Leemon Baird published his work on Hashgraph’s consensus for the first time in 2016, positioning it as an alternative to traditional blockchain architectures. With computer training and a career covering both the academic world and the industry, Baird co -founded Hedera to market technology.
Its academic trajectory is remarkable for its first work in neural networks and learning to strengthen in the 1990s, a period when IAT research was sailing what would later be called “AI winter”. Since then, the Hedera project has evolved in a landscape crowded with competing approaches to the great distributed book, each claiming technical superiority and targeting different segments of the market.
Baird, speaker of the 2025 consensus, transitions easily between technical explanations and commercial strategy, reflecting the double challenges of the construction of a new technology and a viable ecosystem that surrounds him.
This interview was condensed and slightly modified for more clarity.
Coindesk: Your hash algorithm emerged in 2016, during a period when many alternative consensus mechanisms were proposed. What technical limits of previous approaches did you specifically try to resolve?
Baird: I love computer science and the mathematical side of it – Inventing things and solving problems. When I became an entrepreneur 25 years ago, it was the same process. The heart of what I always do is trying to understand the fundamental problem we are trying to solve. What is the real question? What are we really trying to accomplish? And then you build on this and solve this problem. In the blockchain, the fundamental question I asked was: Bitcoin is cool, but it is slow and not as sure as possible with ABFT [Asynchronous Byzantine Fault Tolerance]. It burns a lot of energy and is not as flexible as we want. I was wondering if, on the much lower layer of the consensus itself, there could be a way to avoid burning a lot of energy while being fast and secure. Is it possible to reach ABFT – the strongest type of security – while also being super fast and not burning electricity or not pouring carbon into the atmosphere?
I started working on this in 2012 as one of the many mathematical problems I explored. At the start, I was convinced that it could not be done. I would take the problem, play with and convince myself that it was impossible – again and again. But in 2015, I realized that by throwing two atmosphere, suddenly, everything is being set up. You can have an ultimate speed – essentially at Internet speed – while also having ultimate safety with ABFT. And it’s proof of stake, so you don’t waste electricity.
From the point of view of companies, the question was: what is the right way to govern this? When we look at the blockchains, they often say: “We will not have governance. Anyone can help do it.” But power over time can be consolidated – you end up with a handful of developers or people behind the scenes that control everything.
With Hedera, we started differently. We had governance decentralize from the start. We have brought some of the world’s largest organizations – universities and businesses are spreading on a global scale that people trust and who have a reputation to protect. They balance, creating checks and counterweights, and together they govern the system.
It was a question of answering the fundamental question: what do you want in governance? What will give you real without confidence, or at least a lower confidence bar necessary to fully trust the system? It was our answer: the approach of commercial questions with the same rigor as mathematical questions.
Coindesk: You mentioned the RWA tokenization, carbon credits and stablescoins as key use cases. These are areas where almost all the big blockchain are concentrated. What specific implementations on Hedera have demonstrated significant transaction volumes or the adoption of users?
Baird: I would highlight four key areas:
First of all, AI is extremely exciting at the moment. The dangers of AI are also worrying, which is why we must establish the origin, governance and control of versions for AIS. People need to know if they can trust what’s going on. Hedera helps with AI in several ways, in particular data authorization and potentially manage fees for people who provide training data. The work that Eqty Lab does with Nvidia and Intel on Hedera is particularly exciting.
Second, the tokenization of active active world transforms the way we manage precious assets. We have many projects that tokenize real estate, gold, diamonds, carbon credits and even carbon emissions on Hedera. From the start of blockchain technology, I argued that what is important is not photos of monkeys or games – it is that all things of value on the planet will ultimately be placed in these registers.
Third, stablecoins are essential if you want the adoption of the real world. We have created a stable room studio to facilitate the development of Stablecoin for Hedera. The Hedera Council includes many financial institutions that do impressive work with Stablecoins.
Fourth, immutable data records are almost unique at Hedera via our Hedera consensus service. This allows you to send messages to subjects with access controls and an immutable recording. Companies like Hyundai and Kia use it for monitoring emissions throughout their supply chain.
Coindesk: UCL research on energy consumption has compared various networks, but the methodology is considerably important in these comparisons. Assistance proof systems generally have similar profiles, with differences resulting from the number of nodes and material requirements. Hede’s approach differs fundamentally from other POS networks, or is the efficiency mainly from the configuration of the current network?
Baird: We thought about energy consumption from the start. Everything – of our algorithms to the way in which the nodes are managed and governed, and the fact that we use proof of participation instead of the proof of work – has filed the basics of low shows of the first day.
It created a virtuous cycle. The first adopters who seek to tokensiner carbon credits chose the green blockchain. Then the people who wanted tokenize shows and credits wanted to use the same blockchain where everyone was doing a similar job. This steering wheel effect has made Hedera the most popular blockchain in green technological space.
According to University College London, Hedera has the weakest carbon emissions by transaction of any blockchain. We also buy carbon credits to be negative in carbon. Being green was cooked in our structure from the start, which is why we have become a leader in this area.
Coindesk: We see a lot of projects that try to combine AI and blockchain technologies. Given the different calculation paradigms in which these systems work, what realistic integration points do you see beyond marketing stories?
Baird: The intersection of AI and blockchain is more important than most people think so. On Hedera, we see a real traction in several areas:
Providence and governance are essential. While we enter a world where everything will be generated by AI, we must know that we can trust AI. This requires digital signatures to check the origins – whether human or created by AI.
Data authorization is another crucial intersection. When thousands of people contribute small amounts of data to form an AI, each person needs to control their data – the ability to give or withdraw authorization.
In the meantime, I am very happy to use Hedera for identity and to integrate identity into AI systems. We reach a point where you cannot distinguish the media generated by the AI of reality. The only solution is digital signatures – the content must be signed by the photographer or the journalist. But you then need confidence identity systems to check these signatures.
Coindesk: Having studied neural networks in the 1990s, long before the current AI boom, what is your perspective on current languages? Something has fundamentally changed in technology, or do we just see the results of the scale?
Baird: Many AI developments took place exactly as I expected. When Alphago defeated the world world champion, when Alphazer mastered failures, when Ais conquered the poker – I had anticipated everything. They even used almost the same techniques I had imagined. We just needed faster computers.
Autonomous cars also progress precisely as I had predicted, using the methods I planned.
But Chatgpt and the large language models (LLM) completely surprised me. The architecture of the 2017 transformer – described in the document “attention is everything you need” – represented a breakthrough that no one could have anticipated. In the 90s, we were completely perplexed by the treatment of language – by treating various approaches, but by failing each turn.
The capacities of today’s LLM always surprise me, and their future remains unpredictable. Will they reach superintelligence? Or are they going to hit a ceiling? I don’t know – and I would say that nobody does it.
Humanoid robots have also exceeded expectations. Although their physical development corresponded to my predictions, their conversation capacities – powerful by LLM – have far exceeded what I imagined possible. In the coming years, they will start with basic factory work before moving on to qualified professions such as welding, plumbing and electric work.
These technological advances will make the industrial revolution minor in comparison. Most people have not grasped the magnitude of these changes or how quickly they approach.
