Apple reported fiscal second-quarter results on the evening of April 30 and gave Wall Street the cleanest possible read of the iPhone-and-services thesis. Per the company’s Q2 FY26 press release and the CNBC writeup:
- Revenue: $111.2 billion, up 17% year-over-year — a March-quarter record.
- Diluted EPS: $2.01 vs. $1.96 expected, up 22%.
- iPhone revenue: $57.99 billion, up 22%, a March-quarter record.
- Services revenue: $30.98 billion, up 16%, an all-time record across any quarter.
- Services gross margin: 49.3% vs. 48.4% expected.
- Buyback authorization: an additional $100 billion, with the dividend up 4% to 27 cents.
That is the unambiguously bullish half of the call. The other half — the half that this site exists to read — is a half that simply did not happen. Compare the verbatim sentences from the four most recent megacap earnings transcripts:
- Microsoft, April 29, Amy Hood: “We continue to evolve how we operate to increase our pace and agility, and therefore, we expect headcount to decrease year over year.” (transcript)
- Amazon, April 29, Andy Jassy/Brian Olsavsky: the call paired record AWS growth (+28%, the fastest in 15 quarters) with a reaffirmation of the 16,000 corporate roles cut in January as the operating-leverage path. (CNBC)
- Meta, April 23 memo: 8,000 jobs cut, ~10% of workforce, plus 6,000 open requisitions removed. (CNBC)
- Snap, April 22 memo: 1,000 jobs, ~16% of staff, after activist investor pressure on “over-hiring.”
Apple, April 30: nothing. No buyout program. No “operating-leverage” line. No layoff press release queued for the morning after the print. The only headcount-adjacent comment came from incoming CEO John Ternus, joining his first earnings call to thank Cook, praise the company’s “financial discipline”, and tease an “incredible roadmap” without quantifying anything.
How Apple gets to grow 17% without the buyout playbook
The structural answer is that Apple is not running the AI-capex playbook in the first place. Microsoft is guiding to $190B of FY26 capex. Amazon is guiding to $200B. Meta is in the $70–75B band. Apple’s full-year capex is still tracking around $11–13 billion — roughly one-fifteenth of Microsoft’s number, against a comparable revenue base.
Apple does not own the LLM substrate. They pay Google about $1 billion per year for a custom 1.2-trillion-parameter Gemini model to run the rebuilt Siri shipping later this year. They have no hyperscaler cloud business that needs filling with H100s. They are not depreciating $130 billion of GPUs over four years. So the income-statement identity that forced Microsoft, Amazon, and Meta to put labor-cost reductions in the guide simply does not bind here.
The other half of the answer is that iPhone is still doing the work. The 22% iPhone-revenue growth rate — on top of the iPhone 17 cycle that lapped a comparatively soft year-ago quarter — is a number none of the cloud-AI megacaps have on a single product line. Services compounding 16% on top of 16% is a number none of them have on a recurring-revenue line. Together, those two lines are still the entire company.
Cook’s last call as CEO
Tim Cook’s step-down was announced April 20 — Cook becomes executive chairman; John Ternus, currently SVP of Hardware Engineering, becomes CEO on September 1. The Q2 print on April 30 was Cook’s last earnings call in the chair. The fact that he ended his run with a clean beat, an all-time services record, and zero AI-attributed-headcount language is the single most coherent valedictory paragraph any modern Fortune-10 CEO has delivered.
It is also a lateral comment on the rest of the megacap cohort. Cook spent fifteen years not running the AI-platform race. The other megacaps spent the last 24 months building data centers as fast as they could pour concrete. As of this week, Cook leaves with revenue +17% and headcount approximately flat. The other megacaps are exiting Q1 calendar with revenue growing slower and labor-cost lines being explicitly trimmed to fund the GPU bill.
You can frame that two ways. The bear frame: Apple is going to have to start running the same playbook eventually, because at some point the on-device AI experience becomes the iPhone moat, and at that point Apple has to either license much harder from Google or finally build its own foundation-model training stack — and at that point the capex line lights up and the labor-cost line follows. The bull frame: Apple already paid the optionality premium to Google ($1B/yr is a rounding error against $191B of Q2-trailing operating cash flow), and the hardware moat plus services compounding does not actually require a Microsoft-shaped capex line to keep printing 17% growth quarters. We don’t know which one is right.
What LostJobs is watching
- The Ternus first-call posture on July 30. The Q3 FY26 print is the one Ternus owns from start to finish. If he uses the call to introduce any version of the Hood “headcount to decrease year over year” language, the Apple anomaly closes and the megacap cohort goes 5-for-5 on AI-attributed labor-cost guidance. If he doesn’t, Apple stays the outlier through fiscal 2026.
- WWDC 2026 in June. The “personalized Siri” line Cook has now repeated on three earnings calls finally gets a date and a feature list. The interesting question is not whether Siri ships — it has to — but whether Apple’s framing of the underlying engineering investment requires more headcount or less. The Gemini deal is the implicit answer to that question, but Apple has never said it out loud.
- Whether the missing “headcount” sentence becomes a durable Apple brand. In a calendar year where every other Fortune-50 board has been handed the Microsoft-pioneered template, the company that publicly doesn’t run it has a recruiting and PR moat that is starting to be worth real basis points in the talent market. Cook never named that moat out loud either. Ternus inherits the right to.
The dry coda
The single most precise omission in the entire Q2 release wasn’t on the income statement. It was on the line where every other megacap CFO this earnings cycle inserted a sentence about labor cost. Apple’s was a hyphen. Translate that out of investor-relations English: we still grow without firing anybody, and the next CEO is going to find out whether that survives contact with the AI workload. The 17% revenue beat is the front-end of that translation. The empty headcount-guidance line is the back-end. Cook walks out on both halves working at the same time.