Coffee imports by country (HS 0901) and the 2026 tariff

If you run a coffee brand, your cost of goods walks in the door as green beans someone else grew, priced in a commodity market you do not control, then run through a tariff regime that changed four times in nine months. For 70-plus years that last part was simple: US coffee paid zero duty. That is no longer true. This post maps where US coffee actually comes from under HS 0901 (the customs code that covers coffee), what the new tariff layer does, and why the bigger threat to your margin is the bean price, not the border.

Where US coffee actually comes from (the HS 0901 map)

The US grows almost no coffee, so nearly every bean on a US shelf is imported. In 2025, those imports totaled $12.62B for consumption under HS 0901, and they are concentrated. Colombia led at $2.715B (21.5% of the total) and Brazil followed at $2.478B (19.6%). Together those two origins are about 41% of US coffee by value, and no third country comes close.

The chart has one quirk worth flagging. Switzerland ranks third at $1.286B (10.2%), and Switzerland does not grow coffee. Those dollars are re-exported and processed product: capsules, roasted blends, and soluble coffee that moved through Swiss facilities. It sits in the higher-value roasted bucket, which is how a country with zero coffee farms outranks Honduras, Mexico, Peru, and Guatemala on import dollars. If you are mapping true growing origins, read past Switzerland to the long tail of Latin American and East African suppliers: Honduras ($748M), Mexico ($668M), Peru ($654M), Guatemala ($632M), and the specialty East African names buyers care about, Ethiopia ($312M), Uganda ($133M), and Kenya ($84M).

For an operator, the concentration is the headline. When I talk to founders sourcing single-origin or small-blend coffee, the risk they underprice is not tariff exposure, it is having one or two growing regions carry most of their supply. A frost in Brazil or a logistics snarl out of Colombia moves the price of nearly half the market at once. That is a supply-chain concentration question, and it sits upstream of any duty.

The tariff that didn't exist until it did

Coffee's tariff story was boring on purpose. Green and roasted coffee under HS 0901 entered MFN duty-free because the US grows almost none of it and roasters lobbied for decades to keep the input cheap. Then it got swept into the 2025-26 trade fight, and the rate moved repeatedly.

Where it lands today: coffee was hit with a flat 10% Section 122 across-the-board surcharge in February 2026, scheduled to expire July 24, 2026. But that surcharge has since been struck down. The US Court of International Trade declared the Section 122 tariffs invalid (Slip Op. 26-53) and issued a permanent injunction against collection that has not been stayed; the government has appealed to the Federal Circuit, but as of mid-June 2026 CBP is not assessing the 10% at entry. So the base MFN duty is still 0%, and the surcharge that briefly sat on top of it is currently enjoined, not being collected. When it was in force it was genuinely flat across origins. Even Brazil, the country most exposed to US trade friction, had coffee explicitly carved out of the June 2026 USTR Section 301 determination. So there was never a country you could switch to that paid less duty than another. The duty-arbitrage play that operators reach for first does not exist here.

The one thing to hold loosely is the status itself. The surcharge is enjoined today, but the appeal is live, and tariff policy has moved fast and unpredictably since 2025, so treat both the 10% and the 0% as possibilities rather than settled facts. Re-verify the current rate before you sign a green contract or reprice a subscription.

It's the bean, not the border

Here is the reframe that matters most. Your import value did not nearly double because of a 10% tariff that was only briefly in force in 2026 and is now enjoined. It doubled because the bean did.

Arabica spot roughly doubled, from about $2.10/lb in December 2023 to near $4.10/lb by February 2025 (it touched that level again later in 2025), easing to about $3.17/lb by May 2026. US coffee import value tracked that line almost exactly: $8.0B in 2023, $8.82B in 2024, then $12.62B in 2025, on roughly flat physical volume (inferred from value, not a kilogram series; see methodology). The rise is price, not quantity. And the pass-through to the shelf is well underway. Retail ground roast coffee hit a record $9.72/lb in April 2026, up 29% year-over-year from $7.54 a year earlier.

Put the two forces side by side and the 10% tariff is the smaller line. The pattern we see again and again with founders modeling this is that they spend a disproportionate amount of attention on the tariff headline, because it is new and it is in the news, and far too little on the green-price hedge, which is the variable actually moving their COGS. When the commodity input can swing 30-70% in a year, a 10% duty that is currently struck down and not even being collected is not where your risk lives.

What 82% green means for your COGS

Drill into the HS 0901 subheadings and one number reshapes how you should think about the tariff: 81.7% of US coffee imports are green (unroasted) beans, only 18.3% roasted.

Green not-decaf (090111) alone was $9.685B; roasted not-decaf (090121) was $2.178B; the decaf subheadings are small ($619M green, $135M roasted). For a roaster, this is the whole game. You import raw beans, so any tariff lands on your raw input before roast loss makes it worse. Coffee loses 14-18% of its weight roasting, so 1lb of green yields only about 0.85lb of roasted product. Every cost on the green side, the bean price, freight, and any duty that is in force, gets divided across less sellable weight after the roast.

Walk a single 12oz bag. Call it roughly 14oz of green at $3.50/lb specialty, so about $3.06 of bean. A 10% Section 122 surcharge, when it was being collected, added about $0.31; right now, with the surcharge enjoined, it adds nothing. Add freight and the roast-loss uplift and your landed, roasted cost per bag is dominated by the bean, with the tariff a real-but-minor line even at full force. The takeaway for repricing: if you are going to move a price to defend margin, size the move against the bean, not the duty. The duty is a rounding error next to a $1.50/lb swing in arabica.

What this means for a DTC coffee brand's pricing and margin

For a DTC coffee brand, the data points to three moves, and none of them is origin-shopping for duty arbitrage.

First, protect the fully-loaded gross margin. The coffee brands we model at this size target roughly 52% fully-loaded gross margin (inbound freight, duties, pick-and-pack included), with 60%+ marking a genuinely strong operator. Anything under 40% makes paid acquisition structurally hard to scale. With green up sharply, a brand that has not repriced since 2024 is almost certainly sitting below its own target right now and may not have noticed, because the erosion came through COGS, not through a line item anyone re-reads.

Second, fix your repricing cadence. When the input swings this hard, an annual price review is too slow. Operators who held a list price for 18 months while arabica doubled effectively absorbed the entire commodity move themselves. A quarterly review against a green-cost index, with a pre-agreed rule for when you pass through versus when you eat it, is the difference between defending margin and discovering the loss at year-end.

Third, lean on subscription, because it is where coffee economics actually work. Subscription customers deliver 3-5x the LTV of one-time buyers, and for a replenishable product like coffee the whole P&L hinges on subscription conversion and retention. When we have struggled to make coffee unit economics close, the fix was rarely a new ad channel; it was conversion to subscription on the first order (aim for 35%+, push toward 50%) and hard dunning to recover failed payments, since a large share of early churn is broken cards, not unhappy customers. A price increase lands far softer on a subscriber who is getting brew guidance and can tweak their roast than on a one-time buyer comparing shelf prices.

Operators keep asking which origin to switch to in order to dodge the tariff. That is the wrong question. Every origin faced the same 10%, and that surcharge is now struck down and not even being collected. The bean price doubled and has not lapsed at all. Hedge the green, tighten your repricing cadence, and convert buyers to subscription. The border is the small lever; the bean is the big one.

For more on the unit economics behind these bands, see our coffee financial benchmark and our interim CFO services overview. For the same trade-and-tariff lens applied to another category, see our import-origins work on beauty.

Sources and methodology

US import figures come from the US Census Bureau International Trade imports API, HS code 0901 and the 6-digit subheadings 090111, 090112, 090121, and 090122. We pulled variable CON_VAL_YR (cumulative imports-for-consumption value, in USD) at month 12 of each year for the full-year figure. Years pulled: 2021 ($6.75B), 2022 ($9.47B), 2023 ($8.00B), 2024 ($8.82B), and 2025 ($12.62B). The 2025 country breakdown is the month-12 pull across all partner countries; shares are each country's value divided by the all-countries total.

Arabica prices come from FRED series PCOFFOTMUSDM (Global price of Coffee, Other Mild Arabicas, US cents per pound, monthly, not seasonally adjusted). Key points used: December 2023 = 210.3, February 2025 peak = 409.5, May 2026 = 317.5. The annual averages in the import-value chart are approximations computed from the monthly series.

Retail coffee prices come from FRED series APU0000717311 (average price, coffee, 100% ground roast, all sizes, per pound, US city average, BLS). April 2024 = $6.063, April 2025 = $7.536, April 2026 = $9.723 (record), May 2026 = $9.511. The year-over-year April 2025 to April 2026 change is +29.0%.

Tariff status was confirmed via Parallel.ai deep research with primary citations: the White House Annex I relief (BHFS, November 2025); the Supreme Court opinion in Learning Resources, Inc. v. Trump (24-1287, February 20, 2026); a Skadden analysis of the Section 122 Proclamation 11012 (effective February 24, 2026, scheduled to expire July 24, 2026); and the USTR Section 301 Brazil determination (June 1, 2026, with coffee exempt). Perplexity regulatory research corroborated the 0% MFN base on HS 0901 and surfaced the subsequent US Court of International Trade ruling (Slip Op. 26-53) striking down the Section 122 tariffs: the court issued a permanent injunction against collection that has not been stayed, and the government has appealed to the Federal Circuit, so as of mid-June 2026 CBP is not collecting the 10% surcharge.

Two limitations matter. First, this analysis uses import value, not physical quantity; the "flat volume" claim is inferred from value rising while widely reported volumes held roughly steady, not from a kilogram series pulled here. Second, the tariff environment is moving and the specific surcharge this post analyzes is in active litigation. The 10% Section 122 surcharge was imposed February 24, 2026 and scheduled to lapse July 24, 2026, but the US Court of International Trade has since struck it down (Slip Op. 26-53) and enjoined collection; that injunction has not been stayed and the government's appeal to the Federal Circuit is pending, so CBP is not currently collecting the surcharge. Because the status could flip again on appeal or by further policy action, any reader should re-verify the current rate before acting rather than rely on either the "10%" or "0%" figure as settled. Operator benchmark bands are drawn from the Eightx coffee financial benchmark and Foundry DTC food-and-beverage benchmarks for 2026 and are planning targets, not guarantees.

Frequently asked questions