Kraken’s Product Page Says Commission Free. Kraken’s Legal Page Says 1%.

Four tickers went live for one company that week: SPCXx, SPCXon, SPCXD and plain SPCX.

Only one of them was stock.

SpaceX priced its IPO at $135 a share on 11 June 2026 and opened on Nasdaq the next day around $150, closing its first session at $160.95. In the run-up, Bybit, Binance and Bitget each told customers they could get in at the offering price through tokenized shares. Binance’s campaign alone drew $557 million in on-chain subscriptions across roughly 27,700 addresses and across the three venues the orders passed $1 billion.

Nobody got a share. All three cancelled and refunded. Bybit told users it received no allocation “due to the xStocks’ inability to deliver the underlying assets.” Binance handed out $1 million of its own bStocks SpaceX tokens as a consolation. Kraken customers, buying the same xStocks product through a different door, got roughly 4.28 tokens each.

And on 12 June, Ondo and Dinari listed their own tokenized SpaceX products without any of that drama, because they had never promised IPO allocation in the first place.

Same asset. Same week. Four completely different outcomes, driven by four completely different legal structures that nobody could see from the ticker.

The naming problem is a legal problem wearing a costume

Ticker convention in this market is cosmetic. You take the underlying symbol and bolt on an issuer suffix. AAPLx is Backed’s. AAPLon is Ondo’s. dAAPL is Dinari’s. Robinhood just calls it Apple.

That convention tells you which company minted the token. It tells you nothing about what the token is, and the four issuers have picked four different answers.

I want to give you one frame and then use it for the rest of this piece. Call it the Break Test: you cannot distinguish these products by looking at what you buy, only by looking at what breaks. Four questions do the work.

1. Whose balance sheet is behind it? If that entity fails, what do you hold?
2. What instrument is it? A share, a debt note, or a bilateral contract.
3. Who can redeem at fair value and at what minimum? Selling on a venue is not redemption.
4. What was actually promised on delivery? “Access” is not an allocation.

Every difference that mattered in June 2026 shows up in one of those four answers. None of them show up in a fee table.

What each one actually is

Backed Finance / xStocks — a Jersey bearer note with a European prospectus

xStocks are issued by Backed Assets (JE) Limited, a Jersey company, as bearer debt instruments structured as tracker certificates. Distribution into the EEA runs on a base prospectus approved by the Liechtenstein FMA under the EU Prospectus Regulation.

Kraken’s own risk disclosure is admirably blunt about what you get: holders “do not have ownership in any of the underlying stock or shares” and have “no voting rights, or distribution entitlements, or legal claims to the underlying stocks.” Dividends arrive as a rebase, meaning your token balance grows rather than cash landing in your account.

What Backed does well is reach. xStocks is the only one of the four that shows up natively on Kraken, Binance, Bybit, Bitget and MEXC, plus Solana venues like Jupiter and Raydium and it did more than $10 billion in combined exchange and on-chain volume within six months of its 30 June 2025 launch. Nobody else is close on distribution. One October 2025 analysis criticised Backed for not publishing audits; I’d treat that as a contested claim rather than a settled one, but it’s the kind of thing worth checking yourself before sizing a position.

Break Test answer: your counterparty is a Jersey SPV and primary redemption is gated behind qualified-investor status you almost certainly don’t have.

Ondo Global Markets — a structured note that will redeem $1 at 3 a.m.

Ondo’s tokens are structured notes issued by Ondo Global Markets (BVI) Limited, a bankruptcy-remote SPV, with holder rights governed by Swiss law. The documentation is direct: “you do not have shareholder voting rights, shareholder information rights or other shareholder rights.”

But Ondo pairs that with the best redemption mechanic of the four. Holders have a contractual right to redeem for the then-value of the underlying, the underlying sits with a regulated custodial broker-dealer, and a first-priority security interest is held by Ankura Trust Company as security agent. Since late June 2026, minting and redemption run 24/7 with a $1 minimum and no issuer fee, across 200-plus assets.

The catch is Regulation S. Tokens are offered only outside the United States, to people who are not US persons and are not ordering from inside the US.

One source conflict worth flagging: at least one write-up of the SpaceX listing described Ondo’s product as carrying full dividend and voting rights. Ondo’s own legal documentation says otherwise. Trust the issuer’s own filing.

Break Test answer: an offshore note, but with a real redemption right at a size retail can actually use.

Dinari / dShares — the only one that is genuinely a share

Dinari is the outlier and the reason is boring paperwork. Dinari Inc. is an SEC-registered transfer agent under Section 17A(c) and Dinari Securities LLC is a separately registered broker-dealer and FINRA/SIPC member. That combination is what lets it sell to US persons at all.

On 4 August 2026 Dinari opened 724 tokenized US stocks, covering the entire S&P 500, to eligible US investors buying with USDC from self-custody wallets. Holders get proxy voting, cash dividends paid in USDC, corporate actions, execution at the national best bid and offer and ownership of the backing security.

The constraint is who counts as eligible. US buyers must pass accredited-investor verification under Rule 506(c). If your income and net worth don’t clear those thresholds, the only issuer offering real shareholder rights is the one issuer you cannot use.

Break Test answer: a registered transfer agent’s book entry. This is the only product on the list where “you own the share” survives contact with a lawyer.

Robinhood — one brand, two instruments and the sharpest disclosure in the market

Robinhood is where the naming problem gets genuinely funny, because it can’t even keep its own products straight under one name.

The original in-app EU product is not a token claim on a share. Robinhood’s Key Information Document dated 1 July 2026 classifies it as an over-the-counter derivative contract with Robinhood Europe, UAB, a Lithuanian investment firm supervised by the Bank of Lithuania under MiFID II. Read the rest of that page and it gets sharper. The product “does not allow you to redeem it for shares or units in the Underlying Asset or otherwise.” The summary risk indicator is 7 out of 7, the highest class. And: “Robinhood Europe is the sole counterparty to payment claims arising from the Product for all Underlying Assets,” with no investor compensation or deposit insurance scheme behind it.

Separately, Robinhood now issues wallet-held Stock Tokens as tokenised debt securities from Robinhood Assets (Jersey) Limited, self-custodied as ERC-20s on Robinhood Chain, with dividends handled by an on-chain multiplier and redemption for cash after KYC.

So “a Robinhood stock token” is either a derivative with no redemption or a Jersey debt security with cash redemption, depending on which surface you bought it on. Both are barred to US persons.

Credit where it’s due: that KID is the most honest disclosure document any of these four issuers publishes. It says the quiet part in a regulator-mandated table.

The master comparison table

Everything above, side by side. This is the table the marketing pages don’t print.

Read down the “what you actually own” row and the point makes itself. One share, one book entry at a transfer agent, and three debt-or-derivative claims that behave like a share right up until the issuer has a bad day.

A fifth structure arrived while I was writing this

On 24 August 2026, Coinbase put tokenized US stocks live natively on Base under a new B20 token standard, starting with Apple, Nvidia, Meta and Alphabet. Non-US persons only, under Regulation S. It secured the Financial Services Permission it needed from Abu Dhabi Global Market’s regulator on 11 August 2026, and the shares sit in a bankruptcy-remote structure supervised under that regime. Chainlink supplies the price data, and Coinbase said around 50 DeFi protocols were committed at launch.

Apply the Break Test and one detail stands out. The custodian is Alpaca Securities — the same broker-dealer that already custodies about 94% of tokenized US stocks. A fifth issuer arriving with a sixth legal wrapper does not diversify this market if all of them keep parking the shares in the same building.

That is the sentence I would put at the top of any of these product pages, and none of them do.

Why every fee comparison you’ve seen is wrong

Here’s the part where most write-ups produce a tidy four-column fee table. I’m not going to, because the numbers don’t compare and I can show you why with one venue.

These are all real, all published, all currently applicable to xStocks:

• “Commission free” on Kraken’s xStocks product page.
• “The purchase price for all xStocks includes a market price + 1.0% spread” on Kraken’s own legal risk disclosure for the same product.
• Zero trading fee when buying with USDG or USD on Kraken’s instant buy.
• Maker −2 bps, taker 10 bps on the Kraken Pro order book.
• A fixed 1% trading fee on xStocks conversions.
• “A 5% spread is applied to the final IPO price at allocation” for the SpaceX IPO access programme.
• 0.5% on primary mint and redemption with Backed, reported by third-party analyses, which I could not confirm against Backed’s own documentation.

That’s seven numbers between zero and five percent, for one product, on one exchange, all true at once. “Commission free” and “includes a 1.0% spread” are both accurate, because a spread isn’t a commission. Which one applies to you depends on which button you press.

Now put the other three next to it and the incomparability gets structural.

Robinhood’s KID, which EU law forces into a standard format, discloses 0.10% entry and 0.10% exit costs, 0.00% ongoing, 0.00% transaction costs — a total of EUR 20 on a EUR 10,000 position held a year, or 0.2% annual drag. That is the cleanest cost disclosure in the category and it exists because a regulation made it exist.

Ondo charges nothing to mint or redeem. Your cost is whatever the venue or the pool charges you.

Dinari charges roughly $0.20 per order on L2s as an on-chain settlement fee, gas-based on Ethereum mainnet, with OTC quotes typically inside 10 bps above a $25,000 minimum. A flat twenty cents is not a percentage and cannot be put in the same column as one.

So: a spread baked into a price, a regulator-mandated percentage, a zero, and a flat cent-denominated settlement charge. Four different units of measurement. Any table that lines them up as “fees” is manufacturing a comparison that doesn’t exist.

The rule I’d actually use: ignore the headline number and price the round trip. Buy $10,000, sell it back an hour later and see what’s missing. That figure is comparable. Nothing on the pricing page is.

Case study: the SpaceX IPO, June 2026

Now run the Break Test on question four, the one about what was actually promised.

7 June. Bybit launches IPO Express with SpaceX as the debut listing, routed through xStocks, telling users they can “participate in the SpaceX IPO subscription using crypto and gain early access” before spot trading opens.

9 June. Bitget Wallet promotes tokenized SpaceX exposure through the same pipe.

11 June. SpaceX prices at $135, selling 555.6 million shares for about $75 billion at roughly a $1.78 trillion valuation. Binance Wallet’s SPCXx campaign, described as a “non-guaranteed subscription process,” has by now pulled in $557 million.

12 June. SpaceX opens on Nasdaq around $150 and closes at $160.95, up 19.2%. On the same day, Ondo lists SPCXon and Dinari lists SPCXD, both sourced from the open market after the bell. Bybit, Binance and Bitget cancel. Refunds go out, plus a 10% APR sweetener from Bybit calculated over four days and $1 million of bStocks tokens from Binance. Kraken subscribers get about 4.28 SPCXx each.

The instructive detail sits in Kraken’s own support documentation, published before any of this went wrong. Two sentences: “Allocation is determined by the underwriter, not by Kraken,” and “you participate at the offering price, and a 5% spread is applied to the final IPO price at allocation.”

Both facts were disclosed. Neither made it into the exchange marketing.

So “IPO price access” meant $135 plus 5%, or $141.75, if you got any at all. Against a $150 open that was still a decent trade. But the word “access” was doing enormous work, because the actual mechanism was an unsecured request for retail allocation in one of the largest IPOs ever priced, and Elon Musk had not set aside much for retail.

I don’t think this was fraud. I think it was a delivery promise made by firms that did not control delivery, which is a specific and predictable failure mode when your product is a wrapper and someone else owns the thing being wrapped. Ondo and Dinari avoided it by declining to promise anything they couldn’t buy on the tape.

The one number that surprised me: on-chain SpaceX token volume concentrated to roughly 99% on Solana and cumulative on-chain tokenized stock transfer volume crossed $20 billion for the first time that month. The delivery mechanism failed and the demand didn’t care.

Case study: the OpenAI token, July 2025

Rewind eleven months for the cleaner version of the same lesson.

On 30 June 2025, at an event in Cannes, Robinhood switched on 200-plus stock and ETF tokens for EU users and handed out small allocations of two extras: OpenAI and SpaceX. Five euros’ worth, to anyone who signed up by 7 July.

Neither was stock. Both were wrapped exposure to a special purpose vehicle holding private-company shares.

On 2 July 2025, OpenAI’s newsroom account posted: “These ‘OpenAI tokens’ are not OpenAI equity. We did not partner with Robinhood, were not involved in this, and do not endorse it. Any transfer of OpenAI equity requires our approval — we did not approve any transfer.” SpaceX hadn’t authorised its token either.

By 7 July, the Bank of Lithuania, Robinhood’s lead EU regulator, told CNBC it had “contacted Robinhood and are awaiting clarifications regarding the structure of OpenAI and SpaceX stock tokens.”

Then the sentence that explains this entire market. Vlad Tenev, on CNBC: “In and of itself, I don’t think it’s entirely relevant that it’s not technically an equity instrument. What’s important is that retail customers have an opportunity to get exposure to this asset.”

He is right about the ordinary day and wrong about every other one.

If the price tracks and the venue is liquid, the legal wrapper is genuinely irrelevant. It becomes the only relevant fact in exactly four situations: the issuer becomes insolvent, the underlying company objects, a regulator asks the issuer a question it can’t answer, or you try to get out at fair value and discover the redemption door has a $5,000 minimum and a qualified-investor gate on it.

Those are the four questions of the Break Test. That’s not a coincidence.

The decision table

What’s actually available to you, as of September 2026.

Two things this table won’t tell you. It won’t tell you which is cheapest, for the reasons above. And it won’t tell you which is safest, because “safest” depends on whether you’re more worried about issuer insolvency, regulatory action, or your own inability to exit.

What I think

The market has this ranked backwards.

By on-chain size the leaders are the debt instruments. Token Terminal data reported on 6 September 2026 put tokenized stocks at $3.1 billion, with Ondo around $947 million and xStocks near $693 million — though I’d note that RWA.xyz has published a materially different share split, so treat any single market-share figure with suspicion. The category as a whole was roughly $2.5 billion in mid-August 2026, up from about $329 million a year earlier.

Dinari, the only issuer that hands you an actual share with an actual vote, is nowhere near the top of that list. That’s not because its product is worse. It’s because its product is legal in the one jurisdiction that makes it hardest to be legal, and correctly-structured access to US retail is currently more expensive than offshore exposure to it.

I think that inverts, and not for idealistic reasons. It inverts because the 28 January 2026 staff statement drew a line between issuer-sponsored tokenized securities and everything else, and because Nasdaq’s approved rule change requires tokens to be fungible with and carry the same rights as the ordinary share. Both of those reward exactly the boring paperwork Dinari did and nobody else wanted to do.

Until then, price the round trip, read the risk disclosure rather than the product page, and assume the two disagree.

Because Tenev was right that the legal wrapper doesn’t matter on an ordinary Tuesday.

The trouble is you don’t get to choose which Tuesday you’re on.

This is part 3 of a five-part series on tokenized equities. Nothing here is investment advice. All figures are current as of September 2026; this market moves fast, and some of these numbers will be wrong by the time you read them.