Homomorphic encryption is a complex technology that uses cryptographic algorithms to keep data secure within a network and to third parties, but it has low mass-market scalability and therefore low adoption. Especially since right now, the complexity that makes encryption so effective compromises the security of your data. It is slow and difficult to use widely.
But in a world rife with data breaches and creative and malicious hacks, this approach holds a lot of promise in ensuring long-term data security, so investors are slowly turning the concept into reality. continues to fund startups with the smartest talent.
In the latest development, a Parisian startup called Zama has raised $73 million in a Series A co-led by Multicoin Capital and Protocol Labs, approaching a $400 million valuation. Of note, among the long list of other investors in this equity round is his Metaplanet, the Estonian deep tech investor who cut the first check in DeepMind (among hundreds of other investments). This means that it is included.
The plan is to continue investing in research and development and hire more engineers (the current team is 75 people) to build around the two market opportunities Zama sees in its initial version. ) is planned.
There are solutions for dealing with blockchain transactions and data exchange for the training and use of artificial intelligence. He has also built and published four libraries on GitHub to perform that work, and claims that 3,000 developers are using them.
While there are a number of deep technology efforts underway to improve the way HMEs are used around the world, including in Zama itself, the startup is also working as a business.
“We began commercializing Zama six months ago and signed a contract valued at over $50 million,” co-founder and CEO Rand Hindi said in an interview. Ta. Hindi firmly believes that the big business in the long term is in machine learning, but so far its customers have mainly come from the blockchain camp, and not all of these operate in fiat currencies. Therefore, the amount of $50 million is an approximate value.
“If they have a token, we will claim the token,” he said. “For banks using private blockchains, we charge per transaction.”
Prior to this, Zama raised $8 million in pre-seed and seed rounds, bringing the total raised to $81 million. Sources said the latest round of funding would put the company at a valuation of between $300 million and $400 million, but CEO and co-founder Rand Hindi declined to disclose the amount. refused.
There are several reasons why the company is gaining attention, especially in its current funding climate, where these numbers seem large for a technology that has yet to enter the mainstream market.
The first is a simple market opportunity.
“FHE is the most important fundamental cryptographic primitive for computing in the next decade. Zama's technology is the key to building multiplayer, privacy-preserving applications,” said Multicoin Capital's Money Kyle Samani, partner at Ging, said in a statement. “Zama's groundbreaking work with open source FHE tools is just the beginning. We are proud to help build the next generation of encryption-enabled, privacy-first applications.”
Secondly, it's probably due to the founding team.
With a background in computer science and a PhD in bioinformatics, Hindi is a polymath with an interest not only in AI but also in privacy and how it is protected in modern society. One of his previous startups was his AI voice platform called Snips, which was acquired by Sonos.
His co-founder and CTO, Pascal Paillier, is an expert in cryptography whose patents (he says he has about 25 patent families to his credit) currently apply to smart cards and Used in other applications.
Back in 2016, the two began working on the initial technology that would become Zama. According to Hindi, the breakthrough was in 2019 when they arrived at an algorithm that speeds up calculations by 100 times.
“This allows us to turn this into a business,” Hindi said.
This does not yet represent a useful speed for most of the world's transactions, but given that blockchain transactions themselves are generally slow-moving, this represents an opportunity to provide cryptocurrency developers with a Zama solution. I did. In Hindi, he said that whether you are skeptical of cryptocurrencies or not, it is undeniable given the payroll and other types of financial transactions that are being created. This gives you the opportunity to build even more. ”
As we have previously explained, fully homomorphic encryption is something of a holy grail in the world of security and cryptography. One reason for this is that implementing fully homomorphic encryption is too complex to perform in a realistic time frame.
Some of this problem may be solved over time by the development of compute-optimized chips, which are being developed by both startups and semiconductor giants such as Intel.
Meanwhile, companies like Zama continue to research algorithms and techniques to compress the work required to perform homomorphic encryption on existing infrastructure. Its libraries and previous work include a fully homomorphic encryption library for bringing FHE to machine learning. A compiler that helps convert Python programs to their FHE equivalents. A library that allows entities to interact with the Ethereum Virtual Machine using homomorphic encryption.
There are many other startups in this space, such as Ravel, Duality and Enveil, but the market is very small for now, and they are still trying to prove themselves, Hindi said. And I added that the real goal is to continue to grow. market.
“We're pretty much friends with each other,” he said. “The goal is not to fight, but to build a market. A joint petition. We meet at conferences and talk, and someday we will compete, but not today.”