Your Product Can Grow 27x While Your Token Falls 85%

In just over two years, assets in Ondo’s tokenized products grew from $125m to $3,43b. Over roughly the same period, the ONDO token fell from its $2,14 peak to about $0,31.

That isn’t a bad quarter. It’s how dual-token RWA models are built. We traced more than 20 projects that paired a regulated security token with a separate utility or governance token, from July 2018 to July 2026. The pattern holds across treasuries, real estate, private credit and issuance platforms: demand for the security token, or for the protocol’s technology, did not create demand for the second token.

If you’re designing a token next to a regulated product, this is the part of the model you have to build on purpose. It doesn’t arrive on its own.

The product grew. The token went the other way.

The cleanest cases are the ones where the business worked.

The sector data says the same thing at scale. CoinGecko’s 2026 RWA report shows tokenized RWA market cap rising from $5,42b to $19,32b between January 1, 2025 and March 31, 2026, up 256,7%. Six of the seven largest RWA project tokens fell 44,7% to 98,8% over that period. ONDO lost 80,6%.

The assets came on-chain. The value stayed with whoever had a legal claim to it.

Adoption of the technology didn’t help either

The obvious fix is to make the token required for using the protocol. Polymath tested that in its purest form.

You couldn’t issue a security token on Polymath without POLY: 250 POLY to reserve a ticker, 250 more to create the token. By December 2019, 153 tokens had been deployed. Only 12 had five or more holders. Issuance kept rising through 2019 while POLY fell more than 95%, to $0,0186.

The math was never close. Even if every issuer paid in full, 500 POLY per issuance against a supply of about one billion tokens is noise. Polymath eventually moved to its own chain, Polymesh. POLYX peaked at $0,7488 in March 2024 and traded near $0,036 in July 2026.

The same split shows up in newer protocols:

• Centrifuge. In 2025, users paid $25,89m in fees. Protocol revenue was $2,22m. Revenue to CFG holders was $0, according to DefiLlama. TVL kept compounding while CFG lost 83% from its high.
• Figure. The market valued the company at $5,29b at its September 2025 IPO, with a SEC-registered yield product, YLDS, on its own chain. HASH, the chain’s token, peaked three days later and lost 85% by July 2026. We found no published mechanism linking HASH to Figure’s revenue.
• Stobox. The company still runs a tokenization business, but it bills clients in dollars. STBU is down 99,8% from its high.

Usage is real. A token that sits beside the usage doesn’t capture it.

The split is the design, not the flaw

Dual-token models exist to keep one token outside securities law. To do that, the cash flow has to live somewhere else.

OUSG is a Reg D fund interest. USDY is a note sold to non-US persons. Their fees go to Ondo Finance Inc., the USDY entity and the foundation. ONDO holders get governance and nothing else: no staking yield, no revenue share, no buyback.

That wasn’t an oversight. The SEC’s 2019 framework for analyzing digital assets listed issuer buybacks and burns as signs of an investment contract. RWA teams read that correctly and kept their governance tokens “clean”. The cost of staying clean is that the token has no claim on anything the product earns.

Research on token design points the same way. Malinova and Park show that a generic token without well-designed cash-flow rights is worse than equity. In a dual model, the cash-flow rights are deliberately handed to the security token or the fund. The governance token is generic by construction.

What actually moved the price

With no link to the product, the governance token trades on everything else.

Supply. On January 18, 2025, 1,94b ONDO unlocked, equal to 134% of the circulating supply at the time. EDEN, the OpenEden token, listed on September 30, 2025 with 80% of its airdrop vesting over 120 days. It set its all-time high that day and fell 96,7% by mid-July 2026.

Market beta. BTC ended July 2026 50,3% below its 2025 high. Most RWA governance tokens fell far harder, which means beta explains the direction but not the depth.

Failures in the asset layer. RealT raised about $140m in property tokens and announced a voluntary liquidation on July 2, 2026. Its governance token, REG, lost 77% in ten days. Tangible’s USDR stablecoin, backed by UK property, slipped to $0,51 in October 2023. It counted as fully collateralized only if you included Tangible’s own token.

Regulatory news. It moved prices less than you’d expect. When the SEC closed its probe into Ondo without charges in late 2025, ONDO rose 3,7% that day and kept sliding to year end.

Growth in AUM isn’t on this list. In eight years of data, we couldn’t find a case where it reliably lifted the second token.

The one exception confirms the rule

Maple is the only project in our set where the token tracked the business for a while, and it did two things the others didn’t.

First, it collapsed MPL into a single token, SYRUP, at 1:100 in November 2024. Second, it sent 20% of protocol revenue into monthly SYRUP buybacks from January 2025. SYRUP rose 7,7x between April and June 2025. DefiLlama TVL went from $267m at the end of 2024 to $1,65b by mid-2025. It was the only top-seven RWA token with a positive return in CoinGecko’s January 2025 to March 2026 window, up 28,6%.

Then the size of the link caught up with it. Revenue going to holders in 2025 was $2,88m. Market cap on November 1, 2025 was $485,1m. That’s roughly 0,6% a year. AUM held near $4,5b into August 2026, and SYRUP still fell 79% from its high.

A link works only in proportion to its size. A buyback that is a rounding error next to market cap and unlocks buys time, not value.

Teams without a second token often did better

The counterfactual matters. Superstate and Midas built treasury products with no governance token at all. Harbor, Securrency and Tokeny were acquired by BitGo, DTCC and Apex. The buyers paid for licenses and technology, never for a utility token.

INX ran an SEC-registered token offering. When Republic acquired it in 2025, token holders received a pro-rata share of a cash fund of about $34,3m. No utility token holder in our sample got anything comparable.

A second token is a choice, not a requirement for scale or exit.

When a second token can work

None of this says a second token is always wrong. It says the value link has to be engineered, priced and legally cleared before launch. The cases point to five conditions:

1. A written revenue path. State what share of fees or protocol revenue reaches holders, through which mechanism, from which date. “May buy back” is not a mechanism.
2. A link sized against supply. Compare annual buybacks or distributions with market cap and with annual unlocks. If unlocks are ten times larger, the link is decorative.
3. Unlocks tied to proof. Schedule large unlocks after revenue milestones, not after the narrative peaks. An airdrop at listing is a sell order with a delay.
4. No reflexive collateral. Never back a stable or security product with your own token. Under stress, the collateral and the liability fall together.
5. A legal answer before the economic one. Under the SEC and CFTC classification published on March 17, 2026, a digital commodity should not carry rights to future income or profits. In the EU, a hybrid token with dividend-like rights is treated as a financial instrument. Add revenue rights and you may rebuild the security you split off. Price that before you promise it.

If you can’t meet these, the honest design is a single regulated layer and equity for the team and investors.

In the UAE the same test reads through VARA in Dubai and ADGM in Abu Dhabi: a security token that carries a claim on the asset is a regulated financial instrument, while a governance token without a designed cash-flow right is a speculative virtual asset. The value link this article describes is not a US-only requirement — under VARA and ADGM a second token that only votes still has to justify its demand.

The question to answer before you mint

Every founder with a regulated product eventually asks whether a second token will capture the upside. Eight years of data give a plain answer: only if you write the capture into the model.

So the useful question isn’t “should we launch a token?” It’s “which dollar of revenue reaches the holder, by what right, and how big is it next to supply?” If there’s no answer, the market will price the token as crypto beta plus unlocks. Most tokens in this study ended 85% to 99% below their highs.

Test that link on a spreadsheet before the market tests it on a chart.

Data: prices from CoinGecko and Coinbase Exchange, snapshot of October 15, 2026. AUM, TVL and revenue from project reports, DefiLlama, Allium and CoinGecko’s 2026 RWA report. Fees, protocol revenue and holder revenue are reported as separate layers. This article is analysis, not investment advice.

We apply this framework to the token models we design out of Dubai for founders across the UAE and GCC.