This week the market has been busy assigning grades to capital expenditure. Microsoft got a 16% pop after Azure crossed $100 billion in annual revenue and its spending forecast came in short of the doomsday number. Meta got taken down about 9% for guiding 2026 capex to $135–145 billion and reporting free cash flow of $784 million — down from roughly $8.6 billion a year earlier. Two hyperscalers, one scoreboard: how much are you burning, and is the burn producing anything yet.
Then there is Garmin, which reported on Wednesday morning and does not participate in that conversation at all. Second-quarter revenue was $2.02 billion, up 11% and a company record. Operating income was $615.5 million, up 30%. That works out to a 30.4% operating margin, on gross margin of 62.4%. Pro forma diluted earnings were $2.81 a share, up 29% year over year, against roughly $2.29 expected. The stock rose about 17%.
Management then raised the full year: approximately $8.05 billion in revenue, $10.00 in pro forma EPS, and full-year gross margin of about 59.7% — 120 basis points above the prior outlook. Raising a margin forecast mid-year is a rarer signal than raising a revenue forecast. Revenue can be bought. Margin has to be earned in the mix.
Fitness did the work
The fitness segment produced $756.8 million in revenue, up 25% year over year, or 37.4% of the company. Its operating income was $277 million, up 40% — growing profit at roughly 1.6x the rate of its sales. Outdoor, the segment that carried Garmin through the last cycle, was modestly down. That rotation matters more than the headline: the growth is coming from the part of the business with the recurring-behavior hook, not the part that depends on somebody replacing a handheld GPS.
Eight days before earnings, Garmin bought TrainingPeaks and TrainHeroic — endurance and strength coaching platforms based in Louisville, Colorado, bringing roughly 120 staff. Terms were not disclosed. It is a small deal by dollar volume and a large one by intent: the hardware sells once, the training plan and the coach relationship sell every month.
Our take: Garmin is the control group for the AI capex trade. Same quarter, same tape, same investor base — and a business whose growth requires no gigawatts, no interconnect queue and no depreciation schedule that has to be quietly stretched from 15 years to 25 to keep the numbers presentable, as Microsoft just did. Garmin bought back $43 million of stock in the quarter with about $448 million still authorized and paid a dividend, because it can: cash generated is cash available. Hyperscalers are converting cash into concrete on the promise of returns nobody can yet measure. That may well be the right trade. But the market spent two days this week violently repricing two companies on the credibility of their spending plans, and in the middle of it a watch company quietly earned 30 cents on the dollar and got the cleanest re-rating of the bunch. When capital is expensive, the businesses that don’t need much of it stop being boring.
What to watch
- Whether fitness momentum is product or cycle. A 25% quarter followed new launches. The test is the second quarter after the launch window closes, not the first.
- Outdoor stabilizing. It is still the margin anchor. Fitness growth papering over a structural outdoor decline is a different company than fitness growth on top of a flat one.
- What TrainingPeaks does to the P&L. Subscription revenue at software gross margins, layered onto 62% hardware gross margin, is the whole reason to do the deal. If it doesn’t show up in the mix within a few quarters, it was a defensive buy.
- The gross margin guide. Management raised full-year gross margin to about 59.7% while tariffs and memory-component costs are squeezing every other hardware maker — the pressure that just pushed Samsung’s phone division into a loss. Holding that line is the number to check next quarter.