Less than one year after Mellanox was acquired by Nvidia for US$7 billion, Alon Webman, co-founder of Mellanox, started his second entrepreneurial project. The new company, Chain Reaction, targets the exponential growth to be set off by he calls a "cryptography everywhere" trend. His presentation on Zoom at the first US-Taiwan Startup Forum (UTSF) held in late September generated much interest from various venture capitalists and experts from the blockchain and semiconductor sectors.
Blockchain's potential is widely expected. PriceWaterHouse predicts that the application of blockchain will boost global gross domestic product (GDP) by US1.76 trillion by 2030. IDC estimates that blockchain spending will grow at five-year (2018-2023) CAGR of 55.3% for the Asia-Pacific region and 57.1% worldwide, with global spending reaching US$14.4 billion by 2023. Despite the optimism, Webman and Oren Yokev, Chain Reaction co-founder and CTO, foresee some pain points ahead. They told Digitimes that combining the silicon/software engineering capability of their team and Taiwan's high-tech manufacturing supply chain can provide the best solution.
Q: Why did you pick blockchain to combine with IC design as your second entrepreneurial project? What is your view toward the blockchain IC industry evolution and requirements?
Webman (W): We came from different industries. I came from the high-tech industry, and Oren came from the government institution. But we were classmates some 28 years ago.
All of the people who joined the company wanted to experience something they could start from the beginning. I brought in the experience of Mellanox as a startup, but this time in a different vertical, to build something we did not have 20 years ago. Many people who joined Chain Reaction could have decided to retire at this stage of their lives, but we did not feel what we learned over the years was enough. We wanted to take it to a new company, and this time to do it at the front seat, not from the backseat. This is the first thing. I think this goes along with Oren, goes along with Richard (Richard Lu, EVP of Sales and Operations).
Richard and Joseph (Joseph Chen, deputy director of Global Supply Management) came from TSMC. And my team has been working with me for many years. We want to work together and enjoy a real new start to bring everything that we've learned in the last 20 years.
We want to do something that works and has businesses around. I brought in the IC capability of our team; Oren brought in software, cyber security, and cryptography. And we started to look around different kind of verticals. All of those verticals didn't have two things that are important at the same time: if you go to cyber, you don't have the IC; if you go to networking, you have less security. We think IC, cryptography, security, software, all combined, are important. Blockchain combines them perfectly. But one thing we missed was someone who could help to make sure it would work on an ASIC. We contacted Richard. We brought in the engineering part. Richard brought in the relationships and the experiences in fab, sales, and overall silicon companies around the world.
The combination of ASIC, cryptography, and algorithm - and to do it with TSMC's high-volume manufacturing - brings the maximum of the knowledge and experience that Oren, Richard and I have accumulated over the last 20 years.
Q: You talked about the trilemma - scalability, privacy/security, and decentralization - of the blockchain industry. Could you explain that and how do your solutions solve those problems?
W: You must have decentralization as it is one of blockchain core elements.
You need to have security and privacy for your data. You work in the public domain, so you must secure all information and data. If you want the technology for millions of users, you must have scalability. As a silicon company you always want to work in a segment that has the potential for scalability to justify the business. The problem is that today you cannot have all these three elements taken care of at the same time. All the solutions available out there today are software based, and they only touch and optimize one or two of those three elements.
We think, by providing efficient cryptography, we can optimize all of them. And we also realize that, if we want to do cryptography the right way, it must be through hardware acceleration. The software running the regular and generic processors available today, whether it's Intel processor or the most efficient graphic processing unit (GPU), are far from being efficient for the specific mathematic computation required for cryptography in order to boost all those three elements of blockchain I mentioned before. Doing decentralization basically means that you have a lot of computers which do the same calculation in parallel. Compared to one centralized computer doing one solution, if you do it with 100 decentralized computers, it takes 100 times of computing power to do the same solution.
To do it efficiently and economically, the first step is to reduce the cost overhead coming with decentralization. Accelerating cryptography reduces computation cost of many computers doing the same calculation in parallel. You can use cryptography to improve performance. Performance is measured by two parameters: the number of transactions and the cost. You can use that cryptography to increase the number of transactions per second and reduce the cost, by taking many transactions and signing in with a simple cryptographic signature. The scale in this case is multiplied by the number of transactions you sign together. And last, anytime you make the transactions more efficient, faster, and requiring less resources, you make them cheaper. So that also covers the cost.
After covering the decentralization, the number of transactions and the cost, we go into security and privacy. They are both done by cryptography, and both require lots of computing. To combine them with decentralization and scalability, you need to ensure efficiency. As I said, cryptography is the solution. We believe the most efficient and effective way is by far through hardware acceleration. The first solution we see it being 50-100 times more efficient, and the next solution we expect to be more than 1,000 times better.
They sound like a big number, but you need to understand that cryptography is very inefficient today. Some operations can take minutes, hours, or even days. A thousand times more efficient will bring you to scale and bring in millions and millions of users in the future.
Q: You've been talking about "cryptography everywhere." Any ballpark figure forecast for us?
W: Tell you a secret: cryptography is already everywhere in our lives today. When you use your mobile phone, when you use your car, your computer, everything has cryptography. Today there are already some solutions in cryptography that do not require much computing. However, in the future, cryptography will step up into the next level with the privacy and security elements that we talked about in blockchain. To keep your data outside your premises secured and useful, you take cryptography and do analytics on the data while they are still crypted. That way, you can keep the data secured and still enjoy the benefit of your data. We see database everywhere in the future, in cloud service providers and in storage for data that will involve more advanced cryptography.
Cryptography is already everywhere, but there will be more. That means you will need stronger computing power. Whether it is centralized or decentralized, cloud providers of services on your computer, mobile phones and cars, all of them require strong computing powers, and we will be able to play in all those verticals.
Blockchain is the first area that have scale for us as a company to generate revenues that sustain efforts in other things. But blockchain will be popularized in 3-4 years. Multi-party computation with stronger cryptography will be the next stage, which is expected to happen in 5-7 years. And then we will see V2V communication, which is between cars and everything around them. You do not want anyone to mess around or hack into them. So, security must be strong but efficient. Let us not forget, in 5-10 years, quantum computers will be here. Quantum computers also take up a lot of computing powers. And this just touches the edge of everything. So, you see, cryptography is already here, and there will be much more. We will be able to supply the tools for everyone to use, and that will make Chain Reaction a successful company.
Q: What is Oren's contribution in this regard?
W: He is the smart guy. He is the one who defines how to do it and put the best solution into product. I am just the guy who tries to sell it.
Yokev (Y): We have a lot of smart guys in the company. It is indeed a huge challenge to address the need of data privacy. For example, the COVID-19 situation calls for a lot of data sharing and quick actions. If there is a solution where you can share data but there is no trust between the agencies invovled, it would hurt people and even cause panic. That may be an extreme case, but we see data privacy as a big driver in the future for every cryptography we are doing.
The challenge is to take those abstract ideas and very complicated mathematics and then turn them into transistors that run faster and at lower costs. Fortunately, we have a team of experts that can do all the process. We have a new processor architecture that is supportive of all these new ideas and mathematics, providing better solutions in the coming years.
W: By the way, our product is called CrPU, Crypto Processor Unit or Chain Reaction Processing Unit. It was Oren's idea. He believes it is the way we should be doing.
Q: What is your internationalization plan in terms of recruiting, supply chain management, and customers?
W: It's not just for Chain Reaction, but also for others looking to become big and influential companies. You start a small design team usually located in one area. Because this is the easiest path: you work with people you know, and you have to work together. But to be successful, you have to be international and global. This process takes a while; it doesn't happen in one day. You have a product, and then you must sell it. We want to be a company that is all around the world. We are a young company that is 1.5 years old, and already have two offices in Israel, one in Taiwan, and one in the US. I am sure in the next few years, you will see us elsewhere in the world. There are no boundaries as long as we find the right people. Our company is based on the ability of people, and everyone brings to the table something different, whether it's engineering, or supply chain.
Taiwan is such a strong technology hub. We are going to manufacture a lot of things in Taiwan, and with Richard's help, we are going to enhance our connections in the US and Taiwan. Certainly, TSMC, the best foundry out there, will be our partner. We will sell it in the US, to Europe, to Middle East, to Far East, and to the emerging markets in South America, etc. There will be no limitations. You will see Chain Reaction products everywhere in the future. We hope to change the world by what we do today.
Y: We understand the supply chain is a huge challenge. That is very unusual for an Israeli company to start with a triangle. Many companies do it with the R&D, then sales, and then manufacturing. But we start it rather the opposite. Because of the vision, we know we would need a very strong supply chain and the ability to reach markets globally. Rather unusual, but we believe it's the right path.
W: This is what we bring from what we had in the past. In Mellanox we often said, "high-volume manufacturing starts an architecture." If you don't understand that you will never be able to manufacture and to make a difference. We have Richard first communicate with the architecture guys about what to do, before we can enable the production of the products we sell. If it were the architecture guys to tell Richard what to do, we would only end up with great technology, but no revenues, no supply chain - just prototypes. If you think as a big company while you are a startup, and you know how the company should be, you know the next step to emphasize along the growth of the company, then you will be able to overcome the hurdles along the way. We think that the high-volume manufacturing, the connection with TSMC, and the dedicated sales team are the make or break of the company.
Q: Do you have a timetable for your products to enter mass production? And what processing technology are you planning to use?
W: For the foundry technology, we already know. We are going to use advanced-node processing technology at TSMC. We are going to start with 7nm and 6nm. By the end of 2022 we will already have a 5nm product. We will enjoy the best of what Taiwan's high-tech industry can bring. TSMC is the No. 1 foundry in every aspect. In the first half of 2021, we will have the first solution on top of the cloud, which is still based on our architecture on software. Our first hardware product will be available in the second half of 2021. From there, we will have a continuous line of products.
Chain Reaction CEO Alon Webman
Chain Reaction CTO Oren Yokev