IBM's 25% one-day crash: the mechanism (customers front-running memory prices out of a fixed IT budget) matters more than the headline miss)

IBM had its worst single-day drop ever on Tuesday — down \~25.2%, from $290.23 to $217.07. A bigger one-day percentage loss than Black Monday 1987. About $69B of market cap gone in a session (WSJ). What's unusual is why, and I think the mechanism is more interesting to argue about than the number.

What they actually did. IBM pushed out preliminary Q2 numbers \~8 days early, with a CEO letter filed as an 8-K. Preliminary revenue $17.2B vs \~$17.86B expected; non-GAAP EPS $2.93 vs $3.01. Companies don't usually pre-announce a miss unless they'd rather control the narrative now than let it drop cold on the print date.

The mechanism , and I'll flag up front this is substantially IBM's own framing. In late June, memory/server prices were about to jump, so IBM's enterprise customers , big companies with roughly fixed IT budgets , rushed to buy that hardware first to lock in supply. Because the budget is fixed, that money came out of the "optional" half... which is exactly what IBM sells: z17 mainframe refreshes and the high-margin Transaction Processing software on top. Big, deferrable purchases. So the story is IBM didn't lose to a competitor; it got crowded out of its own customers' budgets by a different line item.

The segment split is consistent with that (though it doesn't prove causation, some of the -7% could be competitive/secular):

\- Infrastructure -7% (the mainframe/TPS hit)

\- Software +5%, Consulting \~flat, Red Hat +11% sequential

\- IBM's own distributed server/storage line +37% — it caught a sliver of the very wave that drowned its mainframe

\- Total sales still grew \~1%

The CEO didn't spin it. Krishna, verbatim in the 8-K: "this quarter we faltered. We did not adapt and move quickly enough, and numerous large deals failed to close on the timelines we expected." He also said IBM "did not anticipate the magnitude of the capex reprioritization."

It wasn't just IBM: MSFT \~-3%, CRM \~-4%, IGV -2% on the same read-through. Cybersecurity went the other way (PANW rallied) — Krishna flagged cyber pulling client attention/spend.

Honest valuation. The easy number is misleading. Trailing GAAP EPS \~$11.31 looks like \~19x at $217, but it's flattered by a one-time \~$1.4B tax benefit in Q4'25, so I wouldn't anchor on that P/E. Cleaner lens is cash: \~$11.6B annual FCF, so at $217 you pay roughly $18 per $1 of annual free cash flow. Balance sheet is fine (Altman Z \~3.0, \~3% dividend). And the \~$315 sell-side targets floating around all predate the crash — stale, not "implied upside." The tape, though, is broken: death cross, \~21% below the 200-day, near the 52-week low, RSI \~33.

The fork. Full Q2 lands July 22, 8 days out, and it reduces to one binary: were those large deals delayed or dead? Delayed = timing air pocket, fixed budgets normalize, deferred deals close in H2, and -25% on a \~4% revenue miss is an overreaction. Dead = enterprise budgets structurally bending toward AI hardware and away from mainframe cycles, in which case the multiple should compress and this is the first leg down. The 8-K wording leans "delayed" — but that's what you'd say either way.

So, genuinely asking: do you read the June front-running as a one-quarter timing air pocket that reverses once budgets reset — making July 22 a buyable catalyst — or as the first crack in enterprises permanently reprioritizing away from mainframes, which means you wait past the 22nd before touching it?

(No position — this is on my watchlist, not a recommendation.)