Kazakhstan now charges 16 percent VAT on foreign online sellers from their first local order, and Uzbekistan opened a 3 to 5 percent bonded warehouse channel in July. The EAEU parcel overhaul that many guides call live was postponed to January 2027.
The rule most logistics guides describe as live since July does not start until January 2027. The rule that is live has no turnover threshold at all, which means it applies from your first order.
Four rules changed the math of selling into Central Asia this year, and one of them is widely misreported. Kazakhstan began charging foreign online sellers 16 percent VAT from their first local order on January 1. Uzbekistan opened a bonded warehouse import channel on July 1 that taxes electronics at 5 percent and clothing at 3. The Eurasian Economic Union’s new parcel regime, which plenty of logistics blogs still describe as live since July, was postponed and now starts on January 1, 2027. On that same date, Uzbekistan cuts its over limit parcel charge from 30 percent to 20.
Two of those four changes make the region more expensive. Two make it cheaper. None of them is the first question a Shopify merchant should be asking about Central Asia, which is where most coverage of this region goes wrong. The tax rules only start to matter once somebody can read your store and wants what is on it.
Roughly four in five people in Uzbekistan do not speak Russian, which means the single localization decision most Western sellers make about Central Asia is wrong before any customs rule applies. The instinct is to file the whole region under “Russian speaking” and localize once. That instinct works in Kazakhstan. It fails badly next door.
Kursiv reported in May 2025, from a presentation at a tech forum hosted by IT Park Uzbekistan, that about 7 million residents of Uzbekistan speak Russian, or 18.5 percent of the population. English came second at 5 million speakers, Turkish third at 1 million. A Russian only storefront in that market is unreadable to four shoppers in five, and an English only storefront reaches fewer people still.
The best evidence on what that costs is global rather than Central Asian. CSA Research surveyed 8,709 consumers in 29 countries in 2020 and found that 76 percent preferred to buy products with information in their own language, and 40 percent said they never buy from sites in another language. That survey did not cover Uzbekistan, so it establishes the direction of the effect rather than its size in Tashkent. The storefront order is Uzbek first, in Latin script, and Russian second, which is a different brief than most translation vendors will quote you by default.
Russian still runs the back office. Wildberries and Ozon, two of the marketplaces the Times of Central Asia names among Kazakhstan’s largest, are Russian companies, and a parcel program spanning both countries will have partners who work in Russian. That shapes hiring rather than merchandising. A support or operations hire in Tashkent will usually be native in Uzbek, and Russian is the skill to test for instead of assuming. Plenty of people there are paying to close that gap themselves. Prep.uz sells a three-month Russian course to Uzbek speakers who need the language for work, built on two one to one mentor sessions and two group conversation classes a week, plus short video lessons. Know its limits before you point anyone at it. By the school’s own framing the target is conversational Russian, not contract or customs Russian, and the site is in Uzbek and Russian only, so it is something your hire enrols in rather than something you administer from abroad.
Kazakhstan is a mature online market that has just been taxed, and Uzbekistan is an early one being opened on purpose, which means the same expansion plan cannot serve both. Treating them as one region is the second mistake after the language one.
Kazakhstan’s Bureau of National Statistics counted 3.77 trillion tenge of online retail in 2025, about $8.2 billion at the Times of Central Asia’s conversion, or 14.3 percent of all retail. That is roughly 50 percent more than two years earlier. Uzbekistan has about 37.2 million people with a median age of 27, according to DataReportal’s Digital 2026 report, yet its ecommerce volume is around $1.3 billion. The president’s office put that at 4 to 4.6 percent of retail in an April briefing, against a global average of 22.
One number belongs in the Kazakhstan decision more than any other. Marketplaces took 86 percent of Kazakhstan’s online retail in 2025, up from 84.9 percent the year before, and the share going through retailers’ own websites slipped from 15.1 percent to 14. A standalone storefront is competing for the small slice, and the pattern I have watched play out repeatedly at the $500K to $2M stage is a founder opening a direct storefront in a marketplace dominated market because the direct channel is what they know. The market is real. The channel is wrong. If the demand signal in your analytics is Kazakh, the first test is usually a marketplace listing, not a localized store, and the one market at a time approach to international selling applies with more force here than in Western Europe.
Uzbekistan is the reverse. There is very little marketplace gravity to fight, the buyer base is young, and the government is actively lowering the cost of getting goods in. The trade off is that almost nothing is built yet, which shows up as a warehousing constraint later in this piece.
A foreign company selling online to individuals in Kazakhstan must register with the State Revenue Committee within one month of its first payment from a Kazakh customer, with no turnover threshold at all. The rule is short and the absence of a floor is the part that catches small merchants.
Under the new Tax Code, that registration happens through a register kept by the committee. KPMG’s summary of the rules lists no turnover threshold. The rate is 16 percent, up from 12, and changes to your company details must be reported within 10 business days. The penalty is not a fine. Edil Azimshayyk, who runs VAT administration at the committee, said last October, in remarks carried by the Times of Central Asia, that for companies ignoring a registration notice, “access to their online platforms will be suspended.”
It would be easy to file this under rules that exist only on paper. The receipts say otherwise. Kursiv reported in August that 133 foreign companies are now registered as VAT payers in Kazakhstan, and the money they send has grown every year since the first version of the tax arrived in 2022.
Look at who pays. The largest contributor in 2025 was not a streaming service or an app store. It was Elementary Innovation, the Temu affiliate, at $28.9 million. iHerb, which ships vitamins, paid $10.6 million, roughly fourteen times what Netflix did. Physical goods crossing the border in small parcels are a serious part of this tax base, and the top five payers covered more than 80 percent of it.
For a Shopify store shipping a few dozen orders a year to Almaty, the honest reading is uncomfortable. The rule as written has no floor, so you are in scope from order one. I found no report of a small independent store being blocked, and the enforcement energy has clearly gone to platforms. That is an observation about the past nine months, not a legal position. Whether to register, sell through a marketplace that already has, or switch Kazakhstan off at checkout is a question for a tax adviser, and it is worth asking before the orders arrive rather than after.
The new EAEU customs regime for cross-border ecommerce starts on January 1, 2027, not July 1, 2026, and any guide telling you it is live today was written before August. This is the single most common factual error in current coverage of the region.
Here is what was supposed to happen. The Eurasian Economic Commission moved goods that individuals buy from foreign online stores into their own customs category, “e-commerce goods”. The 200 euro duty free threshold stays. Above it, the duty becomes 5 percent of the whole order value with a floor of 1 euro per kilogram, and the declaration is filed by an ecommerce operator instead of the buyer.
On July 1, the day it was due, Kazakhstan’s State Revenue Committee announced that the regime could not take effect because one member state had not finished ratifying the amendments to the customs code. On August 7 the commission published the new date. The commission’s August notice says the rules start on January 1, 2027, that Armenia, Belarus, Kazakhstan and Kyrgyzstan have ratified, and that Russia’s final step is still pending. Kazakhstan Today reported the same week that EAEU heads of government have instructed national authorities to complete preparations for the 2027 start.
Until then the old formula applies: a flat 15 percent on the portion above 200 euros. Run a 500 euro order through both. Today it pays 45 euros. Under the new regime the duty alone falls to 25 euros, but the committee has said Kazakhstan’s 16 percent VAT is charged as well. Add that and the same order costs more than 100 euros to clear, over twice today’s bill. According to Tengrinews, the committee names Temu, AliExpress, iHerb and Amazon as the kind of seller affected, and says orders from Wildberries, Ozon and Kaspi are not, because those goods are already inside the union. If you price duty inclusive, and for cross-border you should, then your landed cost model needs a January 2027 version built now rather than in December.
Uzbekistan’s duty free limit is $200 per person per calendar month across every foreign seller that person buys from, which means your parcel can be taxed because of an order someone else shipped. Uzbekistan is not an EAEU member, so nothing in the section above applies there.
The rule has been in place since May 1, 2025: each person may receive $200 of goods a month by international courier without paying anything, down from $1,000 a quarter. For ordinary post the limit is $100. Go over and a single customs payment applies, standing in for duty, VAT and excise. In Asia-Plus’s worked example it is charged on the excess, not the whole value. The rate is 30 percent today, with a minimum of $3 per kilogram. A decree signed on August 27 lowers that to 20 percent and $2 per kilogram from January 1, 2027, as UzDaily and Asia-Plus both reported this month.
The word that matters is “month”. The allowance is cumulative across every foreign seller the customer buys from, so a $120 order from you on the 3rd and a $120 order from someone else on the 20th puts the second parcel over. Buyers know where they stand. The state services portal shows each citizen the allowance they have left, and since May it lets them review every incoming parcel’s description, declared value and weight, then approve or reject it online.
Three product profiles get hurt by this and they are worth naming, because merchants rarely see it coming: bundles, subscription boxes, and anything priced close to $200. If your average order value in Uzbekistan sits between $150 and $220, you are running straight at the threshold, and a customer who has already bought from someone else that month will see a bill you did not quote. Splitting a bundle across two calendar months is a crude fix that works, and it is worth testing before you conclude the market does not convert.
Uzbekistan’s bonded warehouse pilot lets imported goods clear at 5 percent for electronics and 3 percent for clothing, charged only when an item actually sells, which is a fraction of the standard import burden. Presidential resolution PP-136, signed April 13, created a two year pilot that runs from July 1, 2026 to July 1, 2028.
Goods go into a customs controlled bonded warehouse inside Uzbekistan with nothing paid at the border. They are listed on a registered local ecommerce platform, and the customs charge falls due only when an item sells to a consumer.
Kun.uz put the standard import burden at more than 17 percent for electronics and up to about 30 percent for clothing, which is the gap the pilot is designed to close. One caution on the 3 percent. UzDaily and a legal alert from the consultancy BeOne describe it as VAT and duty combined, while Kun.uz’s summary reads it as 3 percent duty plus 12 percent VAT. Have a customs broker read the resolution itself before you build a price list on either version.
This is not a door a small brand walks through alone. The government’s July rules require a warehouse operator to hold a free warehouse licence, run inventory software wired into the customs system, and control at least 15,000 square meters. The selling platform has to be an Uzbek legal entity on a state register, with online cash registers and nationwide delivery. Your way in is a contract with one of those platforms, which makes this a $2M and above move rather than a first international market. Read the fine print with them: customs charges are not refunded when a consumer returns an item, stock that has not sold after two years must be re-exported within a month, and goods can be sold only to individuals for personal use.
Capacity is the other constraint. By the president’s office’s own count the country has 634,000 square meters of warehouse space, 72 percent of it in and around Tashkent and only 34 percent of it Class A, against an estimated need of 2.5 million within five years. The channel is cheap because the infrastructure behind it is thin.
The right move in Central Asia depends almost entirely on your revenue stage, and applying a $10M brand’s playbook to a $500K store is how merchants lose money in a market that was never going to be their problem. The fundamentals are the same everywhere. The sequencing is not.
Under $500K, do nothing structural. Fulfil the occasional organic order at a transparent price, keep Kazakhstan’s no threshold VAT rule on your radar, and put your energy into the domestic fundamentals that are still unfinished. Between $500K and $2M with a real demand signal from Kazakhstan, the first test is a marketplace listing rather than a localized storefront, because 86 percent of the online retail in that country runs through marketplaces and a direct store is fighting for the remaining 14. Between $2M and $10M, the Uzbek market starts to justify the localization work, which means Uzbek first in Latin script and Russian second, and the bonded warehouse channel becomes a conversation worth having with a registered local platform. Above $10M, both markets support a deliberate program, and the January 2027 EAEU change is a landed cost project to schedule now.
Across every stage, two things hold. Check whether your customer can read your store before you check what your parcel will be taxed, because the localization mistake costs more than the customs mistake and is cheaper to fix. And treat the payment side as something you measure rather than something you read about. Uzbekistan’s domestic card schemes and its local payment services do not always clear on a store hosted abroad, and no published figure tells you what share of cards will. The only measurement available is your own: a small capped campaign, and the payment decline rate for that country read against every other market you ship to. If declines run high, you have found your constraint, and it is not the tax code.
The rule as written has no turnover threshold, so a foreign seller is in scope from the first payment received from a Kazakh customer and must register with the State Revenue Committee within one month. The rate is 16 percent. In practice, enforcement so far has concentrated on large platforms, and there is no public report of a small independent store having its site blocked. That is an observation about how the first nine months of 2026 played out, not legal advice or a safe harbour. The realistic options are to register, to sell through a marketplace that is already registered, or to switch Kazakhstan off at checkout. Which one fits depends on how much revenue the market represents, and that is a conversation for a tax adviser before the orders start arriving.
The new EAEU regime starts on January 1, 2027, after being postponed from its original July 1, 2026 date. The delay happened because one member state had not completed ratification of the amendments to the customs code. Armenia, Belarus, Kazakhstan and Kyrgyzstan have ratified; Russia’s final step was still pending as of the commission’s August announcement. Until the new rules take effect, the old formula applies, which is a flat 15 percent on the portion of an order above 200 euros. From 2027 the duty becomes 5 percent of the full order value with a floor of 1 euro per kilogram, the 200 euro duty free threshold stays, and an ecommerce operator files the declaration instead of the buyer. Any guide describing the regime as already live was written before August 2026.
Uzbekistan’s $200 allowance is counted per person per calendar month across all foreign sellers, not per parcel or per merchant. That means a customer who already received $120 of goods from another seller earlier in the month will pay customs on your $120 order, even though neither order exceeds the limit on its own. For ordinary post the limit is $100 rather than $200. Above the threshold, a single customs payment replaces duty, VAT and excise, charged on the excess rather than the full value, at 30 percent with a minimum of $3 per kilogram. From January 1, 2027 that falls to 20 percent and $2 per kilogram. Bundles, subscription boxes, and any product priced near $200 are where this bites hardest.
Lead with Uzbek in Latin script and treat Russian as the secondary language, because only about 18.5 percent of Uzbekistan’s population speaks Russian. That figure, roughly 7 million people, was reported by Kursiv in May 2025 from a presentation at IT Park Uzbekistan, with English second at 5 million speakers. A Russian only storefront is therefore unreadable to roughly four shoppers in five. This is the opposite of the right answer for Kazakhstan, where Russian is far more widely spoken, which is why treating Central Asia as a single localization project produces a store that underperforms in one of the two markets. Russian remains useful on the operations side, since regional marketplace and logistics partners often work in it, but that is a hiring consideration rather than a storefront one.
Not directly, because the entry requirements are built for infrastructure operators rather than individual merchants. A bonded warehouse operator must hold a free warehouse licence, run inventory software integrated with the customs system, and control at least 15,000 square meters of space. The selling platform must be an Uzbek legal entity listed on a state register, with online cash registers and nationwide delivery. A foreign brand’s route in is a commercial contract with one of those registered platforms, which realistically makes this a move for merchants above roughly $2M rather than a first international market. The economics are genuinely attractive, at 5 percent for electronics and 3 percent for clothing against a standard burden of 17 to 30 percent, but the terms include no refund of customs charges on consumer returns and a requirement to re-export unsold stock after two years.
