paint-brush
Toufi Saliba: The Architect Behind HyperCycle's Decentralized Network of AIby@ishanpandey
402 reads
402 reads

Toufi Saliba: The Architect Behind HyperCycle's Decentralized Network of AI

by Ishan PandeyFebruary 15th, 2024
Read on Terminal Reader
Read this story w/o Javascript

Too Long; Didn't Read

HyperCycle, led by CEO Toufi Saliba, is pioneering a decentralized AI network aiming to democratize and enhance global AI collaboration. With significant achievements like selling over 300,000 software licenses and reaching unicorn status, HyperCycle is poised to transform the AI landscape by facilitating secure and efficient machine-to-machine communication.
featured image - Toufi Saliba: The Architect Behind HyperCycle's Decentralized Network of AI
Ishan Pandey HackerNoon profile picture

This interview peels back the layers of HyperCycle's ambitious mission to weave a decentralized web of artificial intelligence (AI) that promises to revolutionize our interaction with technology and each other. In the digital era where innovation paves the path for the future, HyperCycle, under the visionary leadership of CEO Toufi Saliba, stands at the forefront of a technological renaissance in AI and decentralization.

Navigating the AI Revolution: An Insightful Journey with HyperCycle's Toufi Saliba


Ishan Pandey: Hi Toufi, great to have you here for our "Behind the Startup" series. Please tell us about your background, and your journey as CEO of HyperCycle?


Toufi Saliba: Certainly, I’m the co-author of the Toda/IP protocol and I currently serve as the Global Chair for International Protocols for AI Security for the IEEE, the world's largest technical professional organization dedicated to advancing technology for the benefit of all humanity. I have a history of building various AI projects, either cryptography or cybersecurity-centric.


Naturally, I have experienced both failures and successes in my career, starting with my first AI exit occurring in my early 20s in 2002. Then several successes continued a couple of decades later in October 2022 when I  took on the leadership of Hypercycle.ai and helped grow  500 software licenses sold to over 300,000 in one year. We also saw the valuation growth of over 60X reaching a unicorn status in less than a year from inception, and yet we still expect bigger growth to come.

Today I continue to work across many states around the world towards creating a secure Internet of AI and benevolent AGI for the benefit of all of humanity.

Ishan Pandey: Can you provide an overview of HyperCycle's mission and the specific problem it aims to address in the field of AI?


Toufi Saliba: Our mission is to enable the Internet of AI using HyperCycle’s decentralized distributed network. Although it may sound simple on paper it does involve moving a lot of tiny pieces together in order to make this happen. It is crucial that there is no single entity controlling AGI as it would be very detrimental for us homo-sapiens. Think of us as the company building the global AI brain in a pure peer-to-peer (p2p) design.


Ishan Pandey: How do you see the current landscape of AI collaboration, and what inspired HyperCycle to focus on addressing the limitations in the cooperation of AI machines?

Toufi Saliba: We believe that the internet of AI will outsmart and outperform any single AI. Currently, there is little to no collaboration between AI systems. Some might say GPTs are somehow AI-to-AI communication but this is not really the case as it only mirrors how current Web2 apps communicate with each other in a centralized network.


They share data but they don’t learn from the data and certainly do not share the knowledge between each other. In reality, the current AI landscape is completely siloed and AI is not truly getting “smarter” but only more powerful. What good is a super powerful AI if it fails at reasoning properly and if it is controlled by a single big corporation? That’s not beneficial to humanity, that is simply dangerous. If you study the anatomy of the brain you realize how beautiful, yet, how complex and important it is for homo-sapiens or any living species for that matter.


This is why HyperCycle is leveraging very efficient technologies like TODA and Earth64, which I co-authored, to bring this global brain to reality. In essence, we are solving this with HyperCycle, allowing efficient communication between all network AI agents and execution of the communication in sub-second finality, enabling machines to grow in intelligence and become more profitable in the process. So far we’ve seen tremendous traction, with a lot of actors willing to partake in the construction of this network worldwide and B2B clients asking to migrate part of their AI computing on HyperCycle.


Ishan Pandey: In your view, what makes HyperCycle a rare breakthrough in Computer Science, and how does it contribute to the advancement of AI?

Toufi Saliba: If you’ve followed what was previously said you understand that there is no other company doing what we do which is providing a General Purpose Technology securing an Internet of AIs. By 2026, we believe that 90% of machine-to-machine communications will be involved in all transactions.


In addition, HyperCycle will be a solution that makes efficient use of these communications by ensuring each machine on our network becomes smarter. There’s hardly anything that will prevent AI from being continuously developed, therefore we might as well ensure that it will properly learn and grow as a decentralized AI to benefit not only one or two major companies but humanity as a whole.

Ishan Pandey: Could you highlight some of HyperCycle's notable achievements, both in terms of business/revenue and technological advancements?

Toufi Saliba: Certainly. To this day we’ve sold more than 300,000 software licenses, which gives users lifetime access to run their AI machines on our network. Each single license has to be paired with 1024 HyPC which corresponds to a current demand for more than 300,000,000 HyPC tokens.  We’ve already fulfilled part of this demand by liberating 100,000,000 HyPC on our network, currently paired with licenses and activated within real hardware machines that are building their Uptime using an immutable monitoring process called “Tilling”.


On the hardware side, we’ve secured many B2B clients that will provide AI computation for all network consumers. These clients are facilities and hardware providers in Paraguay, Finland, and many other places around the globe. On the B2C side, we’ve already sold out all our HyperAiBox via a satellite company called HyperAppliance. Those machines are plug-and-play hardware devices purposed to run AI computing on the HyperCycle network and it’s our way to deliver AI sovereignty directly to the hands of the people.


These are very cool devices that you only have to plug it in once and then, it can even cover your entire electricity bill, all because the network is using the HyperAiBox's hardware capabilities to resolve tasks requested by AI consumers. Based on the amazing feedback we’ve received we are now pushing a newer version of the HyperAiBox with partners in Korea and we expect to sell many this year.


On the technological side, our network has been operational since November last year. We're in the process of deploying the Computation on our network which will allow any network participant to take part in AI-related tasks on HyperCycle and earn a share of the revenues generated there.


Lastly, and because HyperCycle is a software company, we are working hand in hand with HyperPG to find clients already spending more than 100K$/month on AI Compute and making them save money by migrating part of (or the totality of) their activity onto our network to contemplate how beneficial it is for them to.

Ishan Pandey: How does HyperCycle's business model provide low-cost, high-efficiency computing power for clients and opportunities for hardware owners?

Toufi Saliba: Our secret lies in the efficiency of our design and this is the result of a decade of research in AI, cryptography, and security. Our clients are very aware of how costly AI computation currently is and how inevitable AI usage is becoming in order to stay competitive in the market. HyperCycle allows any machine to operate efficiently in collaboration with other AI machines on the HyperCycle network and that is what allows us to offer very competitive pricing to our clients.


We know we wouldn’t be successful on ideas alone, so decentralized AI needs to be cheaper than the current siloed AI solutions one can find on the market, otherwise progress will slow down. HyperCycle is providing a smarter, cheaper, and decentralized AI computing service for everyone so this recipe is a no-brainer.

Ishan Pandey: What are the near-term plans for HyperCycle, especially regarding the conversion of fiat to crypto for AI computations and the efficiency of deploying the platform for B2B clients?

Toufi Saliba: We’re not exposed to this issue since our requesters can propose payments in fiat. HyperCycle operates as a B2B-oriented software entity catering to AI compute demand. To service this demand, we are supported by both global B2B and B2C contributors on the supply side.


This includes energy or hardware providers, license owners, HyPC token suppliers, AI developers, all of which part a large role in helping to expand and secure our decentralized network. The near-term plan is to provide irrefutable proof of concepts to our current clients when Compute is deployed on our network. It’s very exciting, so far at every stage of our development we have been under-promising and over-delivering and nothing is stopping us from continuing to do so.


On the hardware side, we have gigantic ambitions. As I mentioned earlier, we managed to sell out our first batch of HyperAi Boxes within a couple of days and we are preparing a second batch already. Ultimately we hope to deploy seven figures of HyperAi Boxes around the world.


Don’t forget to like and share the story!


Vested Interest Disclosure: This author is an independent contributor publishing via our brand-as-author program. Be it through direct compensation, media partnerships, or networking, the author has a vested interest in the company/ies mentioned in this story. HackerNoon has reviewed the report for quality, but the claims herein belong to the author. #DYOR