What happens when the world’s largest financial institution tells a cryptography startup that it cannot move forward without a technology that barely existed two years ago?
That is not a hypothetical. According to Fhenix founder Guy Zyskind, J.P. Morgan approached the company exploring the tokenization of roughly $1.5 trillion in assets under management, and the conversation made one thing clear: in his words, they “cannot do it even in theory without privacy for customers.” The exchange, discussed on a recent X Space, captures a tension that has been building inside institutional blockchain adoption for years. Blockchains are transparent by design. That transparency, long considered a feature, is now the primary barrier to the next phase of growth.
The Transparency Problem in Institutional Finance
Public blockchains expose everything. Every balance, every transaction, every liquidation threshold is visible to anyone willing to look. For retail DeFi, this is an accepted tradeoff. For a bank managing institutional client portfolios, it is a structural problem.
J.P. Morgan put it plainly in its Project EPIC whitepaper, noting that “the lack of mature, on-chain cryptographic privacy solutions, coupled with the absence of consensus on implementing privacy-preserving digital identity, continues to create operational friction in tokenized asset interactions.” The bank’s Kinexys platform has processed over $1.5 trillion in transactions since launch, handling an average of more than $2 billion daily, yet its own research acknowledges that solving for on-chain privacy “could broaden adoption” in ways that current volumes do not yet reflect. That gap between what blockchains can do today and what institutions need them to do is exactly the market Fhenix is building for.
The tokenized asset market, for context, is projected to reach $16 trillion by 2030 according to various industry estimates. Today, tokenized money market funds alone sit at roughly $10 billion in assets, a figure that sounds large until you compare it to the $10 trillion traditional money market fund industry. The infrastructure gap is not technological in the compute or speed sense. It is a privacy gap.
What FHE Actually Does, and Why It Is Different
Most people who encounter Fully Homomorphic Encryption for the first time react with some version of the same confusion: if data is encrypted, how can you run calculations on it?
The analogy that helps most people is a lockbox with transparent gloves built into the sides. You can reach inside, move things around, and even rearrange the contents, but you can never see what you are touching, and you can never take anything out. FHE achieves this mathematically. It allows a system to perform arithmetic and logic operations on encrypted numbers without ever decrypting them. The result comes out still encrypted, and only the authorized party can open it. Think of a bank running a credit risk model on a client’s encrypted portfolio: the model produces a score, the client receives that score, and no one at the bank ever saw the underlying holdings.
This is fundamentally different from the two privacy approaches most common in blockchain today. Zero-Knowledge proofs (ZK) allow you to prove a statement is true without revealing the underlying data, but they do not allow computation on that data in a general-purpose way. Trusted Execution Environments (TEEs) rely on secure hardware, meaning privacy collapses if the hardware is compromised. FHE is pure mathematics with no trusted hardware assumption and no limitation on what computations you can run. As Zyskind put it on the X Space, “Privacy is, I believe, the most difficult problem to solve in blockchains. Adding privacy on top of ZK, something like what Fhenix is doing, trying to build and scale Fully Homomorphic Encryption? Those are very, very hard problems to solve. Very few people can do that.”
What Fhenix Has Built
Fhenix started as an FHE-powered Layer 2 blockchain and has since shifted its architecture toward a coprocessor model. The product is now called CoFHE, a stateless off-chain engine that handles the heavy cryptographic lifting. Smart contracts on any EVM-compatible chain, including Ethereum, Arbitrum, and Base, can call CoFHE with a single Solidity import. The coprocessor processes the encrypted computation, and only authorized parties receive the decrypted result.
The performance numbers matter here because FHE’s historical weakness was speed. Earlier systems were too slow for anything resembling real-time finance. Fhenix claims CoFHE delivers throughput improvements of up to 5,000 times over earlier FHE systems, and its separate Threshold FHE Decryption research, accepted to the ACM CCS 2025 conference, recorded 20,000 times higher throughput and 37 times lower latency than prior benchmarks. That paper was accepted alongside research from Microsoft, Google, Meta, Stanford, and MIT. The company has also received a strategic investment from BIPROGY, one of Japan’s largest IT service providers, specifically to expand into the Japanese financial sector.
The practical demonstration of this came through an experiment called Fhenix402, a private version of Base’s x402 micropayment protocol that the team built in a single day. On Base Sepolia, a payment of $0.10 and a payment of $4.02 produced identical encrypted representations on-chain. An observer watching the blockchain could not tell the difference between the two. “You can’t tell which is which, and that’s exactly the point,” Zyskind said.
The Institutional Signal Is Already There
The J.P. Morgan conversation is the most striking data point, but it is not the only one. The bank’s Kinexys Privacy POC explicitly explores on-chain privacy and composability, framing enhanced privacy as “crucial for improving access to digital assets.” J.P. Morgan Asset Management recently launched its first tokenized money market fund, seeded with $100 million of its own capital on the Ethereum blockchain, with Kinexys handling the infrastructure. The fund is open only to qualified investors, but the architecture it represents, public blockchain plus institutional compliance requirements, is exactly the architecture that cannot scale without privacy.
Zyskind also drew a connection that is worth taking seriously: FHE uses the same underlying mathematics as post-quantum cryptography. As quantum computing becomes a credible threat to existing encryption standards, systems built on FHE have a structural advantage that extends well beyond current DeFi applications. This is the kind of long-cycle institutional consideration that banks actually think about, and it suggests the interest from players like J.P. Morgan is not opportunistic but foundational.
“We’re at a true inflection point,” Zyskind said at the launch of Fhenix402. “Circle, Stripe, and global enterprises are moving into blockchain payments. Privacy isn’t optional anymore. It’s the requirement that will make open payments viable.”
My Read on This
The framing Fhenix uses, that it is building the “backbone for confidential DeFi,” is ambitious. But the underlying case is not speculative. J.P. Morgan’s own research documents the problem. The tokenization market’s growth trajectory is measurable. The performance benchmarks Fhenix is publishing are peer-reviewed, not marketing copy.
What I would watch carefully is the distance between demonstrated capability and production deployment. CoFHE is live on testnets and deployed on Base and Arbitrum Sepolia. Mainnet is still ahead. The gap between a cryptographic paper accepted at ACM CCS and a system that processes institutional-grade tokenized assets at scale involves engineering, regulatory, and integration layers that take time. The institutional interest is real, but institutional adoption timelines are slow by design.
FHE also remains one of the more computationally expensive cryptographic operations available, even with Fhenix’s improvements. The 5,000-times throughput gain is measured against prior FHE systems, not against unencrypted computation. That distinction matters for anyone benchmarking against what traditional financial infrastructure can do today.
Still, the trajectory is credible. The problem is documented by the institutions themselves. The technology is advancing faster than most cryptographers expected even three years ago. And if blockchains are genuinely going to hold trillions in institutional assets, the question is not whether on-chain privacy infrastructure gets built. It is who builds it.
Don’t forget to like and share the story!
