MIOSA · Product Research
What the market is actually paying for in October 2026 — measured, not forecast — and which tokenized products the evidence supports.
The tokenization market is real, but it is roughly one-fiftieth the size the headline forecasts implied it would be by now. That gap is the most useful fact in this report: it tells you which categories have found genuine demand, which are still announcements, and where a new product can actually earn a fee.
Every major forecast of the last four years pointed at a multi-trillion-dollar market by 2030. Measured on-chain value today is a small fraction of the low end of those ranges, and the shortfall is not evenly distributed — it is concentrated in exactly the categories that were easiest to announce.
| Forecast | Published | 2030 figure |
|---|---|---|
| BCG + ADDX | 12 Sep 2022 | $16.1T |
| 21.co, "The State of Tokenization" | Oct 2023 | $3.5T – $10T |
| Citi GPS, "Tokenization 2030" | 1 Jun 2026 | $5.5T base / $8T bull |
| McKinsey, "From ripples to waves" | 20 Jun 2024 | ~$2T |
Measured against these: $39.02B distributed on-chain value today. Excluding stablecoins, as McKinsey's figure does.
The forecasts are not wrong about direction. They are wrong about timing, and they systematically over-weighted the categories with the cleanest narrative (real estate, carbon, private equity) over the categories with the cleanest cash flows (treasuries, credit, collateral).
Three categories carry almost all of the distributed value. Everything else in this report is a rounding error by comparison — which is itself the finding.
| Category | Distributed | Represented | Assets | Holders |
|---|---|---|---|---|
| Tokenized US Treasuries & MMFs | $14.64B | $90.29M | 119 | 85,876 |
| Tokenized credit | $8.09B | $41.36B | 2,766 | 205,261 |
| Tokenized commodities | $5.12B | $3.85B | 135 | 477,210 |
| Tokenized equities (Sept 2026) | $4.87B cap | — | — | — |
| Stablecoins | $303.05B | — | — | — |
RWA.xyz category pages read 25 Oct 2026; tokenized equities from CoinDesk Research's October 2026 report; stablecoins from DefiLlama. Note that RWA.xyz's own total stablecoin figure for the same day is $289.66B — the two use different inclusion rules, so do not mix them in one sum.
Within treasuries, the top products have almost no holders. This is the single most decision-relevant number in the report:
| Product | Value | Holders | Yield (7d) | Fee |
|---|---|---|---|---|
| Ondo USDY | $2.37B | 19,490 | 3.68% | — |
| Circle USYC | $2.37B | 29 | 3.38% | 0% + 10% perf. |
| BlackRock BUIDL | $2.17B | 101 | 3.71% | 0.20–0.50% |
| Franklin iBENJI | $1.71B | — | — | — |
| Franklin BENJI | $0.68B | 1,135 | 3.76% | 0.20% |
| Superstate USTB | $0.52B | 95 | 3.11% | 0.15% |
| Ondo OUSG | $0.32B | 70 | 3.69% | 0.15% |
USDY is the only treasury product in the set with genuine retail distribution — 19,490 holders against 29 for USYC, 70 for OUSG, 95 for USTB and 101 for BUIDL. It is also the only one marketed as "permissionless", with 24/7 mint and redemption "even on weekends and bank holidays". The rest are wholesale instruments with retail branding.
Before choosing a product, it is worth knowing how small the revenue pool is. Computed from the reported AUM and fee schedules above:
| Product | AUM | Fee basis | Est. annual revenue |
|---|---|---|---|
| BlackRock BUIDL | $2.17B | 0.20–0.50% mgmt | $4.3M – $10.8M |
| Circle USYC | $2.37B | 10% of ~3.4% yield | ~$8.0M |
| Franklin BENJI | $0.68B | 0.20% mgmt | ~$1.4M |
| Superstate USTB | $0.52B | 0.15% mgmt | ~$0.8M |
| Ondo OUSG | $0.32B | 0.15% mgmt | ~$0.5M |
| Maple Finance (private credit) | $4.80B | 1.06% net interest margin | ~$50.9M |
These are my arithmetic on reported figures, not figures any issuer published. They assume the fee applies to the stated AUM for a full year and ignore fee waivers, tiering and rebates.
Distributed value $8.09B, up 1.31% in 30 days, across 2,766 assets and 205,261 holders. Maple reports $4.8B AUM, $1.67B loans outstanding up 12% year-on-year, a 1.06% net interest margin and 1.19% yield outperformance, with advertised rates of 4.8–5.3% APY. It is growing while the wider DeFi lending market contracted 29% this year.
The cautionary evidence is equally concrete: a Goldfinch FIDU pool default is documented as the first tokenized distressed-debt case, and Figure's $25.49B is almost entirely represented value against just $55.21M distributed. Bridgetower's $11.06B is a single Arizona copper-gold project. Strip out those two and the category is closer to $1.5B of genuinely transferable credit.
Total supply $303.05B, down from $306.13B a month earlier and below the $308.26B peak on 7 October. USDT at $182.42B and USDC at $71.74B are 84% of it. Ethereum holds $142.92B and Tron $94.03B. The GENIUS Act became law on 18 July 2025 as Public Law 119-27, and the OCC published waiver and application guidance on 21 October 2026 — a licensing regime is now forming around an already-concentrated market.
This is not a category to enter with a new dollar token. It is a category to build on.
Market cap reached a record $4.87B in September 2026, up 13.7%, on record monthly volume of $15.6B, up 16.4%. Robinhood became the largest venue at $6.57B, up 407%, for a 42.0% share; bStocks fell 45.5% to $5.42B. Robinhood Chain volume went from $526M in July to $11.31B in September. xStocks has passed $25B cumulative volume in under eight months.
Three things make this hard to build into. The SEC's 17 September 2026 order granting tokenized securities venues relief from the exchange definition is explicitly temporary and conditional. Robinhood's EU "Classic Stock Tokens" are "derivative contracts between you and Robinhood", not shares. And Goldman Sachs, Microsoft and McDonald's are objecting to tokenized versions of their own stock — a legal fight that has barely started.
Distributed value $5.12B across 135 assets, with 477,210 holders. XAUT at $3.38B with 143,499 holders and $4.24B of monthly transfer volume, plus PAXG at $1.76B, are the market. Paxos and Tether gold accounted for almost 90% of the category's growth from $1.43B in early 2025 to $5.55B in March 2026.
Below gold, it thins fast. Silver's largest token is $7.03M with 12 holders. There is no live tokenized copper market with disclosed volume. JSOY_OIL shows $445M but with 2 holders — an issuer-sponsored structure, not a market. XAUT's $4.24B monthly transfer volume against a $3.38B market cap is the highest velocity in the entire dataset, which suggests the interesting product is gold as collateral, not another gold token.
Dubai is the live market: PRYPCO Mint offers property from AED 1,000 with 8–12% net annual ROI, VARA-licensed, with a 3-month lock-in before tokens can be listed. In the US, Arrived reports 1M registered investors, $490M invested and $124M distributed; Lofty sells shares "often around $50 or less" with a 2.5% market-order fee. Tether entered Saudi real estate in August 2026; the Dubai Land Department moved to a second phase in February 2026.
The failure is instructive: RealT is in wind-down and liquidation. Fractional property works where there is a local licence and a title registry that recognises the token, and does not work as a purely on-chain claim on a distant deed.
Ondo launched Private Markets on 5 October 2026 with a structure worth copying: "1 Note = 1 Share", a formula-driven redemption price fixed at the earliest qualifying liquidity event, self-custodiable and transferable between eligible holders, and a prominent "Not Available in US" banner. Coinbase went the other way, launching pre-IPO perpetual futures on SpaceX in June 2026 for international users only — derivatives, not tokens.
The demand argument Ondo makes is strong: 87% of US companies with more than $100M in annual revenue are privately held. The regulatory argument is unresolved.
Bittensor is the only compute token above $1B besides Venice, and the cohort sits 66–96% below all-time highs. Render's on-chain spend is $542K over 30 days and $3.6M all-time — annualised, roughly 0.7% of its $882M market cap. Akash's lease fees are $205K over 30 days. Venice explicitly excludes its subscription, API and card revenue from on-chain figures because none of it touches the chain.
This is the clearest case in the dataset of a category where real usage exists but the token captures almost none of it.
x402 processed 75.41M transactions worth $24.24M in the last 30 days, across 94.06K buyers and 22K sellers, with 99.8% of value settled in USDC. The Linux Foundation launched the x402 Foundation on 14 July 2026 with Coinbase's contribution and 40 member organisations; premier members include Visa, Stripe, Mastercard, American Express, Google, AWS, Shopify, Circle, Cloudflare and Ripple. Google's AP2, announced September 2025 with 60+ organisations, was donated to the FIDO Alliance in April 2026.
Average transaction size is about $0.32. That is the point: this is a rail for machine-scale micropayments, and it is being standardised by the same companies that own card payments.
Liquidity is negligible: 923 listings of which 108 have non-zero supply, totalling 1.74M tonnes, median price $1.00 per tonne and a range from $0.05 to $708.59. KLIMA trades at $0.0015 on $11.24 of daily volume. Klima's own carbon pools are deprecated. Verra banned tokenizing retired credits. The Guardian's 2023 finding that more than 90% of the largest certifier's rainforest offsets were worthless has not been repaired by anything in this research.
The registry layer is being rebuilt properly — Kinexys with EcoRegistry and ICR, Gold Standard with Trovio, Australia's Clean Energy Regulator on a blockchain ledger. That is infrastructure work for institutions, not a product opportunity for us.
The evidence: $14.64B in the category, but the only product with real retail distribution has 19,490 holders while the institutional ones have 29 to 101. XAUT proves retail will hold a tokenized asset at 143,499 holders.
The product: a permissionless yield-bearing dollar note, 24/7 mint and redeem, daily third-party reserve attestation, bankruptcy-remote issuer, positioned outside the US where MiCA and local regimes permit.
What it needs: an issuer SPV, a regulated transfer agent, a custody arrangement, an attestation provider, and a jurisdiction decision made before the first line of code. The fee ceiling is 0.15–0.20% — this is a distribution business, not a margin business.
The evidence: the only category where the economics are demonstrably good — Maple at $4.8B AUM, 1.06% net interest margin, 1.67B loans outstanding growing 12% while DeFi lending shrank 29%, paying 4.8–5.3%. Defaults exist (Goldfinch FIDU) and that is the point: the product is curation and disclosure, not yield maximisation.
The product: a senior-tranche feeder into a curated book of short-duration, secured institutional loans, with borrower concentration, tenor and loss history published per position.
What it needs: underwriting capability, a real loan book or a partnership with one, and a legal wrapper that survives a default. This is the highest-revenue option in the report and the one that requires the most non-crypto competence.
The evidence: a standard with Visa, Stripe, Google, AWS and Circle behind it, doing $24.24M a month at a $0.32 average ticket, 99.8% in USDC, and growing into a market where the float itself is a product.
The product: either a pay-per-call API that meters and settles over x402, or — more interesting — a yield-bearing settlement account for agent float, where an agent's unspent USDC earns treasury yield between calls. That connects the two strongest findings in this report: the retail treasury gap and the agent rail.
What it needs: almost no capital compared with products 1 and 2. It needs engineering and a view on which chain to settle on. This is the cheapest way to be early to the fastest-growing rail.
The evidence: XAUT turns over $4.24B a month against a $3.38B market cap with 143,499 holders — holders are using it as a moving asset, not a collectible. Nobody in this research was shown lending against it at scale.
The product: borrow against tokenized gold, or a gold-denominated yield product. The token already exists; the missing piece is the credit layer on top of it.
The blocker: commodity-lending regulation and custody of the underlying bars. Worth a legal opinion before anything else.
The evidence: Ondo's "1 Note = 1 Share" formula-driven structure is a genuinely clever way to give private-company exposure without issuing equity, and 87% of large US companies are private. Coinbase chose perpetual futures instead, which tells you the token route is harder.
The product: a note that pays the liquidity-event price of one share of a named private company, transferable between eligible holders, sold outside the US.
The blocker: you need a counterparty willing to underwrite the obligation at a liquidity event, and that counterparty carries the valuation risk. Do not build this without one.
One methodological warning to carry forward: the market-size numbers in this report come from four different measurement systems — RWA.xyz, DefiLlama, CoinGecko and CoinDesk Research — and they disagree materially. RWA.xyz puts stablecoins at $289.66B for 25 October while DefiLlama puts them at $303.05B. CoinDesk Research puts tokenized bond and money-market funds at $17.3B while RWA.xyz puts treasuries at $14.64B distributed. None of these is wrong; they measure different things. Every figure in this report is labelled with its source so the comparisons stay honest.
Start with product 2, the senior-tranche credit feeder, because it is the only category where the fee pool is large enough to matter — roughly $50M a year on a single competitor's book, against $15–25M for the whole treasury category. Pair it with product 3, the agent-payments metering product, because it costs little, runs on a rail that Visa and Stripe are standardising, and gives us an early position in the only part of this market growing at triple-digit rates.
Hold product 1 until the credit business has proven our distribution, then use it as the cash-management layer for the same customers. Treat gold collateral and pre-IPO notes as legal-research projects, not product projects.
Do not enter carbon, compute tokens or a new stablecoin at any point. The evidence against all three is unambiguous.
Every source below was read by the research agents on 24–25 October 2026. Grouped by category, in the order the figures appear in this report.