Castle Labs says public blockchains are being pushed toward a different business model as revenue keeps concentrating at the application layer. Over the past two years, 14 crypto companies generated more than $200 million in revenue, and only one of them was a public chain: Hyperliquid.
That gap looks even wider when measured against other chains. In the past 30 days, Arbitrum generated just $430,000 in revenue, while Hyperliquid produced about $58 million, a spread of more than 100x.
The firm’s broader point is that applications have always been better revenue engines because they face customers directly and have to deliver value in a way users can feel. Public chains, by contrast, have spent years supporting ecosystems with grants and protocol upgrades. Castle Labs argues that this approach now needs to change. If chains do not move toward serving paying customers, they risk running down their treasuries.
Block space is no longer enough
In Castle Labs’ framing, the issue is not simply that one chain earns less than another. It is that selling block space no longer works as a standalone business model for public chains.
The report says chains are starting to react. Early moves are already visible in models built around product studios, app distribution, payment rails, and vertical SaaS stacks. Castle Labs expects more chains to move away from the idea of neutral infrastructure and toward owning specific verticals instead.
Ostium exploit puts offchain price paths under scrutiny
The report also revisits last week’s exploit at Ostium. Its LP vault was attacked and lost 23,752,746 USDC after the offchain infrastructure used to supply prices to the protocol was compromised.
According to Castle Labs, the attacker submitted invalid price reports that appeared legitimate, then opened and closed large positions immediately using those reports to extract artificial profit from the vault. In practical terms, the attacker found a way to push false price updates through an approved path, turning losing trades into apparently profitable ones and draining the LP vault.
Castle Labs says the episode was particularly painful for Ostium because the protocol’s core product is built around bringing offchain markets onchain. Stocks, commodities, and foreign exchange listed on Ostium do not have native onchain prices. The protocol has to import them, and it also has to trust them.
Once contracts depend on that trust, users depend on it too. A false price that clears checks can quickly become more than bad data. It can turn into bad execution, bad vault accounting, and real LP losses. For a protocol like Ostium, oversight of the full offchain-to-onchain pricing path sits at the center of the product itself.
Ostium said trader collateral was isolated and unaffected, and that trading contracts were frozen within 60 minutes. Castle Labs describes that as fairly quick, but asks a harder question: how much damage should a single bad price input be able to do before a protocol catches it?
The report notes that Ostium already accepts some of the tradeoffs common in traditional finance. Many of the markets it offers are not truly 24/7 because the underlying assets are not either. If users already accept trading hours, stale prices, market closures, and liquidity gaps, Castle Labs argues that stricter controls around price updates, trade sizes, withdrawals, and timing should be easier to justify, not harder.
It then lays out what those controls might look like. If an authorized path can update prices, that path should be tightly controlled and monitored. If a fresh price update can support large trades or withdrawals, circuit breakers tied to size and timing should be in place. If an attacker probes the system with smaller trades first, monitoring should catch that pattern before the vault is drained.
For protocols that bring offchain markets onchain, Castle Labs says these controls should not be treated as optional security features. They should be embedded into the product and presented as part of the product itself.
Onchain options need abstraction
Castle Labs also turns to adoption in onchain options. After publishing its report on what it called a revival in onchain options, the firm hosted a livestream with Kalshi, Rysk, GammaSwap, and Block Scholes. One point came up again and again: options may be powerful, but marketing them as “options” is often the worst way to sell them.
Most users do not want to think in terms of Greeks, expiries, strike prices, or volatility surfaces. They want yield, leverage, protection, or a simple way to express a view.
That is why the products with the clearest path to adoption are, in Castle Labs’ view, not plain options venues. The firm points instead to yield vaults, short-dated binary options, structured products, and prediction markets.
Rysk’s Dan summed it up in a line highlighted by the report: “Options are not the product; the benefits of options are the product.”
Rysk said its newer product recorded more than $1 billion in open interest last year, driven mainly by experienced DeFi users looking for asset yield rather than users arriving as options traders. Castle Labs says quarterly notional charts show how quickly that product found demand.
Kalshi said it now handles 86% of global crypto binary options volume and about 70% of global prediction market volume. The company’s view is that 15-minute markets are the best time window for crypto binary options because users can understand the payoff, time frame, and risk more easily.
GammaSwap appears in the report as another example of taking options complexity away from the end user. In V1, users could borrow liquidity from an AMM, and the AMM behaved a lot like an options seller. But once Greeks, exotic payoff shapes, and fragmented liquidity become part of every user flow, the product becomes capital-inefficient and harder to use. V2, still in development, is shifting toward predictive markets, order books, and known payoffs, with the goal of offering a clear question rather than another complicated options interface.
Block Scholes adds the infrastructure view. The report says the firm supports roughly 90% of onchain options volume through venues including Derive. Its argument is that native options exchanges may continue to keep a niche audience through their own UX, but structured products are more likely to be the route through which a larger user base gets exposure to options without even realizing it.
Castle Labs’ conclusion is straightforward: for onchain options to keep growing, they need to stop being sold as options. The next wave of growth is more likely to come from packaging payoffs in forms users already understand.
Projects on the firm’s radar
Castle Labs closes with a short list of projects it is watching.
One is Flex, Yearn’s fixed-rate lending product. The report describes it as a fixed-rate money market where borrowers choose their own fixed rate, and says it is worth tracking on DefiLlama as a new protocol.
Another is Base. Castle Labs says two announcements from Jesse and Brian triggered broad frustration on X. Jesse acknowledged a failed strategy around social and creator tokens and is now handing Base App to Cobie, described in the report as a Crypto Twitter trader and the founder of Echo, which Coinbase acquired for $400 million. At the same time, the report says Brian took no responsibility for a memecoin pump-and-dump tied to his profile picture last week. Castle Labs argues that putting Cobie in charge of Base App is effectively the last chance to repair Base’s standing with crypto-native users.
The third name is Plether. The project is building a perpetuals DEX around the U.S. Dollar Index, or DXY, allowing users to go long or short synthetic dollar exposure onchain. Castle Labs highlights several design choices: positions have a defined maximum payoff at entry, LPs are split into senior and junior tranches, and the protocol blocks new positions if it cannot enforce solvency.
The final item is Starknet’s focus on security. Castle Labs says it published a report the previous day on two barriers to the next phase of institutional onchain growth: privacy and durability against quantum threats. In its view, Starknet’s recent work touches both. Better privacy could let institutions disclose onchain information more safely, while quantum durability raises the question of whether today’s infrastructure can survive the next security cycle.