Starbucks Didn't Win the Union Fight. It Outlasted It.

On December 9, 2021, baristas at a single store on Elmwood Avenue in Buffalo counted the ballots and got an answer: 19 to 8 in favor. The first company-owned Starbucks in the United States had a union.¹ It felt like a domino falling. Within a few years, over 12,000 workers at roughly 700 stores in at least 45 states would vote the same way.⁶ By any normal reading of a labor campaign, that is a rout. And yet, as of late 2025, not one of those stores has a signed contract.⁶ The movement won almost every vote it called — and is still waiting for the one thing a union exists to deliver.

The story everyone tells is a morality play: scrappy workers organize, a giant corporation fights dirty, and either justice or capital prevails. Both versions miss the actual mechanism. Starbucks did not crush the union in a single decisive blow. It did something quieter and far more durable — it turned the fight into a process, and then made the process slow.

Even the first victory wasn't a clean sweep

Start with the founding myth, because the strategy hides inside it. The Buffalo vote is remembered as a landslide. It wasn't. Elmwood Avenue passed 19-8 — but on the very same day, a second Buffalo store on Camp Road voted 12-8 against unionizing.¹ Nor was this the first Starbucks union on earth: a store in Victoria, British Columbia had joined the United Steelworkers in August 2020 and signed a three-year agreement in 2021, well before Buffalo.⁸ The accurate claim is narrow and specific — first company-owned U.S. store — and that precision matters, because the whole Starbucks playbook is built on precision. The company would spend the next four years winning on technicalities the public never read closely enough to notice.

The largest violation docket — and almost no consequence

Here is the part that should be impossible to reconcile. In March 2023, an NLRB administrative law judge found Starbucks had violated federal labor law in Buffalo through what he called 'egregious and widespread misconduct demonstrating a general disregard for the employees' fundamental rights' — ordering fired workers reinstated, a shuttered store reopened, and Howard Schultz himself present for a reading of worker rights notices.² That same month, the Senate's labor committee tallied the broader record: judges had found Starbucks broke the law 130 times across six states, including firing or forcing out a dozen pro-union workers, with 70 more cases headed to trial.³ A company can amass one of the heaviest labor-violation dockets in modern memory and still control the outcome — because findings are not the same as remedies, and remedies take years.

This is the mechanism, worked down. A union drive runs on momentum: a fired organizer reinstated this month energizes the next store; a fired organizer reinstated in three years, after appeals, is a footnote to people who have long since quit and moved on. The legal labels — 'allegations,' 'charges,' 'findings,' 'remedies' — are not interchangeable, and the gaps between them are where time accumulates. Over 500 charges sounds like 500 crimes. It is not. Charges are allegations; the 130 are judge-level findings; and even those are subject to Board and appellate review, with some later vacated. Each rung down the ladder is real, and each rung takes longer than the one above it. Starbucks did not need to be innocent. It only needed the clock.

The Supreme Court didn't clear Starbucks — it slowed the referee

The clock got a powerful new ally in June 2024. In Starbucks Corp. v. McKinney, the Supreme Court ruled 8-1 that when the NLRB asks a court for a quick injunction to reinstate fired workers, the court must apply the stricter four-factor Winter test rather than a more lenient standard.⁴ The case is widely misread as a vindication. It was nothing of the kind. Justice Jackson dissented in part, the Court vacated and remanded, and — crucially — it ruled on none of Starbucks' actual conduct.⁴ The underlying dispute even cut against the tidy narrative: the fired Memphis workers had admitted to letting a TV crew into a closed store after hours, and Starbucks' policy-violation defense was real, even as the labor board alleged the true motive was anti-union animus. The Court answered only one question — how fast the referee may act — and it answered it in the company's favor. Emergency reinstatement, the union's single fastest tool, got slower for every employer in America.

Schultz's appearance that March is its own small case study in framing. He is often said to have defied a subpoena; in fact he agreed to testify voluntarily before the committee even voted on one, appearing under its threat but not its execution.⁵ On the stand he was emphatic — the company 'has not broken the law' — recasting judicial findings as mere accusations, and he declined Bernie Sanders' demand to hand the union contract proposals within 14 days.⁵ That refusal is the whole strategy compressed into a single sentence. There is no law that says you must bargain quickly. There is only a law that says you must bargain in good faith — and good faith is, conveniently, slow to prove.

The honest counter: maybe the law is just slow

The fair objection is that this reads too neatly — that no one at Starbucks drew up a master plan called 'exhaust them with the calendar,' and that American labor law is simply, structurally glacial for everyone. That's true, and it's the strongest version of the company's defense. First contracts routinely take years even where both sides act in good faith; the policy-violation defense in Memphis was genuine; and Starbucks can point to rulings where the Board upheld findings that it acted lawfully, as it did after two June 2026 NLRB opinions covering Seattle and Portland stores.⁷ But notice what the steelman concedes. If the law is slow by default, then the rational move for a well-resourced employer is to do nothing to speed it up — and that is exactly what a bad-faith bargaining complaint, filed by the union in December 2024 after talks stalled, alleges happened.⁶ You don't need a conspiracy to exhaust a movement. You need only the patience to let a slow system stay slow, and the lawyers to keep it that way.

So where does that leave a movement that won almost everywhere it asked? Seven hundred stores wear the badge of a victory that hasn't paid out.⁶ The NLRB keeps finding violations — two more as recently as June 2026, for unlawfully interrogating workers about strike plans — and Starbucks keeps replying that the same board upheld many findings in its favor.⁷ Both statements are true, which is the point. The genius of the response was never a knockout. It was the discovery that in a system built on process, the side that can afford to wait usually wins — and that a $100B company can wait longer than any group of baristas can stay angry. The union got the votes. Starbucks got the years. And in a contract fight, only one of those gets ratified.