Starbucks Corporation SBUX

Key data

The business

Starbucks sells coffee, and the habit of buying it. Some 40,000 stores worldwide, most company-run in the U.S., the rest licensed to operators who pay royalties and buy Starbucks beans. The money is made three ways: drinks sold over the counter at company stores, royalty and product streams from licensed stores, and packaged coffee on grocery shelves. The moat is ordinary and powerful: a store on every corner the customer already passes, a rewards app that holds their money and their order history, and a brand that lets a $6 latte feel normal. About a third of U.S. orders now arrive through the app before the customer walks in. The weakness the last two years exposed is that convenience cuts both ways: when the line got long and the app got crowded, the same density that pulls people in pushed them out.

The numbers

The shape is a trough and the climb out. Fiscal 2025, ended last September, was the bottom: net income more than halved to $1.86B and diluted EPS fell to $1.63, less than half the $3.58 earned in 2023. That was deliberate, the first full year of Brian Niccol's "Back to Starbucks" plan, which poured cash into labor and store hours and booked restructuring charges to do it.

The last two quarters are the turn. Operating income rose year over year even as reported revenue slipped, because U.S. comparable transactions grew 4.2% and average ticket 3.6%, the first time in two years that more people, not just higher prices, carried the top line. Traffic is the number that matters in a store business, and traffic turned. The June quarter's headline $1.05B flatters it: that figure carries a one-time gain from selling most of the China business, and the company's own adjusted number was $0.85 a share against the reported $0.91.

Nine months into fiscal 2026 the company has already earned $1.85B, all of last year's profit with a quarter still to run.

Margin is the whole argument. It fell from 16.3% in 2023 to 7.9% in 2025 and clawed back to 10.5% in June. Consensus has EPS at $3.75 by fiscal 2028, above the old peak, which needs margin climbing toward 15% while transactions keep compounding. You pay 23.7 times EBITDA for that, near the top of the twelve-year range and above the roughly 16.5 times peers fetch. What this memo believes and the tape does not: the transaction turn is the durable half and the margin the fragile half, and the next two prints of U.S. traffic decide which one wins.

Management

Niccol arrived in September 2024 from Chipotle with a clear mandate and has spent two years, and a lot of margin, executing it. The tell worth its weight came in November 2025, when board member Jorgen Vig Knudstorp, who turned around LEGO, put $994,500 of his own money into the stock, the single open-market purchase against eleven routine officer sales totaling $1.8M, plan status not disclosed on the sales. Capital allocation shifted: the company spent $1.27B on buybacks in fiscal 2024 then paused them entirely in fiscal 2025, redirecting cash to the turnaround and the dividend, and used China proceeds to cut long-term debt from $17.3B to $13.3B over the past year. Shareholders' equity is negative, a legacy of years of buying back stock above book, stated as filed. The pay-for-recovery logic is honest, and the guidance record since the reset is two beats after two misses.

How it fails or surprises you

Traffic keeps compounding (right tail). U.S. comparable transactions turned positive at +4.2% after two years of decline. Hold that above +4% for two more quarters and the market re-rates from hoping the turnaround works to knowing it does, and the $3.75 fiscal 2028 estimate starts to look low. First print: next quarter's U.S. transaction number.

Margin recovery stalls. Operating margin is 10.5%, still five points below its 2023 level, held down by the labor and hours Niccol added. If it sticks below 12% while wage costs climb, fiscal 2028 EPS misses badly and a 28-times multiple compresses fast. Watch consolidated operating margin each quarter.

The China gain flatters the turn. June's $1.05B profit included a one-time gain from divesting China; adjusted profit was $0.85 a share, not $0.91. Strip the sale and the operating recovery is real but slower than the headline, and the earnings base under that 28-times multiple is softer than it looks.

Closing thoughts

The recovery is mostly priced at 23.7 times EBITDA, the top of its own decade-plus range and half again what peers cost. The people on the other side own it because the turnaround is visibly underway and the brand rarely stays broken for long. What separates the views is one number: U.S. comparable transactions, which converts hope into proof. Sustained above +4% and the margin follows through operating leverage, the $3.75 estimate is conservative and the premium is earned. Fade back toward zero and you are holding a coffee chain at a peak multiple on trough-adjusted earnings, the fatter of the two tails today. The risk if traffic rolls over is a multiple that halves toward peers while EPS stalls near $2.50; the reward if it holds is a business earning back above its old peak. That the turn showed up in transactions last quarter, not just in cost cuts, is the first hard evidence the prior worry, a defensive coffee company wearing a growth multiple, is being answered rather than confirmed.

The bet is still that Americans go back to buying more cups of Starbucks, more often, and that Niccol has fixed why they drifted away. What breaks it is margin that will not climb or traffic that rolls back over. The one pair to watch: U.S. comparable transactions and consolidated operating margin, quarter by quarter. If transactions stay above 4% and margin crosses 13% by fiscal 2028, the read holds; if either fails, you overpaid.