India's Diamond Exports Have Halved. Its Jewellery Exports Just Overtook Them.
India cuts 9 of every 10 diamonds on earth. A map of every export destination, and the one test that shows who really sets the price.
- Yes, India dominates the diamond trade. It cuts roughly 9 of every 10 diamonds in the world, and exported US$ 27.72 billion of gems and jewellery in the year to March 2026. But "gemstones" is mostly diamonds: actual coloured stones are US$ 437 million, about 1.6% of the basket.
- The diamond business is roughly half the size it was in 2018. US$ 25.6 billion of diamond exports in calendar 2018 became US$ 12.9 billion in 2025. Rough diamond imports, the leading indicator, fell another 26% in April to July 2026.
- 2025 was a structural crossover. Finished jewellery exports (US$ 13.70 billion) passed diamonds (US$ 12.86 billion) for the first time in at least fourteen years.
- Run the price-taker test, not the growth test. India shipped more polished carats in April to July 2026 and collected less money for them. More units plus less money equals no pricing power, and that is the whole diagnosis.
- Lab-grown is a volume story and a value disaster. Carats up 30.5%, value down 10.6%, realisation down 31% in one year to US$ 60 per carat, about 8% of what a natural carat fetches. Buildable capacity is the opposite of a moat.
India cuts and polishes roughly nine out of every ten diamonds on earth, and almost all of that work happens in one city. Surat is to diamonds what Shenzhen is to electronics.
It is also a business in retreat. India's diamond exports have roughly halved since 2018, and in 2025, for the first time in at least fourteen years, the country sold more finished jewellery than diamonds.
That crossover is not a rounding error. A different industry is emerging inside the shell of the old one, and it changes what an investor should be looking for.
The numbers below come from the Gem and Jewellery Export Promotion Council, which reports monthly to India's Ministry of Commerce, and from UN Comtrade. Both publish considerably more than the headlines ever use.
Where It All Goes, on One Map
Start with geography, because the geography is the story. Every bubble below is one destination, sized by the value India shipped there in calendar 2025 and coloured by whether that flow grew or shrank against 2024.
Switch tabs to see how completely the picture changes depending on what you are looking at. The destinations for polished stones and the destinations for finished jewellery are almost different maps.
Calendar year 2025, US dollars, free on board. Bubble area is the value shipped; colour is the change against 2024.
HS 7102 + 7103 + 7104 + 7113. Diamonds, coloured stones, lab-grown stones and finished jewellery combined. Total $28.17bn across 42 destinations.
A further $113mn went to small partners and to Comtrade's unallocated aggregates, which have no location to plot. The 42 destinations above are 99.6 per cent of the reported total.
Three cities dominate. The UAE, the United States and Hong Kong together take about three quarters of everything India ships. For all its size, this is a business with a handful of very large customers, which makes it a concentration risk before it is anything else.
The United States is deep red. It is the only large market that fell hard, and it fell by nearly a third in a single calendar year.
Antwerp still matters. Belgium is a small country taking US$ 1.5 billion of Indian stones, essentially all of it diamonds. That is the four-hundred-year-old diamond trading district doing what it has always done, which is act as the clearing house between miners and cutters.
One caution before reading too much into any of it. A trade map is not a demand map. A bubble over Dubai or Hong Kong tells you where goods were consigned, not where anyone eventually bought a ring. Both are trading hubs whose business is to receive goods and send them on. That distinction matters more later.
The Business Is Enormous, and India Mines None of It
The headline numbers first.
Source: GJEPC master report for March 2026, the year-end statement covering April 2025 to March 2026.
Two things get lost in that total.
First, "gemstones" is mostly a figure of speech here. When people say India exports gemstones, they overwhelmingly mean diamonds. Actual coloured stones, meaning rubies, emeralds, sapphires and the rest, are US$ 437 million out of a US$ 27.7 billion basket. Jaipur's coloured stone trade is culturally important and commercially minor. It has been roughly flat for a decade.
Second, India does not mine any of this. India imports rough diamonds, cuts and polishes them, and exports the result. In FY2025-26 it imported US$ 10.48 billion of rough diamondsRough DiamondAn uncut diamond as it comes out of the mine. India imports rough, cuts and polishes it, and exports the polished stone. The gap between the two prices is the industry's entire income.See all terms in the glossary and exported US$ 12.16 billion of polished onesCut & Polished Diamond (CPD)A rough stone after cutting and polishing. India does this to roughly 9 out of every 10 diamonds in the world, mostly in Surat.See all terms in the glossary. The entire industry lives in the gap between those two numbers, and that gap is what you are analysing when you analyse this sector.
India owns no mines of consequence. Its entire position rests on a workforce in and around Surat, commonly put at close to a million people, who can cut a stone more cheaply and more precisely than anyone else. Whether that stays scarce is the question for the next decade.
The Year Jewellery Quietly Passed Diamonds
India's exports to the world, calendar years, US dollars, free on board.
In calendar 2018 India exported US$ 25.6 billion of diamonds and US$ 12.4 billion of jewellery. Diamonds were twice the size of jewellery.
In calendar 2025 India exported US$ 12.9 billion of diamonds and US$ 13.7 billion of jewellery. Jewellery is now the larger business. It has not been in any year since 2012.
The diamond line did not fall because of one bad year. It fell in 2023, again in 2024, and again in 2025. Cut and polished diamond exports fell another 7.99% in the first four months of FY2026-27, to US$ 3.60 billion, and July 2026 alone was down 18.23% (GJEPC, July 2026).
Meanwhile, rough diamond imports, the raw material, fell 26.46% in April to July 2026 versus the same period a year earlier. Rough imports are the leading indicator here. Surat buys rough today to sell polished in a few months, so a sharp fall in rough purchases tells you what the polished numbers will look like next.
None of which means India lost the diamond business and won the jewellery one. What happened is narrower, and worth stating precisely: the diamond business shrank a great deal while the jewellery business did not. Those are different claims, and the difference is where the rest of this article lives.
Cutting More Stones, Collecting Less Money
The framework here is not specific to diamonds.
Most industries never let you run this test, because they only report revenue. The gems trade reports carats and dollars side by side every month, which makes it one of the few places you can.
Natural diamonds, more carats for fewer dollars
FY2025-26: 160.04 lakh carats of polished stones, down 3.85%
April to July 2026: 53.98 lakh carats, up 7.66%
Surat is cutting a similar number of stones, and lately more of them.
FY2025-26: US$ 12.16 billion, down 8.52%
April to July 2026: US$ 3.60 billion, down 7.99%
Average realisation fell from US$ 780 to US$ 666 per carat, a drop of about 15%.
FY2025-26: 160.04 lakh carats of polished stones, down 3.85%
April to July 2026: 53.98 lakh carats, up 7.66%
Surat is cutting a similar number of stones, and lately more of them.
FY2025-26: US$ 12.16 billion, down 8.52%
April to July 2026: US$ 3.60 billion, down 7.99%
Average realisation fell from US$ 780 to US$ 666 per carat, a drop of about 15%.
More stones out, less money back. That is the signature of a price taker, and it has been getting worse rather than better.
Lab-grown diamonds, 30% more carats at 31% less each
Now the same test on the segment everyone calls the growth story.
GJEPC master reports, March 2026 and July 2026. Per-carat figures are export value divided by carats exported.
India shipped 30% more lab-grown caratsLab-Grown Diamond (LGD)A real diamond grown in a reactor rather than dug out of the ground. Chemically identical to a mined stone, but capacity can be built with capital, so prices fall toward the cost of production.See all terms in the glossary and collected 10% less money for them. Average realisation fell 31% in a single year, from about US$ 88 to US$ 60 per carat. In April to July 2026 it fell again, to US$ 68 per carat against US$ 75 a year earlier.
To put that in perspective: in FY2025-26 a polished lab-grown carat fetched US$ 60 against US$ 760 for a natural polished carat, so roughly 8% of the price.
This is what a business with no moat looks like from the inside. Lab-grown diamond capacity is a machine you can buy. Anyone with capital and electricity can add supply, and they have. When supply can be manufactured on demand, price falls toward the cost of production and stays there. Volume growth in a business like this is not evidence that the business is good. Often it is evidence of the opposite, because the only way to hold revenue flat is to run harder every year.
This is the same trap that catches investors in commodity chemicals, in solar modules, and in memory chips during a glut. The mechanism is identical to the one in how commodity cycles actually price, even though the product could not be more different.
The Jewellery Boom That Shipped Less Gold
The headline for FY2025-26 and for the current year is that value-added jewellery is booming. Silver jewellery exports rose 52.21%. Platinum rose 39.32%. Studded gold rose 6.27% while plain gold fell.
Then you notice that GJEPC prints the metal price alongside every one of those numbers, and quietly flags why.
Gold averaged US$ 3,938 per troy ounce in FY2025-26 against US$ 2,578 the year before, a rise of 52.78%. Silver rose 74.57%. Platinum rose 63.60% (all from the LBMA tables reproduced in GJEPC's own report).
So silver jewellery exports rose 52% while the silver inside them rose 75%. When the value of what you sell rises more slowly than the price of the raw material inside it, you are shipping less material than before. The growth sits in the metal price rather than in the business.
Dividing value growth by metal price growth gives a rough estimate of how much metal moved:
| Segment, FY2025-26 | Export value | Metal price (US$) | Implied quantity |
|---|---|---|---|
| Plain gold jewellery | -7.42% | +52.78% | about -39% |
| Studded gold jewellery | +6.27% | +52.78% | about -30% |
| Silver jewellery | +52.21% | +74.57% | about -13% |
| Platinum jewellery | +39.32% | +63.60% | about -15% |
Export values and metal prices both from GJEPC's March 2026 report. The last column is arithmetic on the first two.
Every single jewellery category shipped less metal than the year before. The same pattern repeats in April to July 2026: studded gold up 21.09% in value against gold up 34.79%, implying roughly 10% less metal.
That calculation is cruder than it looks. A finished piece is not a lump of metal. Its export value carries making charges, design, the stones set into it, and whatever brand the buyer is paying for. If India were shifting toward more elaborate, higher-margin work, value could hold up while metal content fell, and that would be a sign of health. The arithmetic cannot separate the two cases. What it does settle is that the jewellery growth in the headlines is not a volume boom.
Whenever a company or an industry reports growth in a product with a volatile commodity inside it, find the commodity's price move for the same period and divide. Steel fabricators, gold financiers, jewellery retailers, oil marketing companies and fertiliser makers all need this treatment. The technique is a cousin of the volume-versus-value checks in how to read an annual report.
America Fell 45% in a Year. Where Did the Goods Go?
The single largest change in the map is the United States.
Source: GJEPC, FY2025-26 year-end release. Comtrade shows the same collapse on a calendar-year basis, down 32.4% to US$ 6.12 billion.
The cause was tariffs. US duties on Indian goods rose sharply during 2025, and gems and jewellery, being discretionary and price-sensitive, absorbed the hit immediately. The US went from India's largest customer to its third in about twelve months.
Two things happened next, and they are easy to confuse.
Some demand really did move. On the same Comtrade calendar-year basis as the map, Saudi Arabia is up 79%, Canada up 46%, and Australia and France both up 19%. These are small in absolute terms but real, and they reflect an effort to sell into markets that were previously an afterthought.
Some of it just changed address. The UAE is up US$ 2.0 billion year on year in the Comtrade data. Hong Kong is up US$ 0.5 billion. Both are entrepôts: their business model is receiving goods and forwarding them. When a direct route becomes expensive, trade reroutes through a hub, and the hub's import statistics rise without a single extra consumer buying a single extra ring.
This is the problem with celebrating the Dubai number. The India-UAE trade agreement in force since May 2022 cut tariffs and made the route cheaper, which is a real advantage. But a jump in shipments to a re-exportRe-ExportGoods that enter a country and leave again largely unchanged. A trading hub's import numbers can look like demand when they are really just goods passing through on the way somewhere else.See all terms in the glossary hub is not proof of end demand, and neither GJEPC nor Comtrade can tell you what share was consumed locally versus forwarded. Anyone who tells you confidently is guessing. Treat "exports to the UAE grew 28%" as a fact about logistics until someone shows you evidence it is a fact about consumers.
Three Businesses Wearing One Name
Three businesses share that export number, and they are moving in different directions for different reasons.
Three businesses inside one export number
Natural diamond polishing
Hard constraintImport rough, cut it, export polished. India does about 90% of the world's cutting by piece count. Volumes are holding but realisation per carat is falling, and rough purchases have dropped sharply, which points to more of the same.
Typical wait: Realisation down about 15% year on year
Lab-grown diamonds
Hard constraintGrow stones in a reactor, polish, export. Volumes up 30% in FY2025-26, value down 11%, realisation down 31% in one year. Capacity is buildable with capital, so supply answers any price signal within months.
Typical wait: Realisation down about 31% year on year
Finished jewellery
TightDesign and manufacture the piece, not just the stone. Now the largest category by calendar-year value. Growth in dollars is real, but is substantially a gold and silver price effect rather than more physical output.
Typical wait: Value up, implied metal volumes down
The first two are commodity processing. The third is not, or at least does not have to be.
Paid for the stone, or for what you do to it
A cutting house earns a spread on someone else's stone. It does not choose the price of rough, it does not choose the price of polished, and it does not have a customer who asks for it by name. Its advantage is cost, which means its advantage lasts exactly as long as its cost advantage does. That is a real advantage and it has lasted forty years, but it is the kind of advantage that erodes rather than compounds. This should feel familiar to anyone who has followed the argument about Indian IT services and labour arbitrage: the same structure, the same slow squeeze, a different input.
A jewellery manufacturer that owns a design, a brand, a retail relationship or a distribution network is playing a different game. Titan does not compete on the price of gold. It competes on trust, store footprint and design, and it earns a making charge that the customer accepts without shopping it around. That is a genuine moatEconomic MoatA sustainable competitive advantage that protects a company's profits from rivals — like a castle's moat. Types include brand loyalty, network effects, switching costs, scale advantages, and regulatory barriers. Warren Buffett's core investing concept.See all terms in the glossary, and the framework for recognising one is set out in understanding economic moats.
The question to ask of any company in this sector is short: does it get paid for the stone, or for what it does to the stone? If the answer is the stone, its economics belong to the commodity and you should value it as a cyclical. If the answer is what it does to the stone, you may be looking at a franchise. Most of the US$ 27.7 billion is the first kind.
The Bear Case Is Winning. Here Is the Bull Case Anyway
In this sector the bear case is doing most of the talking right now.
Lab-grown eats the low end permanently. At US$ 60 a carat against US$ 760 for natural, the substitution at entry price points is close to done.
Producer countries want the work. BeneficiationBeneficiationA producing country's policy of requiring that stones mined on its soil are also cut and polished there, rather than shipped abroad as rough. It moves work away from India.See all terms in the glossary policies in Botswana and elsewhere push cutting toward the mine rather than toward Surat.
Tariffs may not reverse. The US decline is policy-driven, and policy can stay hostile for years.
Realisation keeps falling. Per-carat prices dropped in FY2025-26 and dropped faster in April to July 2026. Nothing in the data suggests a floor yet.
The jewellery growth is thinner than it looks. Implied metal volumes fell in every category.
The pivot is real, even if slow. Studded gold, silver and platinum are all gaining share of the basket, and those are higher value-add than polishing.
Inventory corrections do end. Much of the 2023 to 2025 decline was the trade working off stock built in the 2021-22 boom. That process finishes eventually.
Diversification is happening. Australia, Canada, France and the Gulf all grew double digits, reducing dependence on any one buyer.
India's cost position is still unmatched. A workforce that size and forty years of tooling is not replicated quickly, whatever the policy in Gaborone says.
Trade terms could improve. A US agreement that restores tariff parity would move the largest single number on the map.
Lab-grown eats the low end permanently. At US$ 60 a carat against US$ 760 for natural, the substitution at entry price points is close to done.
Producer countries want the work. BeneficiationBeneficiationA producing country's policy of requiring that stones mined on its soil are also cut and polished there, rather than shipped abroad as rough. It moves work away from India.See all terms in the glossary policies in Botswana and elsewhere push cutting toward the mine rather than toward Surat.
Tariffs may not reverse. The US decline is policy-driven, and policy can stay hostile for years.
Realisation keeps falling. Per-carat prices dropped in FY2025-26 and dropped faster in April to July 2026. Nothing in the data suggests a floor yet.
The jewellery growth is thinner than it looks. Implied metal volumes fell in every category.
The pivot is real, even if slow. Studded gold, silver and platinum are all gaining share of the basket, and those are higher value-add than polishing.
Inventory corrections do end. Much of the 2023 to 2025 decline was the trade working off stock built in the 2021-22 boom. That process finishes eventually.
Diversification is happening. Australia, Canada, France and the Gulf all grew double digits, reducing dependence on any one buyer.
India's cost position is still unmatched. A workforce that size and forty years of tooling is not replicated quickly, whatever the policy in Gaborone says.
Trade terms could improve. A US agreement that restores tariff parity would move the largest single number on the map.
The bull case is not strong enough today to make the polishing business attractive. What is clearer is that the two halves of this industry deserve entirely different treatment, and lumping them together as "India's gems and jewellery sector" is how investors get hurt.
Six Numbers, Twenty Minutes a Month
The data is public, free and updated monthly. Here is the routine.
GJEPC publishes a monthly Exports-Imports Master Report at gjepc.org/statistics.php, usually around the 10th to 15th of the following month. It is a letter to the Ministry of Commerce and it contains everything.
Rough diamond imports lead polished exports by a few months. A sharp fall in rough tells you what the polished number will do next. In April to July 2026 rough imports fell 26.46%.
The report gives both. Value divided by carats is realisation per carat. Rising realisation means pricing power returning. Falling realisation means the opposite, whatever the headline says.
For anything gold, silver or platinum, GJEPC prints the LBMA average alongside. If the metal rose faster than the segment, physical volumes fell.
Studded versus plain gold, jewellery versus loose stones. Share shifting toward finished goods is the thesis working. Share shifting back is the thesis failing.
Growth into the UAE, Hong Kong or Singapore may be re-routing rather than demand. Growth into Australia, Canada or France is more likely to be real.
That is a better monitoring process than most people run on stocks they actually own.
What the Trade Data Cannot Tell You
Trade statistics stop at the border. These are the things only a filing will tell you, and they are the ones that decide whether a company here is a processor or a franchise.
- What share of revenue is making charges and design, versus pass-through metal and stone cost? This is the single most important disclosure and many companies bury it. It is the direct answer to "does it get paid for the stone or for what it does to the stone."
- What is realisation per unit doing, in the company's own numbers? Same test as above, applied one level down.
- How much inventory is held, and at what gold price was it bought? In a rising metal market, inventory gains flatter profits and are not repeatable. The mechanics of this are the same as the ones in how gold, the dollar and the rupee interact.
- How concentrated is the customer base? A polishing house with three buyers has no negotiating position.
- Is the company adding lab-grown capacity? If yes, ask what return it expects at US$ 60 per carat, and what happens at US$ 40.
- What does the company do when demand is weak? Cutting houses cut wages and idle capacity. Brands run promotions and keep the shop open. The difference tells you which one you own.
Working through these is the ordinary business of reading a filing carefully, and the general method is in how to read annual reports. What you do with the answer is a valuation question, which is a separate discipline covered in valuation 101.
Key Takeaways
- Yes, India dominates the diamond trade. It cuts roughly 9 of every 10 diamonds in the world, and exported US$ 27.72 billion of gems and jewellery in the year to March 2026. But "gemstones" is mostly diamonds: actual coloured stones are US$ 437 million, about 1.6% of the basket.
- The diamond business is roughly half the size it was in 2018. US$ 25.6 billion of diamond exports in calendar 2018 became US$ 12.9 billion in 2025. Rough diamond imports, the leading indicator, fell another 26% in April to July 2026.
- 2025 was a structural crossover. Finished jewellery exports (US$ 13.70 billion) passed diamonds (US$ 12.86 billion) for the first time in at least fourteen years.
- Run the price-taker test, not the growth test. India shipped more polished carats in April to July 2026 and collected less money for them. More units plus less money equals no pricing power, and that is the whole diagnosis.
- Lab-grown is a volume story and a value disaster. Carats up 30.5%, value down 10.6%, realisation down 31% in one year to US$ 60 per carat, about 8% of what a natural carat fetches. Buildable capacity is the opposite of a moat.
- The jewellery boom is substantially a metal price effect. Set the value growth against the LBMA price and every gold, silver and platinum category implies less metal shipped than the year before. The pivot up the value chain is real, but it is a mix shift, not a volume surge.
- The US fell 44.92% on tariffs and the UAE took first place, but an unknown share of that shift is goods changing address on the way somewhere else. Treat entrepôt growth as logistics until proven otherwise.
- Ask one question of any company here: does it get paid for the stone, or for what it does to the stone? Most of this industry is paid for the stone, and should be valued as a cyclical processor rather than a franchise.
Figures from the GJEPC master reports, which run April to March, and UN Comtrade, which runs January to December, for HS headings 7102, 7103, 7104 and 7113. Retrieved 8 September 2026. Metal prices are LBMA averages. The two run on different calendars. Educational material, not investment advice; no company mentioned is a recommendation.
Disclaimer
Nothing on this site is investment advice. All content is for educational and informational purposes only. Do your own research and consult a registered financial adviser before making any investment decisions.
Finished reading? Mark this article to track your learning progress.
Software Engineer, Self-Taught Investor
Software engineer who started learning about money in 2016 after a layoff coincided with a new home loan. Went from bank deposits to mutual funds to picking stocks in India and the US, learning through YouTube, screener.in, TradingView, and the hard way. Still learning. This site is her notes made public — for education and sharing only, not financial advice.
