The Market Got the Earnings It Asked For. Then It Punished Them Anyway.

April 24, 2026

---

There's a particular kind of market cruelty that reveals itself in earnings season. A company beats. It raises guidance. The stock drops 18%. You've been here before, but it never stops feeling like a philosophical problem.

ServiceNow fell almost 18% Thursday — its worst single-session performance on record — despite reporting first-quarter earnings that came in ahead of Wall Street's estimates. The company's CFO Gina Mastantuono acknowledged she "took a little bit of incremental conservatism" into guidance due to Middle East developments, and the market heard "conservatism" and reached for the sell button with both hands. Salesforce sank 8.7%, Workday dropped 9.4%, Adobe fell 7%. The iShares Expanded Tech-Software ETF — IGV — dropped roughly 6% on the day and is now down about 19% for the year.

IBM fared slightly better only in the sense that an 8% decline is less cinematic than 18%. IBM beat on adjusted EPS by $0.10 and came in $300 million above revenue expectations, then reiterated its 2026 guidance unchanged. CFO Jim Kavanaugh noted on the call that the company has never raised guidance in a first quarter. The market treated this candid piece of calendar context as a confession of failure.

Stop and sit with what actually happened here. Two enterprise software companies posted results better than expected on every fundamental metric that used to matter — earnings, revenue, backlog, margins — and collectively lost somewhere north of $60 billion in market cap in a single session. The question worth asking is not whether the sell-off was rational. It clearly was, in a mechanical sense. The question is what it tells you about the scaffolding beneath all these valuations.

What the market is really repricing isn't Q1 2026. ServiceNow's net new ARR hit a record $820 million, up 47.5% year over year. RPO ended the quarter at $27.7 billion, up 23.5% in constant currency. That's not a business in trouble. What's being repriced is the assumption that geopolitical friction — war-related deal delays in NOW's case — is a one-quarter headwind rather than a structural tax on software spending cycles. Once that shift takes hold in the narrative, every number gets filtered through a new lens.

---

Then there's Tesla, which executed a different kind of earnings theater entirely.

Shares initially rose about 4% in after-hours trading when the EPS print crossed the tape. Then CFO Vaibhav Taneja said capex would top $25 billion for 2026 — up from $20 billion flagged just last quarter, and up from $8.6 billion in all of 2025. He also warned that the spending ramp will likely mean negative free cash flow for the rest of the year. The after-hours gains turned into a 3.6% decline by Thursday's close.

There is a version of this story where you are impressed. Tesla is simultaneously running six new production lines across vehicles, robots, energy storage, and battery manufacturing — Cybercab, Semi, the Houston Gigafactory, a dedicated Optimus facility. The robotaxi operation has expanded to Dallas and Houston. There's a $3 billion research chip fab in Texas being built jointly with SpaceX and xAI. Energy storage margins hit a record 39.5%. This is not a company standing still.

But Musk's own words undercut the urgency his capex commitment was meant to project. On whether robotaxi revenue would be "super material" this year, he said probably not — but likely significant "next year." On unsupervised FSD reaching customer vehicles, he said he was "just guessing," and offered Q4 as a rough estimate. When the person most responsible for the company's $1.45 trillion valuation is openly guessing at timelines on his flagship product during the earnings call, investors who bought the autonomy thesis are permitted to feel mildly unsettled.

---

The market's Tuesday/Wednesday relief rally — the S&P at 7,137 and the Nasdaq at new all-time highs after Trump extended the U.S. ceasefire with Iran — looks considerably more fragile from Thursday's vantage point. The S&P gave back 0.41% on Thursday and finished at 7,108. The Nasdaq dropped 0.89%. Not catastrophic. But the structure of the session — intraday records followed by afternoon retreat — has been the recurring shape of every attempted breakout this year.

The real throughline in all of this isn't the geopolitical uncertainty, though that's the convenient explanation being offered for every miss and guidance caveat. The actual tension is simpler and more uncomfortable: the companies spending aggressively into an AI future (Tesla, $25 billion) are being punished for the cash burn that requires, while companies delivering AI-adjacent software revenue at scale (ServiceNow, record ARR) are being punished for any language that implies the enterprise deal cycle might slow. There is no correct posture. Either you're burning too much cash building the future or you're exposed to the present slowing down.

Texas Instruments offered the session's one clean vindication, surging 18% — its best single-day performance since October 2000 — after beating expectations on the back of industrial and data center demand. Semiconductors keep their clarity while software loses its narrative and EVs try to buy their way into a robot future. The leaderboard tells you something.

Certuity's Scott Welch noted earlier this week that the market "was not cheap before the war started," and the recent rally had only brought it back slightly past breakeven for the year. That observation reads differently now. The ceasefire gave everyone permission to stop worrying. Earnings season is not extending the same courtesy.