Tomasz Stanczak, the founder of Nethermind and former co-executive director of the Ethereum Foundation, said he wants to spend the next decade working on robotics, automation and the physical world after stepping away from the foundation and day-to-day crypto operations.
In the first xBuilders interview in The Round Trip, a series produced by PANews and Web3.com Ventures, Stanczak looked back on his path from finance to Ethereum infrastructure and explained why he now sees the biggest opening outside the digital layer. His argument was straightforward: intelligence in software is becoming cheaper and more common, while manufacturing, logistics, automation and data centers still contain a large amount of friction that machines and humans have not yet removed.
From London finance to Nethermind and the Ethereum Foundation
Stanczak said he entered crypto in 2017. Before that, he had been working in finance in London, including roles in banking and hedge funds. He later left that world and founded Nethermind in 2017.
Over the following years, Nethermind grew into a company with several hundred employees and what he described as stable revenue. In March last year, he joined the Ethereum Foundation and stepped down as Nethermind’s CEO. He said one major reason for that move was the need to remove potential conflicts of interest after joining the foundation, which meant cutting many earlier business ties.
After leaving the foundation, he kept only a board seat at Nethermind and no longer handled daily management. That, he said, gave him room to think clearly about what he actually wanted to build next.
The three areas he wanted to study after his departure were AI, robotics and the physical world. In his view, software itself used to be the value layer, but AI is rapidly turning digital intelligence into a commodity. The harder and more promising work now sits in the real world, where there are still large operational gaps.
He said he initially intended to explore the space loosely. A few days after leaving his role, he was invited to visit HF0’s residency program. That visit quickly clarified his direction, and about a month later his new company was registered.
Starting again meant learning again
Asked what it felt like to start over, Stanczak said building a company means repeatedly putting yourself in an uncomfortable environment. He described the HF0 residency as a strong experience because it forced him back into what he called “pain mode,” where he had to learn new things from scratch or recover skills he had not used recently.
He said he had not done a large amount of coding over the past three to four years. At the same time, coding itself has changed in the AI era, which means founders now need stronger architectural thinking and a better grasp of how AI agents can be used to build systems.
He also mentioned a recent trip to Detroit, where founders in manufacturing were building drones, robots and other physical products. The city’s energy, he said, reminded him of the early blockchain community, when people met frequently in person, collaborated and built together. Robotics, unlike purely digital systems, naturally pulls people into labs, factories and test grounds.
He expects many people who once worked in blockchain to move naturally into these real-world “physical nodes.”
Why he thinks robotics is at an inflection point
Stanczak said the key shift is that technologies that used to advance separately are now maturing at the same time.
AI models and training capabilities are improving quickly. In parallel, robotics teams are exploring different technical paths and trying to answer the same question: how to make robots actually useful. He said AI is not only making robots smarter. It is also changing the process of building them, including materials design, mechanical design and system optimization.
He contrasted that with earlier manufacturing cycles, where a single design change could take six months to verify. Today, more mature simulation environments let teams run many experiments in virtual settings before touching hardware. Add falling actuator costs and better hardware supply chains, and the pace of research and development changes sharply.
He said people inside the field see this year as the most exciting one in robotics history. While critics still point out that humanoid robots have not been deployed at scale, he argued that the lasting value of robotics is not simply replacing people. It is filling labor gaps in the physical economy.
His example was data centers. According to Stanczak, the global data center industry is short by hundreds of thousands of skilled maintenance technicians needed for operations and upkeep. Robotics companies are moving into that market in an effort to close the gap through automation.
For him, the question is no longer whether robots will show up. It is which industries they will enter first.
Humanoid robots are only the first step
On the current wave of humanoid robot startups, Stanczak said humanoids are a beginning, not the final form.
He said the industry will initially make robots look more like people because that is the fastest route into the existing world. A human-like body can use work environments already built for people, and remote operation is easier in that format.
That will not last forever. Over time, he expects the industry to realize that many jobs are not suited to a human shape at all. Future robots will be built for specific tasks. Some will not need two legs. Some will not need two hands. They will be designed around the job rather than around the human silhouette.
He expects that pattern across agriculture, healthcare and manufacturing. Logistics and warehousing have already shown part of this path, he said, with large warehouses becoming highly automated long before the rest of the economy.
That is one reason robotics reminds him of Ethereum in 2016 and 2017. Back then, Ethereum was criticized for being slow, immature and exposed to security concerns. Even so, people could already see a roadmap and the outline of a system that might solve real problems. He pointed to stablecoins as one of blockchain’s most successful applications, saying usage and adoption have risen quickly over the past two years.
He also said early enthusiasm around AI agent payments on blockchain faded after an initial burst, but he expects some of that attention to return and said he has seen “some interesting data.”
In his telling, robotics is at a similar stage now: products remain immature, many are still in laboratories, and large-scale deployment has not happened. Yet the potential is visible enough that belief in the path is growing, and once that belief spreads, progress tends to accelerate.
He added that his interest in autonomous machines did not begin recently. The company behind Nethermind was named Demerzel Solutions Limited, with “Demerzel” taken from Eto Demerzel, a humanoid robot character from Isaac Asimov’s Foundation and Robot series. He said his new company continues that tradition. To him, this is not a sudden turn. It is a seed planted about a decade ago that now happens to have reached the right moment.
Blockchain payments, in his view, are more likely for agents than for people
When the discussion turned to crypto payments, Stanczak said he is not currently focused on blockchain or crypto features inside robotics. His work is centered on automation in the physical world, especially automated deployment of operating systems and autonomous business processes.
His company is only about one and a half months old, he said, and much of the vision is still being defined. Payment systems are not the first issue he needs to solve.
Still, he said it is natural to expect blockchain-based payments to be used by AI agents and robots in the future. He has long argued that blockchain should function as a base layer that becomes effectively invisible to both humans and AI. If the infrastructure is good enough, he said, it should disappear into the background.
He repeated a view he often gives in interviews: blockchain is not something he sees as directly user-facing. Most people do not verify a blockchain directly, except for a narrow set of users who run their own nodes and understand the cryptographic and mathematical details. In practice, most people stand behind wallets and block explorers. If those access layers are centralized, they can become the weakest link and may even present false information. He said a recent Ethereum Foundation announcement also highlighted this access-layer issue.
That is why he sees AI as critical between users and blockchains. AI can help users receive and verify information, especially when real-world information is hard to check. Over the long run, he said, the stack needs verifiable sensors and verifiable AI to expose what is actually happening in the physical world, even though building that remains difficult.
On payment disputes, he said chargeback mechanisms will likely reappear on top of blockchain systems. The base transaction layer still needs to preserve success-or-failure finality, but higher layers can build dispute or verification mechanisms that reduce trust assumptions. In his view, that could help robots balance trust against latency.
He also said robots may end up using Ethereum, a Layer 1, a Layer 2 or a dedicated chain, but control would ultimately sit with an AI agent. While robots may look localized from the outside, he expects many of them to operate as group systems, with one agent coordinating multiple edge devices or machines.
A company built around deploying data centers
Pressed on what his new company is building, Stanczak said the current focus is “deployment of data centers.” He tied that directly to the decentralization logic behind blockchain nodes.
He said many people complain that Ethereum nodes have become demanding enough that they can only run in the cloud or in data centers. Real decentralization, in his view, would mean being able to deploy a node extremely quickly and pair it with its own data center so it can plug into physical space fast.
Robots fit into that picture by helping data centers come online and operate autonomously. He said the goal is not to replace humans but to address labor shortages and optimize the entire deployment workflow, including logistics, supply chains, legal approvals and permissions.
His team, he said, is working on everything tied to building in the physical world, whether that involves energy, data centers or drone fleets.
He framed the problem from the perspective of AI as a future customer. If AI systems need to generate their own intelligence sources across different locations, then today’s human deployment interfaces are too slow. He described the product ambition as taking something like GitHub and extending it into physical space.
He connected that idea to a concept he had discussed in talks at Devcon in Cannes and Bangkok: a “social GitHub” where pressing a button and supplying capital would let a system deploy itself. He compared that idea to Factorio, the factory-building game in which everything starts from computation and energy. He also used the phrase “recursive deployment” to describe a loop where agents deploy robots and robots deploy more robots.
He acknowledged that it sounds close to science fiction. Even so, he said AI, materials and automation are advancing fast enough that the loop is becoming more plausible.
What he wanted to change at the Ethereum Foundation
Looking back on his time at the Ethereum Foundation, Stanczak said he joined with a clear goal of pushing organizational change and that the foundation was ready to accept difficult adjustments.
The changes he wanted were a more structured core organization, a more goal-driven operating style and better transparency. In his account, one of the main problems at the time was that the foundation was not communicating boldly enough, largely because of the exceptionally tough regulatory environment in earlier years. Conditions had started to improve before he arrived, which meant part of his task was simply to say that the organization could now speak more openly.
He said many infrastructure operators, including L2 operators, DVT teams, staking projects, Consensys, Bitmine and Nethermind, were already becoming more active in public discussions around Ethereum. That mattered because it meant the foundation was no longer the only node speaking for the ecosystem.
He argued that internal and external communication are linked. If an organization is opaque externally, internal communication also deteriorates because people fear leaks. He said the Ethereum Foundation improved the situation by activating its official Twitter account and speaking more often with large companies, banks and other institutions.
In management terms, he said his core objective was simple: maintain a pace of two fork upgrades per year. That target, in his view, defined the Ethereum Foundation’s proper role. The foundation should not dictate what must be built. It should coordinate, provide strong testing, security review, research guidance and funding, and help core developers ship on time.
Once the delivery target is explicit, he said, responsibility inside the organization can be reorganized around it. In earlier years, the foundation may have been too focused on absolute decentralization and too suspicious of hierarchy. Some degree of goal orientation was necessary, in his view.
Foundation direction, enterprise work and the fading private-chain story
Asked about reports that the Ethereum Foundation is shrinking while another nonprofit, Etherealize, will take on more research and development work, Stanczak said he did see a noticeable shift after his departure: less focus on enterprise and more focus on privacy.
He said that is probably healthy for the foundation because it does need to return to its core role. At the same time, he said he did not like the form that some of the shift took, including the idea that people needed to sign a compulsory agreement to stay. He also said a “Milady” style culture is not his vibe and that he does not want Ethereum forced into one particular shape or aesthetic.
His preference is for Ethereum to remain completely neutral, open to different paths and equal participation. He also said the foundation should periodically pull its course back toward its central mission: values, cypherpunk principles, censorship resistance, open source, privacy and security. If the foundation sees itself as one node among many, its remit can become narrower and clearer.
He also mentioned EthLabs, saying the newly formed group includes highly capable leaders who are more focused on Ethereum adoption in finance and even AI. In his view, both the people who stayed and the people who left to form new teams are strong, and technical disagreement of that kind is productive.
On Nethermind’s recently announced work with UBS to explore privacy technology on Ethereum, Stanczak said the development was gratifying. He credited the current management team, led by Daniel Salazar, and said the group has worked together for years and remained deeply aligned with the company’s vision while building a strong enterprise position.
He said Nethermind had identified 2025 and 2026 early on as key years for deeper enterprise conversations. With regulation thawing, stablecoin usage broadening and tokenization gaining traction, innovation teams inside banks that had previously been watching from the sidelines were beginning to build. Nethermind, he said, used its engineering and research base to work with top financial teams globally.
Enterprise privacy remains a hard problem because systems must protect user privacy, meet compliance requirements and prevent abuse once they are usable at scale. He added that, to his knowledge, Etherealize is also focused on enterprise privacy and tokenization of asset classes.
He said that when major financial institutions finally choose public blockchains, it validates the message his side pushed last year: do it on public chains, because there is no point in experimenting with private chains anymore. In his account, earlier non-public-chain efforts such as R3 and various so-called alternative chains that marketed themselves as public while functioning more like private systems are now being pushed out by the market. Ethereum, he said, achieved permissionlessness in a much stricter sense.
For that reason, he sees the return to public chains as a better outcome for everyone because it means higher security.
The through line: removing latency from coordination
At the end of the conversation, Stanczak was asked what unifies his work across finance, Ethereum and robotics.
His answer was that he has always been building the same thing: a coordination layer. More precisely, he said he tries to optimize coordination and remove latency from systems.
He used Nethermind as an example. “Fastest sync time” was always one of the team’s ultimate goals. He said the core engineering group used to track node market share every morning, and when Nethermind reached 33% market share, the team would actively help other client teams in the Ethereum ecosystem. He called that collaborative culture one of the ecosystem’s strengths.
At the same time, the team remained deeply focused on performance. How fast can the mainnet sync under different modes? How fast can other chains sync? Those are all latency questions. The same logic extends to infrastructure: how quickly can a node be deployed, and how quickly can it catch up to the latest block so that it can actively participate in block-building decisions?
He said the same reasoning now shapes his work in the physical world. The problem there is how fast a data center that can run a node can be deployed into physical space. At that level too, the work comes down to identifying bottlenecks and deciding which one matters most right now.
That, he said, defines not only what he did in the past, but also the company he is trying to build now.