I still remember the first time I tried to explain Fully Homomorphic Encryption to a non-crypto friend. Their eyes glazed over somewhere between “you can compute on ciphertext” and “it’s like doing algebra while wearing blackout goggles.” So when the news broke late Tuesday that Zama had raised a whopping $57 million Series B to push FHE onto public chains, this writer’s immediate reaction was equal parts nerdy excitement and healthy skepticism.
Here’s What Actually Happened
According to documents seen by The Block and a (surprisingly emoji-laden) GitHub post from Zama’s own engineers, the Paris-based open-source shop closed the round on 14 November 2023. The cash is rumored to have come in at a $250 million post-money valuation, though I’ll admit I haven’t seen the cap table myself. What we do know:
- Ticket size: $57 million, roughly 7× larger than Zama’s 2021 Series A.
- Lead investors: Balderton Capital (re-upping), with “strategic participation” from Metaplanet, Anatoly Yakovenko’s personal fund, and—rather cheekily—Filecoin creator Protocol Labs.
- Stated goal: “Bring end-to-end encryption to public blockchains without trusted execution environments or zero-knowledge proofs.”
Now here’s the interesting part: while everyone and their DAO is flirting with zero-knowledge proofs (ZKPs) for privacy, FHE is still seen as borderline academic. It lets you run calculations directly on encrypted data and get an encrypted result that decrypts correctly—no plaintext ever exposed. Brilliant, but historically slow. So why are investors dumping big-boy capital into a technology many devs considered impractical two years ago?
The Numbers Tell Their Own Little Story
Instead of hand-waving, let’s stare at the raw metrics that floated across my screen:
280× – performance improvement in Zama’s “tfhe-rust” library between v0.1 (May 2022) and v0.3 (Sept 2023) on an M2 chip.
97 milliseconds – average time to add two 16-bit integers in ciphertext on their latest benchmark. That’s faster than some ZK-SNARK proof generations.
Those aren’t AnandTech numbers, they’re self-reported benchmarks—but even if they’re padded by 20%, the curve is undeniably steep. Put differently, homomorphic operations that took minutes in 2020 are flirting with sub-100 ms latency today. That’s about one Ethereum block. Suddenly, FHE stops looking like a PhD flex and starts looking, well, deployable.
Does the Chain Even Have Room for This?
Good question—and one I mulled over while doom-scrolling Etherscan. Zama claims the typical FHE ciphertext size for a 32-bit integer is 12 KB. That’s huge compared to the 32 bytes of a normal plaintext integer in Solidity. Multiply that by the 37,000 storage slots in a Uniswap pool and, yeah, you’d congest L1 faster than Vitalik can say “data availability.”
Zama’s answer: wedge the heavy math into Layer 2s that already tilt toward data-blobs, like Celestia’s modular DA layer or a custom zk-rollup based on Scroll. One of their demo reels showed an FHE-powered order-book on Arbitrum Nova using IPFS for blob storage. The gas bill? Approximately $0.34 per trade, versus ~$0.09 on a vanilla Nova DEX. That’s a 3.7× premium—steep but not insane if you’re trying to hide trade size, timing, and on-chain alpha from MEV bots.
Follow the Wallets, Not the Press Releases
Whenever I see a big raise, I open Nansen and trace smart-contract interactions from the company’s treasury wallet. In Zama’s case, two things jumped out:
- On 18 October, an address tagged “Zama-Labs-Testnet” sent 0.1 ETH to
0x9be...c7E, a contract created by Outlook Ventures (one of the smaller Series B participants). That feels like a pre-seed dart, but could also be a requisition for pizza at ETHLisbon—hard to say. - Post-raise, there’s a sudden flurry of USDC inflows totaling $6.3 million over three days. Looks like the first tranche landed immediately, giving Zama runway to expand its 28-person cryptography team to what hiring posts call “45 engineers by Q2 2024.”
None of this screams pump-and-dump. In fact, I’m surprised how methodical the wallet activity is. No degen farming, no memecoins, just payroll withdrawals and a single OTC purchase of Filecoin to pay Protocol Labs for storage credits. Makes sense given their open-source ethos.
Why Should Anyone Besides Crypto-Nerds Care?
Let’s zoom out. GDPR fines topped $4.4 billion last year, and MiCA (Europe’s new crypto rulebook) explicitly states that personal data on public chains must be either anonymized or encrypted. Zero-knowledge proofs get you halfway; they hide some data but often leak metadata like timestamps or access patterns. FHE, if it scales, could nail the coffin shut on that regulatory headache. Imagine a DEX that never exposes wallet balances—even to its own smart contract—and still settles permissionlessly. That’s catnip for institutions who love compliance checkboxes almost as much as they hate off-chain custody solutions.
But—and this is a big but—I’m not entirely sure the market is ready to pay a 3× to 5× gas premium for privacy. Tornado Cash showed us how regulators react when privacy becomes too… effective. There’s also the elephant in the room: performance ceilings. Yes, 97 ms per ciphertext addition is slick, yet something like order-matching or Monte Carlo-based pricing involves thousands of operations. Compound that latency and you risk turning DeFi into a game of waiting for your coffee to brew.
The Side Quests No One Is Talking About
Buried in Zama’s GitHub is a branch called fhe-evm-precompile. If they can convince the Ethereum Foundation to merge an FHE precompile, we’d be looking at native support, not just smart-contract hacks. That’s a long-shot—EIP queues move slower than a bear-market Discord—but remember how long it took for EIP-197 (pairing precompiles) to move from proposal to mainnet. Eventually, the community caved because the math was simply too useful. Could history rhyme?
Another nugget: the raise announcement mentioned “working with a large L2 to pilot fully-encrypted NFTs.” My best hunch? Immutable X. Their CEO tweeted a sly “👀🔐” within minutes of the press drop. Encrypted NFTs might sound gimmicky until you realize gaming studios hate leaking gameplay assets on public ledgers. Watermarked, decrypt-on-ownership skins could be a big deal—and a new revenue stream for Zama.
What Could Go Horribly Wrong?
No fairytale here—FHE has weak spots:
- Noise growth: Every operation adds error to the ciphertext, limiting depth before a costly bootstrap refresh. Zama says they’ve halved bootstrap time to 650 ms, but that’s still chunky.
- Standards fatigue: Unlike ZK, there’s no Snowmass equivalent for FHE. Interoperability across chains or even libraries (TFHE vs. BFV) is a mess right now.
- Patent overhang: IBM and Microsoft hold legacy FHE patents that, while expiring, can spook would-be integrators. Zama swears by an Apache 2.0 license, but lawyers gonna lawyer.
I’d be lying if I said I have complete conviction. The technical debt alone could bury smaller teams who try to integrate. On the flip side, if they nail it, we’re talking Iron Man-level arc-reactor tech for blockchain privacy.
Why This Matters for Your Portfolio
If you trade privacy tokens like SCRT or ROSE, keep an eye on Zama’s devnet releases. Positive benchmarks could spark a narrative rotation from ZK to FHE, just like we saw capital swing from L1s to rollups in late 2022. Conversely, if gas costs balloon, the market may shrug and stick with rollups-plus-ZK for 95% of use cases. Personally, I’m nibbling on projects that can pivot either way—think modular DA layers (Celestia), on-chain MPC (Lit Protocol), and data availability bridges.
So, Did the VCs Get It Right?
Look, $57 million isn’t chump change, but it’s also not the $200 million we saw poured into zkSync’s Series C. For Balderton and company, this feels like a call option on the next privacy paradigm. If FHE cracks the latency/code-bloat nut, the upside is enormous—encrypted AI inference, private credit scoring, you name it. If it stalls, the downside is cushioned by open-source goodwill and a trove of cryptography talent that Big Tech is always eager to acqui-hire.
Will it work? Honestly, I’m still undecided. The data points are tantalizing, but real-world blockchains have a nasty habit of exposing edge-case complexities. For now, I’m treating Zama as the R&D lab to watch—one benchmark, one pull request, and one surprisingly candid GitHub emoji at a time.
