Khan Capitals | April 2026
Key Takeaways
- ServiceNow recorded its worst single-day decline on record, falling roughly 18% on 23 April despite a reported 22% growth in subscription revenue, as investors rejected a Middle East deal-slippage narrative and re-rated the business on AI disruption risk.
- IBM shed close to 9% even after beating on both the top and bottom line, with management’s reaffirmed full-year guidance taken as an implicit downgrade versus a market that had been pricing acceleration, not stabilisation.
- Texas Instruments rallied close to 19% the same session, its strongest one-day move since October 2000, as analog chip demand tied to AI data centres and reshored industrial capex drove an upside revenue guide of $5.0 to $5.4 billion for the second quarter.
- The 23 April session delivered one of the widest intra-sector dispersions inside US technology in recent memory, with software multiples compressing while analog and data centre silicon re-rated higher, ending the Nasdaq’s 13-session winning streak, the longest since January 1992.
- The signal is structural, not tactical: large-language-model agents are migrating from tailwind to competitive threat for seat-based enterprise software, while compute infrastructure absorbs the associated AI capex windfall. Beating estimates no longer clears the bar if the business model is the question.
A Record Rally Broken by a Record Rout
For twelve consecutive sessions beginning on 2 April, the Nasdaq Composite climbed without pause, clawing back the Iran-war drawdown and pushing technology indices to fresh highs. By 17 April, the index had posted a 17.7% gain and its thirteenth straight green close, the longest positive streak since January 1992. The reflexive narrative framed the melt-up as relief: a two-week ceasefire between Washington and Tehran, a reopening of Gulf shipping lanes, a Federal Reserve apparently willing to look through the oil-driven inflation impulse, and an earnings season led by the largest banks that had flattered trading revenue and reaffirmed credit quality.
That reflexivity ended in a single session. On Thursday 23 April, ServiceNow shares collapsed around 18% in the largest intraday decline in the company’s public history. IBM, reporting the same evening as ServiceNow, lost close to 9%. The iShares Expanded Tech-Software sector ETF (IGV) closed down over 3.5%, its worst session in three months, while Texas Instruments surged almost 19%, its best one-day return since the dot-com peak in October 2000. The Nasdaq snapped its streak and finished broadly lower as a straightforward single-name disappointment metastasised into a sector-wide re-rating.
The instinctive reading, amplified by ServiceNow management on the call, is that geopolitics delayed deal signatures and compressed subscription growth. That framing does not survive contact with the Texas Instruments print, which showed the same operating environment delivering accelerating analog demand from data centres and industrial customers. When investors reject the macro excuse inside one session and price a structural change in another, the market is telling observers something about business models, not shipping routes.
ServiceNow: When Beating the Print Is Not Enough
ServiceNow’s first-quarter results, released after the close on 22 April, were in most respects unremarkable for a company trading on a forward revenue multiple close to eleven. Subscription revenue rose 22% year on year to $3.67 billion, in line with analyst consensus. Current remaining performance obligations, the most-watched bookings proxy, grew in the high teens. Free cash flow margins remained over 40%. Chief executive Bill McDermott used the earnings call to raise the 2026 AI revenue guide to $1.5 billion, up from the $1 billion marker set at the start of the year.
Against this, the stock lost nearly a fifth of its market capitalisation in one trading day. The proximate catalyst was a colour remark from the chief financial officer acknowledging that a handful of large on-premise closures in the Europe, Middle East and Africa region had slipped, with the Iran conflict cited as a contributing factor. For a business that had been priced for continued acceleration rather than hairline maintenance of the growth algorithm, the combination of in-line subscription growth and any deceleration commentary was sufficient to reset the valuation.
The more revealing question is why institutional positioning proved so fragile. Coming into the print, ServiceNow carried a premium within large-cap software, was widely flagged by sell-side strategists as a primary AI-platform beneficiary, and sat in virtually every growth and quality factor portfolio. The tape response suggests a late-stage crowding pattern: buy-side incumbency leans heaviest on the names perceived as most insulated from AI substitution risk, so the marginal disappointment in exactly that cohort produces the most extreme response. The same mechanism compressed Salesforce, Workday, Snowflake and Datadog in sympathy.
IBM: The Penalty for Reaffirmation
IBM’s quarter, reported the same evening, beat consensus on both revenue and adjusted earnings per share. Software revenue grew at a high-single-digit pace, with Red Hat and automation contributing outsized growth. The message that unsettled the market was the decision to reaffirm, rather than raise, full-year revenue and free cash flow guidance of around $13.5 billion. Management defended the decision by pointing to foreign-exchange headwinds and the conservatism appropriate to a more volatile macro backdrop.
The market’s response, a 9% single-day decline, highlights a pattern visible throughout this earnings season: the hurdle has moved from beating the quarter to demonstrating sufficient evidence of AI-driven revenue acceleration to justify continued multiple expansion. IBM’s watsonx bookings, while showing meaningful sequential growth, did not clear that implicit bar. The stock traded as if investors had concluded that the AI build-out is predominantly an infrastructure story, and that enterprise consulting and middleware are neither asymmetric beneficiaries nor rapidly disrupted victims. That intermediate outcome is precisely the worst positioning in a sentiment-driven market.
Texas Instruments: The Silicon Counterpoint
The cleanest contradiction to the software tape sat one section over in the sector map. Texas Instruments reported first-quarter revenue of $4.83 billion, up 19% year on year and over $300 million ahead of consensus. Adjusted earnings per share of $1.68 landed roughly 32% above the Street. The company guided the second quarter to revenue of $5.0 to $5.4 billion, approximately 10% above consensus at the midpoint, with earnings of $1.77 to $2.05 against expectations near $1.58.
Crucially, management attributed the strength to exactly the segments that software management blamed for weakness: industrial and data centre demand. Analog chips, the unglamorous building blocks that regulate power delivery, translate signals, and manage heat inside AI accelerator systems, have re-rated from a late-cycle commodity category into a secular growth line item. The implicit message to the market is that AI capex flows are both larger and more durable than the consensus model implies, but that the distribution of those flows is narrower than most software bulls have assumed.
Anatomy of a Divergence
| Company | Segment | Headline Number | Guidance Posture | Share Reaction (23 Apr) |
|---|---|---|---|---|
| ServiceNow | Enterprise SaaS | Subscription revenue +22% to $3.67bn | Mid-East deal slippage cited; AI guide raised to $1.5bn | Approx. -18% (worst on record) |
| IBM | Hybrid cloud and consulting | EPS beat by $0.10, revenue beat by ~$300m | Full-year guidance reaffirmed, not raised | Approx. -9% |
| Texas Instruments | Analog and embedded semis | Revenue +19% to $4.83bn; EPS $1.68 vs $1.27 | Q2 guide $5.0 to $5.4bn; ~10% above consensus | Approx. +19% (best since Oct 2000) |
| Salesforce / Workday / Snowflake | Horizontal and vertical SaaS | No print; traded in sympathy | Not applicable | Range of -4% to -7% |
| IGV Software ETF | US software composite | Largest constituent down ~18% | Not applicable | Approx. -3.5%, worst session in three months |
The AI Agent Thesis Moves from Tailwind to Threat
For the better part of two years, the consensus framing within enterprise software has been that generative AI is additive: it drives new product cycles, higher attach rates, and premium pricing tiers that expand the addressable market. ServiceNow’s Now Assist, Salesforce’s Agentforce, Microsoft’s Copilot family and Workday’s Illuminate have all been pitched as revenue pools layered on top of seat-based subscription businesses, not replacements for them.
The tape is now testing a different thesis. Large-language-model agents that can read documentation, execute workflows, query databases and integrate with enterprise systems raise the possibility that a substantial fraction of seat-based software spend is substitutable by direct model access. An organisation can plausibly replace several categories of user licence with a combination of foundation-model API spend, a thin orchestration layer and a smaller human operator headcount. The economics are not yet proven, but the direction of travel is sufficient to compress multiples on businesses whose value depends on the continuation of seat-count growth.
This is the structural backdrop against which ServiceNow’s slipped on-premise deals look less like a Middle East story and more like a demand-hesitation story. Chief information officers weighing a multi-year on-premise renewal must now model the probability that agent-based architectures will displace a portion of that workload inside the contract window. Delay is rational even in the absence of a war. For a company priced at over ten times forward sales, any lengthening of the sales cycle is punished aggressively.
Why Silicon Gets the Other Side of the Trade
The same forces that compress SaaS multiples inflate demand for the compute infrastructure on which agent workloads run. A foundation-model agent performing the function of ten seat-based software licences requires meaningful inference compute, high-bandwidth memory, advanced packaging, and the analog componentry that ties the system together. Texas Instruments sits at the unglamorous end of that supply chain, but its addressable market expands roughly in proportion to the wafer-starts consumed by the AI accelerator complex.
The aggregate picture is a widening bifurcation inside technology. Hyperscalers continue to lift capital expenditure guidance: Alphabet’s cloud backlog has reached $243 billion, Microsoft’s commercial remaining performance obligations sit above half a trillion dollars, and Tesla has just added $5 billion to its 2026 capex plan. The downstream beneficiaries are accelerator designers, memory vendors, foundries, advanced packaging specialists and analog silicon. The middle of the stack, where horizontal enterprise software has historically captured disproportionate operating leverage, is the portion now being re-rated.
What the Market Is Underappreciating
Three observations follow from the 23 April session that mainstream commentary, fixated on the Middle East narrative, is at risk of underweighting.
First, the reaction function inside software has changed. Throughout 2024 and 2025, in-line subscription growth and stable net-retention were sufficient to sustain premium multiples. That clearing bar has risen. Investors now require visible evidence that AI-linked revenue is growing faster than the core business, and that agent-native product lines are both defensible and high-margin. In-line is the new miss.
Second, the crowded positioning in high-quality enterprise software compounds the downside convexity. Factor-model portfolios, thematic exchange-traded funds and long-only growth mandates have all concentrated in a narrow set of winners. Liquidity is abundant on the way up and scarce on the way down. The skew of outcomes from a similar disappointment at a peer name over the coming week is meaningful.
Third, the market continues to price AI capex intensity as if it will obediently recycle into the revenue lines of the software incumbents. The 23 April evidence is the first genuine data point suggesting otherwise: the capital is flowing predominantly to silicon, hyperscalers, and the hyperscalers’ internal AI stacks, rather than to the horizontal SaaS layer that historically captured such spend. If that reading is correct, forward consensus for the large-cap software universe is too high on revenue, too generous on margins, and too complacent on multiples.
Investor Implications
The cross-asset implications of this single session deserve careful decomposition. The equity implications are most direct. Allocators with overweight positions in horizontal enterprise software may wish to review exposure concentration, stress-test portfolios for a scenario in which the IGV multiple compresses another 15 to 20% from current levels, and consider whether existing AI-theme baskets are genuinely diversified across the stack or merely long the same twenty names via different tickers. The tactical pair of long analog and data-centre silicon against short horizontal SaaS has carried well year-to-date and may continue to do so into the 29 April mega-cap earnings cluster.
Within fixed income, the implication is more subtle. Investment-grade spreads for large-cap software issuers, which have tightened alongside their equity multiples, are exposed to a softening of revenue visibility. The issuers themselves remain cash-generative and carry strong balance sheets, but the spread cushion over equivalent industrial names has compressed to levels that no longer obviously compensate for the uncertainty now visible in the equity market. Credit-focused allocators may find better risk-adjusted carry in the utility, industrial and select semiconductor segments that capture AI capex without exposure to SaaS-model disruption risk.
From a cross-asset positioning standpoint, implied volatility in single-name software remains underpriced relative to the potential for further earnings-driven gapping. The VIX complex, which traded in the mid-teens heading into the week, still reflects the Nasdaq’s recent momentum rather than the tail risk now emerging at the sector level. Hedgers focused on concentration in the large-cap growth cohort may find QQQ and IGV puts attractively priced for rolling exposure through the 29 April Alphabet, Microsoft, Meta and Amazon prints. Gold and the US dollar, both bid through the Iran-war period, continue to offer portfolio ballast against the combination of geopolitical tail risk and equity-sector rotation visible in current price action.
Conclusion
The week ending 24 April delivered a single-session lesson that is likely to echo through the remainder of earnings season and into the Federal Reserve’s 28-29 April meeting. The record-setting Nasdaq rally was broken not by a macro catalyst, not by a Federal Reserve surprise, and not by a new geopolitical shock, but by the re-rating of two of the most widely held enterprise software names. The immediate cause was an earnings print that, on headline metrics, would historically have been acceptable. The deeper cause is a shift in how the market values businesses whose growth algorithm depends on the continuation of a seat-based software paradigm that AI agents increasingly threaten to collapse.
Texas Instruments’ counterpoint clarifies the trade rather than contradicts it. AI capex is not slowing: it is concentrating. Compute, networking, power and the analog silicon that surrounds them are absorbing the spend that the market had assumed would flow more evenly across the technology stack. The horizontal middle layer, long the market’s preferred expression of software operating leverage, now faces a credible disruption thesis that has moved from theoretical to testable. Institutional allocators entering the 29 April mega-cap earnings cluster must contend with a market whose reaction function has changed, whose positioning remains crowded, and whose patience with incremental in-line quarters has visibly run out.
Sources: CNBC, Software stocks plunge on ServiceNow, IBM results; Bloomberg, ServiceNow posts lukewarm sales on Middle East deal delays; The Motley Fool, Solid earnings from IBM and ServiceNow fail to quell AI concerns; Benzinga, IBM shares drop despite Q1 earnings beat; CNBC, Texas Instruments stock soars on Q1 earnings as AI demand jumps; Yahoo Finance, Nasdaq scores longest winning streak since 1992; Schaeffer’s Investment Research, Nasdaq 13-day win streak.
Related Reading: The divergence explored above builds on themes developed across the Khan Capitals catalogue. The broader AI capital-expenditure cycle and its distribution across the technology stack is charted in AI Capex Boom: $600 Billion and Counting, while the software sector’s vulnerability to sharp sentiment shifts was first visible in Tech Volatility in 2026: Software Stocks Drop $400B in a Week. The capex-guidance reset that now frames the 29 April mega-cap cluster is dissected in Tesla’s $5 Billion Capex Surprise Opens the Mag 7 Earnings Reckoning. The macro backdrop is covered in The Warsh Doctrine: Regime Change Pricing Arrives Ahead of the April FOMC, and the strategic framework for navigating the year sits in 2026 Outlook: Navigating Tariffs, AI, and Geopolitical Risk. For the next leg of this story, see our analysis of Dell’s $51.3 billion AI server backlog.
Leave a Reply