SINT · The first TST · Shares.ink native token
The world's most advanced chips are made 68 miles from China. SINT accumulates the American alternative.
- The risk.
- Over 90% of the leading edge, and every chip Nvidia sells for AI, comes through Taiwan, an island China says it will take.
- The trade.
- If that supply ever stops, the value does not vanish. It moves to the one company making leading-edge chips on American soil. That company is Intel, and the United States government owns 10% of it.
- The vehicle.
- SINT is a treasury token with one mandate: hold as much Intel as possible before that day. It raises no money to get there. Its own volume and volatility turn into Intel, and a share of every other trade on Shares.ink flows to its treasury.
- If that day never comes.
- The shift still happens, only slower. The Taiwan risk is plain to everyone, Washington included, and chip production is being pulled onshore because of it. SINT is then a growing position in America’s only leading-edge chipmaker, with a floor redeemable in Intel.
- Advanced AI chips from Taiwan
- 100%
- Musk, March 2025
- US chip demand from Taiwan
- 95%
- Lutnick, September 2025
- US stake in Intel
- 10%
- since August 2025
The pitch
The entire AI build-out, every model, every data center and every trillion of market value sitting on top of them, is manufactured in one place. TSMC makes over 90% of the world's most advanced chips and, by Elon Musk's count, 100% of the advanced AI chips. Nvidia designs them. Taiwan makes them. Nobody else does at that scale.
Taiwan sits 68 miles from China and 9,500 miles from the United States. Xi Jinping said in November 2026 that complete reunification “will certainly be realised”, and gave his military a 2027 readiness goal. Elbridge Colby, who went on to run Pentagon policy, said on the record that disabling or destroying the fabs is table stakes if China takes the island, and President Trump has said that the last thing America needs is a war 9,500 miles away.
Intel is the only company in the West that designs and manufactures leading-edge chips at home. The US government has owned 10% of it since August 2025 and Nvidia put $5 billion into it. The market prices it as a turnaround story. If Taiwan goes dark, Intel is not a turnaround. It is the national-security asset that the contracts, the funding and the valuation move to. SINT is the on-chain position on that move, and its only job until then is to hold as much Intel as it can when the move comes.
Every layer stands on the one below it. The bottom layer is one company on one island, 68 miles from the country that says it will take it. Intel is not in the stack: its leading-edge fabs stand on American soil, and it is the only place the designers can go if the point gives.
Three trades in one token
- 1
The AI trade
Every model, every data center and every company valued on them runs on leading-edge chips. Intel is a claim on the hardware the whole industry is built on.
- 2
The Taiwan thesis
The supply of those chips sits 68 miles from China. SINT accumulates the one place that supply can move to, before it has to.
- 3
The on-chain treasury
SINT is the first Tokenized Stock Treasury in operation: a treasury that grows from volume and volatility, redeemable in kind, with no broker, exchange or bank in between.
Each of these is a position investors already hold on its own. SINT is the one place all three sit in a single token: the hardware AI is built on, the risk that hardware carries, and the on-chain structure that turns the position's own volatility into more of it.
Single point of failure
- Leading-edge chips made by TSMC
- 90%+
- of global advanced-node output
- Taiwan to China
- 68 mi
- the width of the Strait
- Taiwan to the United States
- 9,500 mi
- 140 times further
- Nvidia and Apple silicon made by TSMC
- 100%
- every data-center GPU, every A and M chip
Beijing is 140 times closer to TSMC than Washington is. Every deterrence argument starts from this line.
Everyone who can manufacture a leading-edge chip, by distance from China. One of the three is at home, and it is the one Washington owns a tenth of. Nvidia’s Blackwell and Rubin generations are both made by the first one. The question is who is still standing if the first one is not.
The twelve
- Guatemala
- Haiti
- Paraguay
- Eswatini
- Belize
- Saint Lucia
- Saint Vincent and the Grenadines
- Saint Kitts and Nevis
- Marshall Islands
- Palau
- Tuvalu
- Holy See
The 193 member states of the United Nations. Eleven recognize Taiwan, plus the Holy See. Together they hold about 40 million people, fewer than California. No G20 member, no nuclear power and no permanent Security Council seat is among them. The United States is not. Taiwan Ministry of Foreign Affairs, 2026.
Taiwan's standing
The island that makes the world's advanced chips is not a member of the United Nations, is not recognized by the country it relies on for defense, and has no treaty obliging anyone to defend it. Beijing calls its absorption inevitable. That is the counterparty risk under every AI balance sheet.
Fall in world output in a single year. The two Taiwan bars are Bloomberg Economics’ scenarios, January 2024, raised to $10.6 trillion in February 2026; the lighter extension is the 14% case where Taiwanese chips cannot be replaced. Covid and the financial crisis are what actually happened, world GDP at market exchange rates per the World Bank (the IMF’s purchasing-power measure shows 2009 as nearly flat, because China kept growing). The Great Depression is drawn hatched because it is not a single year: about 15% off world output, spread over 1929 to 1932. The model does two thirds of the Depression in twelve months. No single year in a century has done that.
Bloomberg Economics modelled it twice. In January 2024 a war over Taiwan cost the world economy about $10 trillion in the first year, 10% of global GDP, a larger blow than the Covid pandemic, the global financial crisis or the Ukraine war. The February 2026 update put the extreme case at $10.6 trillion. If the companies that depend on Taiwanese chips cannot substitute them, the loss reaches 14%.
Taiwan itself loses 40% of its economy in the war case. The United States loses 6.7% of GDP in one year, a Depression-year number, three times Covid. Even the quiet version, a blockade with no shots fired, takes 5% off the world and more than 3% off America.
A hit of that size is not something Washington absorbs. It intervenes, the way it did in 2008 and 2020, and this time the instrument is obvious: the one leading-edge fab on American soil, in which the government already holds 10%. That stake is what turns a commercial turnaround into a national-security asset, and national-security assets get the contracts, the funding and the protection. Intel cannot replace TSMC's volume, and nothing in this memo assumes it can. It does not have to. It has to be the only leading-edge capacity Washington controls on the day Washington decides where the capital, the contracts and the protection go.
- Nvidia's next generation, Rubin, is built on TSMC's 3 nanometer process in Taiwan. Blackwell wafers also come out of TSMC Arizona, but as of late 2026 every wafer returned to Taiwan for packaging, with no US packaging line due before about 2028.
- A 2021 US Army War College paper proposed that Taiwan destroy its own fabs on invasion. Bloomberg reported in 2024 that ASML can remotely shut off the EUV machines TSMC runs on if China invades. The capture scenario does not exist. The outage scenario does.
Who runs on TSMC
TSMC names no customers and discloses only those above 10%. Nvidia overtook Apple in 2025 after doubling its spend in a year. AMD, Broadcom, Qualcomm and MediaTek each sit below the line. TSMC annual financial statements.
What is made there
The companies at the top of this list are worth many times TSMC and Intel combined, and not one of them has a second source for the chips listed here. That is the exposure. Intel is the only American company on the other side of it.
The rehearsals
Every major encirclement drill since the Pelosi visit, placed on the calendar with its duration. The cadence has gone from one a year to two or more, the two largest fleet deployments came with no name at all, and each one practises a piece of the same operation: blockade of the ports, seizure of air and sea control, strikes on the island, and keeping the Americans out.
“They are not exercises. They are rehearsals.”
The same commander warned a year earlier that the tempo was approaching the point where a force able to launch a major operation would already be in the field under the cover of an exercise, and that traditional warning intelligence could no longer reliably tell the practice from the real thing.
That is the analysts' concern in one line: the drills are not only training, they are conditioning. Each one moves ships and aircraft into the positions an invasion or blockade would use, each one is called routine, and each one makes the next one less alarming. The day the real operation comes, its first forty-eight hours will look like every rehearsal before it, and the world will spend those hours deciding whether to believe it.
December 2025 was the first time Chinese rockets landed inside Taiwan's 24-nautical-mile contiguous zone. Beijing's New Year message, delivered as the ships sailed home, called unification unstoppable. In November 2026 the largest deployment to that date came with no name, no announcement and no end date, a foreign intelligence source telling Reuters the silence was the point: pressure without the diplomatic cost of a named drill. The world barely noticed. That is what conditioning looks like.
The trigger on the table
The December 2025 arms sale was answered with the largest drill to that date, eleven days later. A bigger package cleared Congress in January 2026 and was still unsigned when Trump met Xi that November. The precedent is one data point, and it says the ships sail within two weeks of a signature. An arms sale is a dated, observable trigger, and it is the kind of thing a position should be in place before, not after.
What the principals are saying
“The complete reunification of the motherland must be realised, and it will certainly be realised.”
“The reunification of our motherland, a trend of the times, is unstoppable.”
“Currently, 100% of the advanced AI chips are made in Taiwan. If China were to invade in the near term, the world would be cut off from advanced AI chips.”
“Taiwan, they stole our chip business. And they want protection.”
“Taiwan is 9,500 miles away. It’s 68 miles away from China.”
“The last thing we need right now is a war that’s 9,500 miles away.”
“China could turn the world off, potentially.”
“So the idea that I pitched them was, let’s get to 50-50. We’re producing half, and you’re producing half.”
“Disabling or destroying TSMC is table stakes if China is taking over Taiwan.”
“They stole our chip industry. They should cool it a little bit.”
Read together: Xi says reunification will certainly be realised. Trump says Taiwan took the business and that the last thing America needs is a war 9,500 miles away. His commerce secretary says 95% of American chip demand is met from Taiwan and wants half of it made at home. Musk, who builds the largest AI clusters, says an invasion cuts the world off from AI chips. Colby, at the Pentagon, says disabling the fabs is table stakes. There is one American company with a leading-edge fab.
The day the fabs stop
Everything above says the supply can stop. This is what happens next, and why it ends at one company. None of it is a forecast of war. It is the order in which the problem reaches each desk once the chips stop arriving, and every step uses a tool Washington has already used.
- Day one
The supply stops.
Blockade, invasion or sabotage, the result for a buyer is the same. RAND’s finding is that output from Taiwan’s fabs falls to zero quickly once they are disrupted, and the Pentagon policy chief quoted above has called disabling them table stakes.
- Weeks
Every buyer is short at once.
Nvidia, Apple, AMD, the clouds, the carmakers and the Pentagon all draw on the same inventory and it runs down together. 2021 was the rehearsal: a shortage of cheap, older chips, with no fab lost, cost carmakers $210 billion of sales and 7.7 million vehicles. This time it is the leading edge and it is all of it.
- Weeks to months
Washington rations and funds.
The Defense Production Act lets the government put its own orders at the front of any American production line and allocate what comes off it, and lets it finance new capacity directly. Both powers were used in 2020. In a chip outage there is one American leading-edge line to point them at.
- Months to years
Capital and customers follow.
Designers that never needed a second source need one at any price. Prepayments, long-term supply agreements and government purchase commitments go to whoever can make a leading-edge wafer on safe soil. Intel’s problem flips from finding customers to building capacity fast enough, with the bill underwritten.
The reason this ends at Intel is arithmetic, not sentiment. A leading-edge fab takes years to build and cannot be bought in a crisis at any price. On the day the supply stops, the capacity that exists is the capacity there is. Outside Taiwan that is Samsung, on a peninsula with its own front line, and Intel, in Arizona and Oregon. Washington can put its orders at the front of exactly one of those two lines, and it owns a tenth of that one.
In 2008 the government had a weekend to decide which companies it could not let fail, and bought its way in afterwards. This time the decision is already made and the stake is already on the books. Intel is not the casualty that gets rescued in this crisis. It is the instrument the rescue runs through.
No published plan says any of this, and Intel cannot replace TSMC's volume. The chain above is inference from the tools the government owns and how it used them the last two times. Scarce capacity with a sovereign buyer is the claim, not a one-for-one replacement.
What Washington did the last two times
- 2008Financial crisis
- Congress authorized $700 billion within weeks of Lehman failing. The Treasury ended up owning 60.8% of General Motors, and the government took 79.9% of AIG against $182 billion of support. It bought equity in what it could not let fail.
- 2020Covid
- A $2.2 trillion package passed in March. The Defense Production Act was invoked on General Motors for ventilators, and Operation Warp Speed committed about $18 billion to vaccines before one existed. It paid in advance for capacity it needed.
- NextA Taiwan outage
- The equity is already bought: 10% of Intel since August 2025. The purchase commitments have started, with up to $3 billion for the Pentagon’s Secure Enclave. The emergency powers are the same ones. Both moves are already on the board, before the crisis.
Congressional Research Service and GAO on TARP and the Defense Production Act; AlixPartners, September 2021, on the chip shortage; RAND on a Taiwan fab outage.
Why Intel is the only answer
- The only fab Washington can call its own. Intel designs and manufactures its own chips on American soil, and the American government owns 10% of it. The only other leading-edge fab is Samsung, in Korea, 700 miles from China and under 50 miles from North Korea, with a Texas plant years from leading-edge volume. Nobody replaces an exposed island with an exposed peninsula while holding a stake in the fab at home. Everyone else is fabless and depends on TSMC.
- The process is in volume on US soil. Intel 18A entered high-volume manufacturing in Arizona and Oregon in late 2025. Fab 52 in Arizona is live. 14A is next, conditional on customers.
- Washington is already a shareholder. The US government converted $8.9 billion of CHIPS Act grants into a 10% stake in August 2025. A national-security crisis is not a bailout case; it is a shareholder protecting its asset.
- The names are already inside. Nvidia closed a $5 billion placement in Intel in December 2025 and is co-developing client and data center products with it. SoftBank put in $2 billion. Microsoft and Amazon have chips committed to 18A, the Pentagon awarded Intel up to $3 billion to build its Secure Enclave, and Tesla is reported to have picked 14A for future chips. Soros Fund Management made Intel its largest disclosed US holding in the third quarter of 2026: 2.83 million shares, about $340 million at the end of September, up from 100,000 three months earlier. The relationships exist before the emergency, not after.
- Policy is pushing volume onshore. Lutnick's 50-50 pitch, chip tariffs and Trump's demand that Taiwan move production to America all point at the same beneficiary.
- The alternative is nothing. A TSMC outage does not reroute to a second supplier. It reroutes to whichever fab is standing, and the one standing on American soil is Intel.
The Taiwan discount
Market value, November 2026. TSMC trades near 25 times earnings, a multiple analysts describe outright as carrying a Taiwan discount. The dashed extension is the part of the franchise the market refuses to pay for while it sits 68 miles from China.
$1.9 trillion is not what TSMC's franchise is worth. It is what the franchise is worth after the market subtracts the risk of being 68 miles from a country that says, in public, that it will take the island. The same monopoly on the leading edge, incorporated in Delaware, fabbing in Arizona, under the US legal system, the US defense umbrella and US industrial policy, would carry no such discount and a far higher multiple. The discount is real, it is large, and it is sitting there unclaimed.
There is exactly one listed company positioned to earn that franchise on safe soil, and the market valued it at $320 billion in November 2026: a turnaround that ran about sixfold into June 2026, fell 45% that October on weak guidance and a plant shutdown in Israel, and had no significant external foundry customer at that date.
That is the asymmetry. The turnaround is optional. The hedge is structural. If Taiwan runs, Intel is the one leading-edge fab with no China or North Korea risk in its price: an American company on American soil, backed by its own government. If it does not, Intel is the only fab standing and the discount the market applies to TSMC becomes the premium it has to pay for Intel.
Where the value goes
Value does not vanish. It moves.
The headlines after an invasion will say that trillions of dollars were wiped out. They will be wrong about the verb. The world's need for chips is not destroyed with the fabs: every data center, phone, car and weapons programme that needed a leading-edge chip the day before still needs one the day after. What is destroyed is the supply. The value that sat on TSMC, and on every company that depended on it, comes loose and looks for the next place that can do the job.
Markets behave like physics here. A vacuum does not stay empty. Air rushes in from wherever it is, at once, without being asked. Take the largest franchise in the industry out of the world and capital, customers and government money rush toward whatever can fill the space. On the leading edge, outside Asia, there is one candidate.
Intel cannot fill that space on day one, and it does not have to. Scarcity is repriced first: the only leading-edge wafers left are worth far more than they were the day before. Capacity follows, built with other people's money, because every customer and every government has the same problem and the same single address. That is why we believe Intel does not merely recover in this scenario. We believe it ends up as the most valuable company on earth, because it inherits the most valuable position on earth: the one the whole stack rests on.
That is our belief about one scenario, not a forecast. The bear case below sets out what stands against it.
Catalysts and the bear case
What we are watching
- A Taiwan arms package being signed. The December 2025 sale was answered with the largest drill to that date within eleven days. The next one, about $14 billion, cleared Congress in January 2026 and was still unsigned that November. If a blockade or an invasion has a trigger, a signature like this is the most likely one on the calendar, which is why it is the event we watch most closely.
- A volume external foundry customer on 18A or 14A. Microsoft and Amazon are committed, but Intel's FY2025 filing said no external customer was significant. In late 2026 its CEO said a couple would be announced within two quarters, and bought $9.7 million of stock himself after the October drop.
- Chip tariffs and the 50-50 production push turning into signed commitments.
- Further sovereign capital: a second government tranche, defense procurement, or allied orders.
- Any escalation in the Strait. The position is designed to be held before that, not bought after.
What could go wrong
- 2027 is a readiness goal, not a date. The 2026 US intelligence assessment says Beijing has no fixed timeline, and the PLA purges have set readiness back. The hedge can stay unexercised for years.
- TSMC keeps moving to Arizona and the dependence thins before the risk lands.
- Intel's own filings say 14A could be dropped without an external customer. The execution risk is real.
- Intel cannot absorb TSMC's volume. In an outage it is the scarce capacity, not the replacement.
- Intel is also a TSMC customer. Some of its recent processors are built on TSMC nodes, so an outage hurts its product business before it helps its foundry.
- A sovereign shareholder cuts both ways. Capacity directed in the national interest is not necessarily priced for shareholders.
- In the outage itself every risk asset sells first. An on-chain token is not spared the first leg down.
- Intel is volatile in both directions and the treasury carries that volatility in full.
Two scenarios, one position
One position, two ways to be right. The larger one is the one nobody wants.
The mandate is simple: accumulate as much Intel as possible before the thesis plays out. Every mechanism below exists to do that without raising capital, issuing anything or asking anyone's permission. The stock is bought by the flywheel, on every trade, in every market, for as long as the platform trades. When the move comes, what matters is how much Intel is already in the treasury.
Why we built it
Shares.ink exists to prove one idea: a token is worth more than its ticker when its treasury holds a real asset and a real argument for holding it. We call that a Tokenized Stock Treasury. A new kind of instrument needs a first example that can be judged on its merits, so we did not launch a platform token that stands for nothing. We took the one position we would want to own ourselves and built the native token around it.
That position is the Taiwan thesis. It is specific enough to be wrong, which is the test of a real thesis: a named asset, named triggers, a dated record and a bear case anyone can check. SINT is that thesis placed inside the structure, so a reader can see in one token what every treasury on the platform can be: community-owned, bound to a mandate, and accumulating a real equity by rule rather than by promise.
It is one use of the structure, not the only one. The same contracts carry any stock and any conviction, and none of it touches a broker, an exchange or a bank. It all happens on the chain.
- A community and a company
- A treasury on a small company’s stock lets a community build a position in it together, in the open, with every share visible on the chain and the rules fixed in the contract.
- A flywheel on a flywheel
- A treasury on a Digital Asset Treasury company, a Strategy, a SharpLink or a BitMine, accumulates that company’s shares as its own volatility grows, while the company accumulates Bitcoin or Ether underneath it.
- A meme with a balance sheet
- A token can be nothing more than a joke and still be backed by real stock, with a treasury that grows with every swing in its price and a floor it can be redeemed against.
We laid the foundation. What gets built on it is open. SINT is the one we put our own name on.
Why a TST and not the stock
SINT launches on a bonding curve against INTCx, the xStock for Intel, and graduates into a Uniswap v4 pool with its own staking contract and governor. The same contracts as every other TST, with no special powers, no presale and no allocation to the protocol.
After graduation, 10% of every SINT trade runs the flywheel: on a buy that slice is burned, on a sell it becomes Intel in the treasury and streams to stakers. Volatility in the token adds Intel to the treasury instead of diluting holders.
Every TST on Shares.ink sends 2% of each trade to the protocol. That share goes to the SINT treasury, which funds liquidity, activity incentives and buying back and burning SINT. SINT is the one treasury that compounds with the whole platform.
Any holder can redeem SINT in kind, at any time, for a pro rata share of the unallocated Intel in the treasury. Stakers receive the Intel the treasury streams, and dividends compound inside it. Holders vote on reward schedules and treasury sales.
How to measure it
A treasury company is judged on one number: how much of the asset stands behind each share, and whether that number is rising. Strategy reports Bitcoin per share. SINT is read the same way, and all three figures below come straight from the chain.
- Intel per SINT
- Intel in the treasury ÷ SINT not burned
- How much Intel stands behind one token. The equivalent of Bitcoin per share, and the figure to track over time.
- NAV per SINT
- Intel per SINT × Intel’s price
- The net asset value behind each token in dollars. Redemption in kind works from this figure, so it is the floor.
- Premium to NAV
- SINT price ÷ NAV per SINT
- What the market pays above the backing for the thesis, the accumulation and the platform. Treasury companies trade the same way. Below 1, redeeming is worth more than selling, which is what holds the floor.
Intel per SINT is the number the mandate exists to raise. It goes up when a purchase burns supply and when the treasury takes in more Intel than it streams out, and stakers collect the stream itself. The share price of Intel can stand still while it climbs.
Questions
Why SINT and not just Intel?
If you want Intel, buy Intel. It is the cleaner instrument and nothing here says otherwise. SINT is built for a different job: more exposure to the same thesis for each dollar, and a position that keeps growing without anyone putting more money in.
Size. Intel was a $320 billion company in November 2026. For the share to triple, the market has to find another $640 billion. SINT starts as a small token on top of a treasury of Intel, so the same repricing of the thesis moves it much further. It is a higher-beta claim on Intel and on the Taiwan risk, and that cuts both ways.
Accumulation. An Intel share is a fixed claim: one share today is one share on the day the thesis plays out. SINT is not fixed. Every sale of SINT turns into Intel for the treasury and its stakers, every purchase burns supply, and 2% of every trade in every other treasury on Shares.ink arrives on top. The Intel working for holders grows with volume and volatility while the supply it is spread over shrinks.
The platform. SINT is the native token of Shares.ink, so it carries the platform as well as the stock. The more treasuries launch and trade here, the more flows to SINT. An Intel share has no such second engine.
The clock and the floor. SINT trades around the clock, which matters for a thesis whose trigger will not wait for New York to open. And any holder can redeem it in kind for a pro rata share of the unallocated Intel in the treasury, which puts a floor under the token that moves one to one with the stock.
Higher beta means it can fall harder than Intel as well as rise faster. Read the bear case above before the comparison.
- Is SINT a tokenized Intel share?
- No. INTCx from xStocks is the tokenized share. SINT is a token whose treasury holds INTCx. Its price is set by its own market, and only its redemption floor moves one to one with Intel.
- What if China never moves on Taiwan?
- Then the calm branch plays out: Intel is an American leading-edge fab backed by its own government, winning customers one at a time, and the treasury keeps accumulating it the whole way. The position does not need the crisis to work. The crisis is what makes it work fastest.
- Where does the Intel in the treasury come from?
- From volatility, not from fundraising. Every swing in the token is trading, and trading is what feeds the treasury: 10% of every SINT trade runs the flywheel, where a buy burns that slice and a sell turns it into Intel. The more the price moves, in either direction, the more Intel the treasury takes in. On top of that, 2% of every trade in every other treasury on Shares.ink goes to the SINT treasury.
- Can anyone change the rules or move the treasury?
- No. SINT runs on the same contracts as every other treasury on the platform, with no admin key. Nobody, Shares.ink included, can change its flywheel, mint more of it or move its treasury outside a governance vote of its holders.
- Why does the treasury accumulate now?
- Because the position has to exist before the event, not after. Every trigger on this page is public and dated, and by the time one of them fires the repricing has already happened. What counts on that day is how much Intel is already in the treasury.
- What is the main risk?
- Intel is volatile and SINT is more volatile than Intel. The thesis can take years, and in the first days of a real crisis every risk asset sells before anything is repriced. The bear case above lists the rest.
The record
What SINT is not
SINT is not a tokenized Intel share and does not track Intel's price one to one. INTCx from xStocks is the tokenized share. SINT is a token with a treasury, and the treasury is what holds Intel, as INTCx. Its price is set by its own market, its supply shrinks on buys, and its floor is what the treasury returns on redemption.
This memo describes a thesis, not a forecast. Every quote and figure is sourced to public reporting and filings and carries its own date. Prices and market values are as of the date shown and move every day; the argument does not rest on any one of them. Nothing here is investment advice.