Why IBM just suffered its worst stock crash of all time—and what it says about the market’s two bubbles

AI-market “dual bubble” concerns surfaced as IBM crashed 25% after a Q2 revenue miss of $17.2B vs ~$17.9B consensus (-3.7% YoY expectation vs +5% anticipated) and adjusted EPS of $2.93 vs $3.02. Offsetting that, JPMorgan posted $21.2B net income (record for US banks) and Goldman reported an 84% earnings jump to $6.4B with revenues up 39%, highlighting a divergence between bank profitability and tech/IT earnings sensitivity. The article argues the bigger risk is an “earnings bubble” in AI-linked capex cycles, with profit downside often detected only after stocks fall—making IBM’s reaction a potential early crack rather than a valuation-only issue.

Analysis

The first-order winner is not “AI” broadly; it is earnings quality. A market that is now punishing small revenue misses this hard is likely to re-rate lower-duration, lower-growth tech and services names first, while rewarding balance-sheet compounding and fee engines that can absorb volatility. That makes JPM the cleanest large-cap beneficiary versus IBM, while GS is a more mixed expression: still helped by active markets now, but more exposed if risk appetite fades and the fee pool normalizes.

The second-order risk is contagion from one software miss to the broader enterprise IT complex. If this is a real earnings-bubble regime, the next damage should show up in bookings, renewal rates, and forward guidance across legacy software and systems integrators before it reaches the AI darlings; NVDA and GOOGL are not immediate shorts, but they become de-rating candidates if hyperscaler capex growth slows or ROI scrutiny tightens over 1-3 quarters. The tail risk is a sudden stop in private-credit and capital-markets support, which would hit GS, UBS, and HSBC faster than JPM.

Contrarian view: this may still be a single-name repricing, not a regime break. Strong bank prints argue liquidity is not yet tightening, so the market may be over-interpreting one weak software quarter into a sector-wide thesis. The thesis is falsified if subsequent software and AI capex commentary holds up; it is confirmed if another 2-3 large-cap IT names guide down over the next earnings cycle.

Key Decisions for Investors

• Long JPM / short IBM for 1-3 months. This is the cleanest relative-value way to express earnings quality versus earnings fragility; JPM should keep compounding if liquidity stays ample while IBM remains vulnerable to any further guidance disappointment. Exit if IBM reaccelerates organic revenue growth or JPM trading/NII momentum rolls over.
• Short GS / long JPM into strength over the next 1-3 months. GS has more beta to capital-markets exuberance and would likely de-rate faster if the market shifts from rewarding activity to penalizing volatility; JPM is the more defensive compounder. Best entry is on any post-earnings bid in GS rather than chasing weakness.
• Set a sector alert on IGV/XLK rather than forcing an immediate AI short. If two more large-cap software names cut forward guidance or booking expectations in the next earnings cycle, buy put spreads on the software basket as the higher-conviction contagion trade. Falsifier: stable renewal/consumption trends and no capex deceleration from hyperscalers.
• No outright short on NVDA yet; watch for hyperscaler capex guidance and AI ROI commentary over the next 1-2 quarters. If spend growth slows materially, NVDA becomes a better 6-18 month multiple-compression short than a near-term earnings miss trade.