TechFlow Logo
Login/ Sign up
ETH Gas
Gwei
Fear
gas
Ondo has made a decision.

Ondo has made a decision.

2026.08.13
Share
If a project runs out of funds or the technology collapses completely, then the decision has already been made.

Written by: Thejaswini M A

Translated by: Block unicorn

As someone who hates making decisions, I understand Ondo's dilemma.

The difficulty lies in facing the permanent loss from unchosen paths, accepting that time is linear, and bearing the grief of unchosen potential.

Wasting 20 minutes in front of supermarket shelves, holding a tub of protein powder, Googling the difference between isolate and concentrate, calculating the price per scoop, and comparing the 5-pound tub on Amazon with others. Plus the struggle of choosing flavors, it's really headache-inducing! Finally at checkout, I just grabbed a Snickers bar and temporarily shelved the decision paralysis.

Similarly, obvious decisions are easy to make. If a project runs out of funds or the technology collapses completely, then the decision has already been made.

Time is the only resource you can't buy; every month wasted on the wrong thing is a month wasted in vain. If you fail at the last hurdle, it means everything invested before is lost in vain.

Ondo had to give up a year and a half of effort. They had a functioning testnet and a tokenized treasury settlement system that JPMorgan had tested. This was not easy, so we dive deep.

On July 27, Ondo Finance published a blog post introducing the Ondo Network. This was vastly different from what they announced 17 months ago.

Ondo announced the launch of Ondo Chain, a Layer 1 network for tokenizing real-world assets, featuring permissioned validator nodes, passive income staking functions, and native bridging capabilities. Its major supporters include Franklin Templeton, WisdomTree, and Wellington. Kinexys, under JPMorgan, had settled tokenized treasury bonds on the testnet using Chainlink.

But now, CEO Ian de Bode confirmed that the company will not run both systems simultaneously. They abandoned the blockchain project.

They removed the execution and validation layers, keeping Ethereum as the settlement layer.

Code executes within hardware enclaves (also known as Trusted Execution Environments). These enclaves are sealed modules on server chips. Code runs within them, and server operators cannot read or modify it. Each enclave generates a hardware fingerprint for its active code. Changing a single byte will change this fingerprint, causing verification to fail.

Verification relies on a certain number of independent verifiers. Before launching the enclave, they compare the enclave's fingerprint with approved code. They also hold split key fragments, meaning no single operator (including Ondo) holds the complete key. Keys can only be reassembled within an enclave whose code has been verified.

Settlement still occurs on Ethereum. Asset transfers still occur on the public chain, same as before.

Ondo believes blockchain bundles four independent functions: consensus, replication, transparency, and final state settlement. Consensus and replication reduce system speed, while transparency makes all information public. These trade-offs are an unnecessary burden for order matching. Therefore, Ondo moved execution off-chain while keeping settlement on the blockchain.

They said: "The matching engine needs a deterministic sequencer, and replication operations add latency to every order."

I don't know about you, but I remember Accenture saying this in 2022.

The Australian Securities Exchange spent over seven years trying to replace its clearing and settlement system CHESS. They attempted to replace it with Digital Asset's distributed ledger system, even acquiring an 8.5% stake in the company in 2016. The launch target for the new system was repeatedly delayed until finally settled after 2021. By the end of 2022, Accenture's audit found that the software development progress was only 63% complete, and the launch date was indefinitely postponed.

It pointed out: "Distributed systems introduce higher latency."

Due to the project's failure, the Australian Securities Exchange (ASX) wrote off losses of up to 255 million Australian dollars, and the Australian Securities and Investments Commission (ASIC) sued the exchange for issuing misleading market updates. Additionally, market brokers also lost tens of millions of Australian dollars after connecting to a system that was never deployed.

The New York Stock Exchange and Nasdaq use private matching engines operated by a single company. They do not replicate order books across hundreds of computers, and traders do not want their real-time orders visible to the entire market. Then, trades are submitted to DTCC, which is slower, but the entire industry shares its service, and its settlement amount exceeds any other institution on earth. DTCC's subsidiaries processed securities transactions worth 4.7 quadrillion US dollars in 2025.

They say that currently no blockchain can handle such large trading volumes. DTCC is tokenizing anyway because settlement only occurs on business days. DTCC officially went live with trading on July 15 and plans to scale up in October. These tokens sit on top of existing post-trade processing mechanisms.

Crypto once promised to merge trade execution and settlement into one machine while retaining the advantages of both. Rollups achieved Ondo's vision to some extent.

Speed is not a reliable indicator for judging whether a blockchain is needed. Every trading platform wants to pursue speed, and most platforms can enhance speed by paying. A more reasonable criterion is: besides yourself, is there anyone else who needs to write data to this ledger?

Ondo operates a trading platform. It matches perpetual futures contracts for Nvidia, crude oil, and the S&P 500 index. These contracts track prices and settle in cash, so no one actually receives stocks. The tokenized part is the collateral. You use tokenized treasury bonds as margin. They hold the order book, and no external party needs write access to their margin engine. Putting it on the blockchain would mean running multiple database copies (when they only need one), making the order book public (their market makers would strongly object), and managing a validator set.

Instead, without any blockchain technology, the platform reached a total contract value of 6 billion US dollars in three weeks, averaging over 300 million US dollars per day.

Ondo failed this test, so blockchain makes no sense for them. Some companies passed this test. The difference lies in who their target customers are. If your users are the only ones who can access the ledger, then you are operating a trading platform. If strangers create content on your ledger without permission, then you are operating a market, and markets need blockchains because the entire product is the part you cannot control.

The situation with Base is completely different. Coinbase sells block space for others to build projects. Therefore, the data Coinbase publishes reflects other people's activities. In Q2 2026, over 90% of proxy stablecoin trading volume ran on Base, and 97% of on-chain proxy trades used the x402 protocol, which processed 160 million payments in the past year. Stablecoin trading volume on Base increased 7-fold year-over-year, reaching 19 trillion US dollars so far in 2026. And almost none of this trading volume was occupied by Coinbase itself.

If Coinbase moved Base's execution to off-chain private zones, the product it sells would cease to exist.

Robinhood's situation is quite awkward; it is in a middle ground. The Chain went live on July 1. As an integrated product of Arbitrum Orbit, stock tokens have circulated in over 120 countries/regions. But no matter how you hold them, stock tokens are debt instruments issued by Robinhood Assets Jersey Limited, not stocks. This makes Robinhood violate the guidance issued by the U.S. Securities and Exchange Commission (SEC) in January 2026, which drew a line favoring issuer-backed tokenized stocks over third-party synthetic stocks. Self-custody can protect you from Robinhood losing your tokens, but it cannot protect you from Robinhood's bankruptcy.

What Robinhood really gains from blockchain is a distribution channel. Uniswap launched its token issuance platform pools.trade on Robinhood Chain on August 5. Letting other people's apps stand shoulder to shoulder with your 23 million users is still a good business.

For the past three years, companies have been building their own supply chains, but the results were not ideal.

dYdX conducted a large-scale test of the app-chain concept. At the end of 2023, the protocol detached from StarkEx and launched its own Cosmos chain, performing order matching directly on a decentralized validator set. Architecturally, it won this debate, but the market share was completely different.

In early 2023, dYdX accounted for 73% of the total trading volume of decentralized perpetual futures. By 2026, its market share had fallen below 3%. The platform's current monthly trading volume is about 2.5 billion to 3 billion US dollars, while Hyperliquid's trading volume is as high as 180 billion to 208 billion US dollars. dYdX's total value locked is 100 million to 150 million US dollars, far lower than Hyperliquid's 6.2 billion US dollars.

Speaking of Hyperliquid, it constitutes the best argument for having one's own blockchain. Its matching engine HyperCore runs consensus directly. It uses 27 validator nodes (only 5 at launch) and does not impose slashing. The project does not open-source the code but releases the final program. You can run it, but you cannot understand how it works. Joining the active validator set requires over one million HYPE.

Suppose there are 27 validator nodes running publicly unreadable code, processing 200 billion US dollars in transactions monthly; while another solution is an enclave running audited code checked by certifiers. Both models rely on a tight operational circle and cryptographic techniques. Although calling one "blockchain" and the other "off-chain" makes people feel they are completely different from reality.

Unichain exists to capture Uniswap's trading volume and send sequencer fees to UNI holders. But Uniswap processes about 1.5 billion US dollars in transactions weekly through the existing network, while Unichain holds only 532 million US dollars of its total value locked (TVL) of 5.76 billion US dollars. Additionally, Uniswap's fee switch mechanism already consumes 90 million US dollars worth of UNI annually without needing to build its own chain.

From a broader perspective, among all secondary collateral, three holders hold 91% of the total value locked (TVL), and only twenty hold any TVL. Ethereum itself holds 65% of primary collateral (L1), while Base holds 53% of secondary collateral (L2).

Blockchains are shutting down faster: Swell ended the Swellchain project in June, Mint Blockchain collapsed in April, and Polygon closed its zkEVM sequencer on July 1. RootData records show that by early 2026, approximately 100 crypto projects had perished, with networks accounting for a large portion.

Why is everyone launching products?

Until recently, applications were priced based on revenue; chains were priced based on potential.

When Ondo announced the launch of a blockchain with staking rewards, its token price surged 11%. However, when the network officially went live, nothing changed. The ONDO token does not provide any yield, and the growth in protocol trading volume did not bring any direct demand. Management stated that the token's utility will be gradually realized in subsequent proof-of-stake and slashing mechanisms.

But security enclaves are not easily obtained. Over the past year, security chip hardware has suffered heavy blows. Hackers repeatedly bypassed Intel and AMD's security protections using cheap hardware. They first cracked memory protection with tools worth 50 US dollars, then stole Intel's master security key with equipment worth 1000 US dollars. Last October, researchers released the TEE.fail attack, named after the security vulnerability of TEE (Trusted Execution Environment, or "security enclave"). Security enclaves are effective because chips promise no one can peek inside. Researchers used a circuit board worth 1000 US dollars connected to the computer's RAM memory to eavesdrop on plaintext data transmitted between memory and processor, thus breaking this guarantee.

They stole Intel's signing key, which is the digital authentication credential used to prove code runs securely in a real security environment. Using the stolen key, they created a forged security certificate on BuilderNet (an Ethereum system for organizing pending crypto transactions). This allowed them to secretly view private transaction instructions while the system automatically checked and reported everything was completely secure.

These attacks require physical access to the machine. Therefore, Ondo's top security measure is essentially physical security, such as who installed the device, which warehouse the device is stored in, and who transported it. Is this worse than trusting 27 validators who will not be slashed? Perhaps not. But this is physical security after all, it is real-world risk, so pricing should reflect this.

Ondo has been facing reality and plans to open certification to guarantor operators, while splitting code verification, key custody, and server hosting into different entities. The final state will be published on-chain so external observers can question bad transactions, and proof-of-stake and slashing mechanisms will be added later.

Might as well just build the Ondo Chain...