath, and both had already endured a rough stretch — Microsoft down 23% and ServiceNow down 43% over the preceding year, per Motley Fool. The fear wasn’t really about IBM’s specific execution; it was that IBM’s commentary confirmed a broader reallocation of enterprise budgets away from software licensing and toward physical AI infrastructure, a shift that could squeeze any company dependent on big-ticket software deals.
Inside IBM’s Numbers: Delay or Demand Destruction?
The central question on IBM’s earnings call, as TradingView’s detailed writeup makes clear, was whether the lost deals were merely deferred or genuinely gone. CFO James Kavanaugh offered a telling breakdown: roughly 80% of IBM’s annual software revenue is recurring and continued to grow in healthy fashion, while the more transactional, enterprise-license portion of the business — the part most exposed to one-time capital budget decisions — was hit hardest, with transaction processing revenue down 9%. Red Hat revenue, by contrast, rose 11%, and data revenue climbed 18% at constant currency. Annual recurring revenue reached $24.6 billion, up 8% year over year.
Infrastructure told a similarly split story: overall infrastructure revenue fell 7%, driven by a steep 42% decline in IBM Z mainframe revenue, but distributed infrastructure jumped 37% — its best growth quarter on record — with IBM exiting the quarter carrying roughly $500 million of backlog in that category. Krishna insisted the mainframe franchise itself remains strong, noting the z17 product cycle is running at nearly 130% of its predecessor’s pace on a comparable basis, and said IBM saw no evidence of clients abandoning the mainframe altogether.
Pressed by analysts from Evercore ISI and Jefferies on whether the softness was timing or something worse, Krishna said about a third of the slipped large capital-expenditure deals had already closed within the first three weeks of the third quarter — his central piece of evidence that the problem was a scheduling hiccup rather than a change in underlying demand. IBM trimmed its 2026 constant-currency revenue growth outlook to 4%-5% from a prior forecast of 5%-plus, while still targeting roughly $1 billion of year-over-year free cash flow growth and raising its operating pretax margin expansion target to 100 basis points.
The Bigger AI Market Story
Motley Fool’s broader analysis frames IBM’s stumble as a symptom of where the AI investment cycle currently sits. Early AI winners were chip and cloud infrastructure providers — Nvidia, AMD, Broadcom, and cloud operators like Amazon Web Services. More recently, the spotlight has shifted toward memory and storage suppliers, as AI agents and real-world deployment drive a surge in demand that has outstripped supply, pushing prices higher and lifting shares of SK Hynix, Micron and Sandisk. IBM’s own warning was, in effect, a confession that it got caught on the wrong side of that reallocation — its enterprise customers chose to lock in hardware ahead of price increases rather than sign new software and consulting contracts with IBM.
This is not IBM’s first brush with AI-related stock turbulence in 2026. Back in late February, news that Anthropic’s AI tools could tackle COBOL coding tasks — a core part of IBM’s legacy consulting business — sent the stock tumbling, leaving it roughly 20% below its 52-week high at the time, according to an earlier Motley Fool piece. That drawdown pushed IBM’s price-to-earnings ratio down to about 23 times earnings, below its five-year average of nearly 30 times, though its price-to-sales and price-to-book ratios remained above historical norms — a reminder that even a battered IBM wasn’t trading at obvious bargain levels by every measure.
Meanwhile, IBM continues to position itself in the next frontier of computing: quantum. Krishna reiterated on the July call that IBM plans to invest more than $10 billion in quantum computing over the next five years and remains on track to deliver a large-scale, fault-tolerant quantum computer by 2029. Motley Fool’s broader survey of quantum stocks places IBM among a small group of ‘giants’ — alongside Microsoft, Alphabet and Nvidia — running serious quantum research programs as a side project funded by already-profitable core businesses, distinct from pure-play, loss-making quantum start-ups like IonQ and D-Wave.
Our Take
In our view, the most important thing IBM’s July shock revealed isn’t about IBM specifically — it’s about how unevenly distributed the gains from the AI boom have become, even among companies that are all, nominally, ‘AI plays.’ Clients aren’t cutting AI spending; per Krishna’s own account, they are simply choosing where to spend it, and right now that choice favors physical infrastructure — chips, memory, servers — over software contracts and consulting engagements. That’s great news for Micron and Sandisk shareholders and a genuine headwind for software-heavy enterprises like IBM, ServiceNow and, to a lesser extent, Microsoft.
Whether IBM’s framing of this as pure timing rather than demand erosion holds up is, we think, the question that matters most over the next couple of quarters. Krishna’s evidence — a third of slipped deals already closing in early Q3 — is suggestive but not conclusive, and it came from the company itself rather than an independent verification. The durability of IBM’s recurring software base (that 80% figure Kavanaugh cited) does offer some cushion, and the mainframe cycle’s reported strength suggests IBM’s bread-and-butter enterprise relationships haven’t evaporated. But a downgraded revenue growth target, a sliding infrastructure segment, and a stock whose single-day crash was described as historic are not things any Canadian investor tracking IBM.NE should wave away as mere noise.
For Canadian businesses and consumers more broadly, the episode is a reminder that AI’s economic footprint is being built unevenly and that the companies supplying the picks and shovels — memory, storage, security — may continue capturing outsized value relative to the software and consulting layer for some time yet. Given how exposed Canadian pension funds, bank-run mutual funds and retail portfolios are to large U.S. technology names, swings of this magnitude in a company as systemically followed as IBM are worth watching closely, not as a verdict on any single stock, but as an indicator of where the broader AI capital cycle is actually headed next.
This is a Commentary piece: analysis and editorial perspective from Canadian Business News, clearly distinguished above from the reported facts it’s based on. It is not financial, investment, or legal advice.
Sources
