Superfluid Founder: Why I Decided to Leave the Crypto Space

Author: Francesco Renzi, Co-founder of Superfluid

Compiled by: Gu Yu, ChainCatcher

Since 2017, I have been working in the crypto space. In 2019, we founded Superfluid with a grand vision: to change the way money works.

Perhaps it was youthful and naive. It was indeed a bit too early.

I still believe that cash flow is a logical evolution of recurring payments, but I am now more acutely aware of how difficult it is to overcome deeply ingrained behaviors, especially regarding such fundamental matters.

Earlier this year, I decided to step down from my position at Superfluid. Miao has taken over and is leading the team to launch an exciting new product, Superfluid Wallet, a new wallet dedicated to creating the best UX for coordinating your agent's interactions on-chain.

I am incredibly proud of what we have built and remain deeply invested in Superfluid's success. But my own attention is turning elsewhere.

What Has Changed

Superfluid's cash flow introduced a radically new UX to an extremely entrenched system.

This is a transformation that requires decades, a lot of coalition building, and possibly top-down pressure, rather than the crypto grassroots counterculture.

Although we saw early signs of adoption as early as 2020, the market simply wasn't mature enough, and unresolved issues around accounting, taxation, and security consumed much of the goodwill we initially built up.

From 2017 to 2021, I was very optimistic about crypto creating fundamental social change. This was one of the reasons I was fully invested. I believed that technological innovations in trust, financial access, and transparency would change the relationship most people have with money, investments, and broader institutions.

From that idealistic perspective, Superfluid winning didn't seem too crazy!

I do believe that crypto has made astonishing progress in creating alternatives to existing financial institutions, and I am very impressed by the advancements in this area. I am also extremely amazed by how some new behaviors have become ubiquitous, mainly in the areas of new asset issuance or new market creation (e.g., AMMs/prediction markets).

But if I pose a narrower, more sobering question: "In what areas does crypto continue to prove itself sufficiently better to replace an existing mainstream behavior?" the list remains disturbingly short.

The clearest cases are not particularly glamorous: ransom payments, evading sanctions, and perhaps gambling deposits and payouts. Stablecoins are a more constructive success, but they largely reinforce access to the dollar rather than create an alternative to it.

I once thought crypto would be a tailwind for a truly multipolar world. It helped people escape certain parts of the traditional financial system, reduced certain forms of government control, and at least in the early imagination of Bitcoin, provided a way out of dependence on the dollar. Contrary to my hopes and predictions, institutional capture, performative preferences, and broader geopolitical trends have intertwined to re-centralize crypto around the U.S. and the dollar.

Greater Social Change

Meanwhile, a massive shift is happening elsewhere.

Creating software has been one of the most lucrative skill sets over the past 30 years, and now it has almost been automated in about a year. I chat with my AI agent more than I do with friends and family. My AI Assistant makes more calls on my behalf than I do myself.

The social consequences are clear, immediate, and highly disruptive.

They push me toward a series of broader questions:

Who can see our data?

Who can change the software that acts on our behalf?

Who holds its keys?

Can we leave a system without giving up our history, relationships, and normal functioning?

As software becomes more powerful, these questions become more urgent. Since January, I have been an enthusiastic user of OpenClaw and Hermes, and it is evident to me that the future will involve significantly more digital experiences for all of us.

But when software can read our messages, coordinate our work, and act on our behalf, a new class of custodial and control issues arises.

In crypto, we have learned to question whether we truly control our assets. In an agentic world, we need to question whether we truly control the systems that act in our name.

Just as crypto has created some optionality for leaving the financial system, creating alternatives to centralized AI is also extremely important.

Whether intelligence is "managed" by centralized providers or is widely available and decentralized (or at least multi-centered!) will become one of the decisive battles of the next five years.

Any single winner would be a catastrophic outcome for human agency. I believe this needs to be discussed and understood more broadly.

A Pragmatic Path to Private Infrastructure

Recently, I have been delving deeper into Trusted Execution Environments and Confidential Virtual Machines: TEEs and CVMs.

These technologies allow software to run in hardware-protected environments designed to prevent even the machine's operators from freely inspecting its memory and data. Through remote attestation, these environments can also provide evidence of what software is actually running.

They are not magic. Attestation does not prove that an application is secure, and confidential hardware cannot prevent authorized software from leaking information through allowed channels.

But I increasingly believe that TEEs and CVMs are the most pragmatic short-term answers to many trust issues we will soon encounter.

More importantly, they offer a hopeful way to change people's expectations of infrastructure. Today, we generally accept that whoever operates the cloud has final access to what runs within it. This should no longer feel inevitable.

Similarly, we currently assume that as software deployers, we can always inspect, fix, or update the software we deploy. CVMs challenge this assumption. Like smart contracts, they can be deployed as immutable and opaque systems: once running, even the deployer may not be able to inspect or modify them. The huge difference is that they run on a complete Linux computer.

The new expectation should be infrastructure without permanent plaintext access: in such systems, privacy relies less on the operator's promises and more on boundaries that operators cannot arbitrarily cross.

Building an immune system against AI centralization requires stronger data sovereignty. For businesses and individuals, data defines and distinguishes us. Keeping data away from data predators is becoming a fundamental form of self-defense.

This is the direction I want to invest in now: providing private, verifiable infrastructure for software that needs access to the most sensitive parts of a person's life or business.

I do not think my time in crypto was a detour. On the contrary. Crypto taught me to focus on trust boundaries, adversarial systems, and the roots of control and privacy. Its hardest lesson is that compromises on fundamental values can accumulate in ways that are not immediately visible. What seems like a small compromise today can grow and spread like cancer.

AI needs what crypto once had: more sovereign, more private systems, and crucially, systems that remain useful even if their operators are not (or no longer) benevolent.

These lessons remain core; I am simply shifting from one expression of this work to another.

The question is no longer just who owns an asset, but who owns and controls the computer that thinks, remembers, and acts around us (or on your behalf!).

That is where I am currently investing my time, and I am excited to soon announce my first product in this field.