Six approaches in production. Three trade-offs that decide which fits your stack. Honest comparison of the incumbent US custodian model, Dinari, Ondo, Backed, Superstate, and Flo Finance.
I keep getting asked the same question on intro calls. "How is Flo different from the incumbent US custodian model / Ondo / Dinari / Backed." I've answered it about forty times in the last quarter, and the version I gave the first time was bad. The version I gave the fortieth time is what's below. I'm writing it down so the forty-first conversation can start one level deeper.
The tokenized US equity category isn't abstract anymore. As of April 2026, more than $1 billion sits in tokenized public equities and ETFs across a handful of platforms. At least five distinct architectures are in production. The marketing language across them is similar enough to be confusing. Tokenized stocks. 1:1 backed. Regulated infrastructure. The actual stacks differ in ways that matter for any fintech, neobank, or wallet picking a vendor.
Honest disclosure first. Flo is in the map too. I'm a builder, not a neutral analyst, and I'll flag that wherever it matters. The architecture decisions speak for themselves.
The first is custodian-native tokenization, offered by the incumbent US custodian direct. The fintech integrates that custodian's API. The custodian holds the underlying as a self-clearing FINRA-regulated firm. Tokenization sits inside the custodian itself. That custodian self-reports ~94% market share of tokenized US stocks and ETFs and ~$480M in tokenized AUC (December 2025). Strong because the custodian is mature and self-clearing means no second-tier dependency. Exposed because the fintech inherits custodian-specific quirks with no abstraction layer, and the incumbent is incentivized to capture more of the stack over time.
The second is the regulated tokenization wrapper, used by Dinari. Dinari operates as a US-registered Transfer Agent and runs a FINRA-registered subsidiary (CRD 329672). The user deposits stablecoin or USD, Dinari's subsidiary routes the order to the incumbent US custodian, that custodian holds the underlying, and Dinari mints the dShare on-chain backed 1:1. dShares retain shareholder rights including dividends and voting where permissible. Public footprint: 85+ countries and 150+ tokenized US equities live on dinari.com (200+ planned for the Flow Traders 24/7 launch). $12.7M Series A in May 2025, led by Hack VC and Blockchange Ventures, with VanEck Ventures, F-Prime, and the Avalanche Fund. Plus the Dinari Financial Network on Avalanche and a Flow Traders partnership for 24/7 liquidity. Strong because they're the first regulated tokenization stack in the category and full shareholder rights are real. Exposed because they still route through the same incumbent custodian for clearing, so the same single-custodian concentration shows up underneath. Public-equity-only.
The third is DeFi-native wrapped certificates, Backed Finance's xStocks. Swiss/Jersey-domiciled. Tracker certificates (debt claims) that are 1:1 collateralized by the underlying US equity, with the underlying held by a regulated European custodian. The token represents a claim against Backed as the issuer; bToken holders do not have direct share ownership. Designed for permissionless DeFi composability: trades on DEXes, can be used as collateral, no on-chain whitelist restrictions in the default form. ~$160M TVL across Solana and Ethereum, 60+ tokenized equities live, $20B+ cumulative trading volume since the June 2025 Kraken launch (RWA.xyz, Kraken blog). Strong because DeFi-native composability is real and the European wrapper sidesteps US securities-law complexity for non-US users. Exposed because the tracker certificate model means no shareholder rights, issuer-counterparty risk concentrated at Backed, and the wrapper doesn't fit US-domiciled users at all.
The fourth is the custodial wrapper with chain-agnostic distribution, Ondo Global Markets. Ondo issues GM tokens redeemable through collateral agents. Underlying shares are held at US-regulated institutional custodians, with the incumbent as primary. Available on Ethereum, BNB Chain, Solana. Ondo Global Listing in February 2026 added same-day tokenization of IPOs. ~$3B in tokenized assets across Treasuries and equities, with the equities subset around $700M (RWA.xyz, April 2026). Strong because they're the largest scale by TVL, multi-chain distribution is out of the box, and same-day IPO tokenization is novel. Exposed because GM tokens don't pass through shareholder rights, the single-custodian dependency is intact, and the universe is public-equity-only.
The fifth is native regulated tokens, Superstate Opening Bell. The token itself is the regulated share. Not a wrapper, not a synthetic, not a tracker certificate. Public companies issue or convert their exchange-listed common stock directly onto Solana or Ethereum. Superstate's USTB and USCC funds together exceed $1B in AUM (Q1 2026), with USTB transitioning to Invesco management in May 2026. Strong because it's the cleanest legal model with no wrapper or counterparty layer between token and share, and the best long-term position if regulators converge on "the token is the share." Exposed because it requires the underlying public company to convert or issue native, which is high friction, and the universe is small. Not a fit for a fintech that wants to give users AAPL exposure today, because AAPL isn't a Superstate-native token.
The sixth is the multi-custodian, multi-asset SDK layer. That's what I'm building. Flo Finance is a developer SDK that abstracts over multiple institutional custodian partners and supports both public and private market tokenization through the same surface. Buy, sell, positions, borrow. Each endpoint routes to the right custodian integration based on security type, jurisdictional gating, and execution quality. Every primitive is asynchronous via an on-chain order_id: the create-order tx is atomic at the chain level (stablecoin pulled or tokens burned alongside the order record), and the terminal action lands via settle(order_id) once the custodian leg confirms on the standard cycle (T+1 for US equities, typically T+2 elsewhere). Because settle reads the on-chain order before doing anything, the reorg / double-spend window is closed at the contract level — not by waiting for finality, not by trusting an off-chain indexer. Where I think we're strong: two-custodian redundancy is structurally different from any other public-equity tokenizer, public-and-private through one SDK isn't covered by anyone in the current market, and the settlement framing is honest — there is no inline-from-pool fast path, just an atomic on-chain claim that makes the custodian cycle the only thing the user is actually waiting on. Where we're exposed: no production scale comparable to the incumbents yet. Regulatory positioning sits downstream of our institutional custodian partners.
There isn't a single right answer. The right architecture depends on your users' geography, your appetite for vendor concentration, your need for shareholder rights, and whether your roadmap requires private-market tokenization. The map above is a tool, not a verdict.
Two questions I'd ask any vendor in this category, including me.
What happens when your primary custodian has an outage. If the answer is "we'd be down too," you're inheriting their concentration. If the answer is "we route to the secondary custodian automatically," ask for the runbook. The runbook is the artifact, not the slide.
What exactly is settling instantly, and what is settling on standard timelines. If the vendor can't separate the user-facing leg from the broker-exchange leg crisply on the first ask, the marketing is ahead of the architecture, and that's the kind of gap that survives a venture pitch and dies in a regulatory review.
The category is still forming. The vendor map will look different in twelve months. The fault lines above will still be the right way to evaluate any new entrant, including, eventually, ours.
I might be drawing some of these lines wrong. We're all figuring this out at the same time. If you're seeing something different in your own diligence work, I'd actually like to hear about it.
Note on figures: competitor and ecosystem statistics in this post reflect public disclosures as of writing. They move quickly. Verify against the cited primary sources (incumbent custodian PR, Ondo docs, Dinari docs, RWA.xyz) before using them in your own decisions.
