Etherean to Crypto Moderate, Part 4: Equilibrium

2026 May 27 See all posts
Etherean to Crypto Moderate, Part 4: Equilibrium

2025/early 2026 was a pivotal time for crypto. I was not the only one for whom these core crypto questions bubbled to the surface. The announcement of EVM-based Tempo and Arc in mid-2025 as regulator-compliant, whitelisted validator chains with native privacy features was met with a lot of pushback from people in the Ethereum community. Major Ethereum contributors like Dankrad Feist left to work on Tempo. I think Tempo and Arc jolted more people awake. Projects with serious backing were taking a much more pragmatic approach to solving real world problems and that involved compromising on the cypherpunk dream of a future world where everyone self-custodies assets and carefully manages their onchain privacy to conduct transactions through immutable protocols. Similar to the realization in the 2000s that people would never run their own servers, I think people realized crypto adoption, at least for the foreseeable future, would inevitably involve intermediaries. It's exactly the same phenomenon that Moxie talked about with the early internet. The technologists of the 1990s assumed that everyone would become like them, run their own infra, learn to read code etc. We fell into that same trap again with crypto thinking people would self-host wallets and read smart contracts.

Stablecoins achieved clear success. Total market cap went 100x from ~3B in 2020 to ~300B in 2026. The vast majority of this growth came from centralized stablecoins - coins like USDC/USDT which have admin keys and freezing capabilities. Financial plumbing, especially internationally, is already benefiting from fast, global settlement and I have minimal doubts it will continue to grow. Ironically, USDC was ridiculed initially as centralized by most of the industry, but now for the most part stablecoins are championed as the clear PMF of blockchains and people have forgotten about the earlier stronger ethos.

There was massive churn in the Ethereum Foundation about what the priorities should be. Solana's much clearer focus on UX had started to move meaningful capital away from the incumbent leader. First, the Ethereum community responded by creating teams dedicated to functions more associated with a traditional startup like marketing and business development. I think there was this realization that even virtuous protocols still need outreach to fight for market share. VCs welcomed that new type of thinking. Many crypto teams that were historically deemed the most decentralized or "Ethereum aligned" announced big reductions in workforces, narrowed priorities or were outright acquired by less cypherpunk teams (e.g. Optimism, Celestia, Axelar). Then the Ethereum Foundation pivoted back to cypherpunk priorities and many key EF leaders left. Vitalik's point was "the world will benefit from at least someone doing the cypherpunk stuff and that someone will be us." VCs did not like this.

The industry also started to focus more on privacy, as it became ever more obvious that it was a blocker for a bigger wave of adoption. When it comes to money, privacy has always raised a lot of challenging policy questions. Crypto purists will draw analogies to encrypted messaging or existing physical cash. When strong encryption first became widespread in the 1990s, there was a very similar debate between the technologists and regulators. Technologists argued basically that people have a right to communicate privately, even if that means criminals can communicate and organize via such technology. Furthermore, there was the practical argument that even if you tried to ban it, you can't stop people "multiplying numbers together." The regulators argued that law enforcement needs exceptions in order to do their job and they would be hindered otherwise. End to end encrypted messaging nevertheless prevailed, despite law enforcement never really being on board with it. Today we're basically in a limbo state, where the metadata is often subpoena-able by authorities and they get by doing their jobs with basically that. Money is even more relevant than messages for law enforcement, so it's an even harder debate for the pro-technological side to argue. Tornado Cash and Thorchain are actively being used by the Lazarus group to systematically launder billions of crypto-hack proceeds to North Korea. That is obviously a Bad Thing. The crypto industry's modern answer to all this is "compliant" privacy protocols. Zcash supports selective disclosure of transaction history for taxes/audits but it can't easily prove innocent provenance of funds, since that would require every upstream party in the transaction graph to also disclose. That problem was solved with "proofs of innocence" originally stemming from a Vitalik et al paper in 2023 and led to systems built on that idea like Privacy Pools and Railgun. Private transactions via smart contract pools which allow users to prove that the source of funds was not from a known tainted source, which is a reasonable compromise.

In 2026 especially, AI rose to the absolute forefront of everyone's mind, but especially tech people. There was a big shift in crypto mindshare to "agentic payments" as a clear new use case, which maybe we'd been building towards all along. Most paid API services are designed around a human registering, obtaining an API key then embedding that in software. That works fine for software that has a relatively small number of dependencies, but if the software is an agent working on your behalf you may want it to interact with a large number of services to best accomplish the task. You also may want it to dynamically discover paid endpoints and use them directly without pre-registering. This led to the revival of the ancient HTTP code 402 (payment needed) via Coinbase's x402 system as well as the competing Machine Payments Protocol (MPP) from Stripe/Tempo. While this is exciting, it's noteworthy that MPP also allows non-crypto payments and there are plans for non-crypto compatibility in x402 V2. Letting your agent dynamically discover and use paid services without human approval presents obvious risks and challenges. There are also benefits of the old-school API key approach where I can declare upfront a small number of service dependencies for which I'm actually comfortable letting my agent send money to.

My view today is that crypto started off in a quixotic cypherpunk place, attracted extremely vocal critics, experienced a clear reckoning with the realities of regulation and economics, then reoriented in a more sustainable direction. I'd summarize my stance on the industry now as a "crypto moderate":

Going forward AI will certainly accelerate solutions to the technical drawbacks of crypto today like privacy and scalability. A lot use cases have been tried and several have been proven clearly to not make sense. There are real use cases which are now scaling up, but they certainly came with compromises on the original vision.

Questions or comments? Email feedback@cstein.xyz