Owned hardware, AI chips, and governance: how Bitfury and Gonka’s creators view the network’s next phase

At a recent community AMA, Vadim Krutov, COO of Bitfury (the project’s leading investor), and David Liberman, one of the creators of the Gonka Protocol, discussed the strategic partnership, network stability, specialized hardware, governance, developer demand, and external inference markets.

Vadim Krutov, COO of Bitfury, and David Liberman, co-creator of the Gonka Protocol. Illustration by Gonka Community Blog.

Below, Gonka Community Blog takes a closer look at several parts of the discussion we think matter most from the AMA (hosted by Demian G. and S. Paranko), edited for clarity and length.

Bitfury sees owned hardware as key to network stability

Beyond its financial commitment, Bitfury has been in active conversations about where the Gonka ecosystem goes from here.

Since stability is one of the key challenges for decentralized infrastructure, Vadim Krutov highlighted an important point at the very beginning of the AMA: Bitfury is thinking about becoming one of the network’s anchor miners by deploying its own hardware on the network.

“Today, most of the hardware running on the network (and possibly all of it) is rented. As rental contracts expire, miners can decide whether to keep their machines in the network or move them elsewhere.

On one hand, that competition is healthy. It encourages everyone to think about tokenomics, demand, and the long-term attractiveness of the ecosystem.

On the other hand, it also increases volatility. Strategically, we would like to see a more stable network. A stable network needs what you might call anchor miners: participants who put hardware on for the long term rather than moving in and out based on short-term market conditions.

As David and I have talked about, moving from rented infrastructure to owned hardware would improve the network’s stability and create a foundation that’s less exposed to fluctuations in the GPU server market.

We are considering buying part of that hardware ourselves and operating it until it becomes obsolete. Whether that point arrives via Kimi 2.6, MiniMax, DeepSeek, or something else, we don’t know yet. Our expectation is that the equipment will stay useful for at least one or two years and act as a reliable backbone for the network.”

Bitfury looks beyond GPUs to specialized AI chips

Vadim also pointed to a longer-term trend that could eventually benefit both Gonka and the broader AI infrastructure market: the emergence of specialized AI hardware optimized for specific workloads rather than general-purpose computing.

In his view, that kind of hardware could be a more efficient use of capital than today’s general-purpose AI infrastructure.

“Tenstorrent, a company that David and Daniil introduced to us, is doing exactly this. They are building specialized hardware. Not expensive, general-purpose systems designed for every possible use case, but solutions optimized for specific models.

We hope to start testing their hardware in our data centers within the next month. If the results look good, it could become part of our investment strategy.

Other startups are also working in this space. As a company that both invests in and incubates startups, we are already seeing a number of similar opportunities.

One example is Axelera, one of our portfolio companies. Some of you may know it as a European champion in AI chips. The company is already approaching a billion-dollar valuation. We built it internally and continue to work closely with the team.

Demand for AI chips is growing so quickly, and Nvidia’s margins and valuation have grown so large, that more companies are bound to enter this market with their own chip designs.”

Clarifying the ecosystem investment plan

Vadim also gave more context on Bitfury’s previously announced $12 million ecosystem allocation, which forms part of the company’s broader $50 million commitment to Gonka.

The funds are already in a dedicated smart contract, but Bitfury intends to deploy them gradually, focusing on risk management and governance quality.

“First, we look at network security. No matter how well everyone works or how many security audits are done, bridges occasionally fail. If you look across the industry, bridge exploits have hit almost every major protocol, including mature ones.

Because of that, we want to move step by step, both in our own interests and in the interests of the community, to reduce risks.

The second focus is governance efficiency. This is an extremely important topic for us. How these funds are ultimately spent is one of the most important questions we’re looking at.

We plan to work in two directions.

The first is improving the maturity of proposals, voting, and result tracking. We intend to share our views on why we support or do not support particular proposals. We’ll also explain what we consider a high-quality proposal in terms of preparation, commitments, and business case. We want to see mature proposals and mature discussions.

The second direction is the Foundation.”

Donate

How Bitfury sees the Foundation’s role

According to Vadim, decentralized decision-making is valuable and well suited to a protocol like Gonka. But like any form of direct democracy, it has trade-offs.

“The main drawback is the slow speed in decision-making, and sometimes the absence of a common direction that helps align decisions across the ecosystem.

We are currently discussing the creation of a Foundation with community members and, first and foremost, with David and Daniil. Many successful protocols have done something similar, often with a Foundation registered in Switzerland as a non-profit, with delegated responsibilities for management and coordination.

Here’s a simple example. If an organization wants to enter into an agreement with the ecosystem today, it can’t sign a contract directly with the protocol. It has to find a party that can act as a counterparty. Sometimes that’s Bitfury, sometimes it’s someone else. But what is needed is a systematic mechanism. A Foundation can be that mechanism.

Today, many initiatives come from community members based on their own understanding of the situation. The Foundation would take the roadmap we’ve all approved as a set of development priorities and help make sure the right initiatives are proposed and carried out.

We believe that when we create this Foundation together, it should not be controlled by Daniil, David, or Bitfury.

We may act as its incorporators, but the plan is then to appoint a reputable board of directors. People who understand how Foundation structures work and would run it independently from us, following the roadmap the community has approved.”

Why the creators believe the Foundation should not weaken the community

David Liberman, one of the creators of the Gonka Protocol, described the Foundation as a mechanism that could make governance more organized without taking authority away from the community.

“For us, it was very important that the Foundation would not become a formal structure behind which we simply hide our own interests. We wanted the community to form first. People who genuinely believe in the protocol and support it. Not those who join for a week, but those who stay with the protocol for months.

We wanted the Foundation to emerge as a result of a community decision, so that it could ultimately serve the role defined by the community itself.”

The recently approved roadmap, in his view, provides a democratically approved direction for the ecosystem.

“At the same time, having a mechanism that helps optimize this process, led by people with strong reputations not only within Gonka but also beyond it, seems to me to be the right path forward for the community.

As Vadim already mentioned, this does not change the fact that the main resources remain in the hands of the community. If the community decides that something else needs to be done beyond what the Foundation directly coordinates, it can still do that.

Now that the community is much more empowered and understands that the future of the project depends on each active participant, I don’t think the creation of a Foundation would take anything away from the community. It wouldn’t reduce its role or diminish its influence. And I hope the community will remain active.”

Demand, developers, and real usage as roadmap priorities

Vadim described the emergence of a community-approved roadmap as one of the most important milestones in Gonka’s development over the past quarter.

With that roadmap in place, he turned to the question of execution. Which priorities should get the most attention as Gonka moves into its next phase?

Speaking about Bitfury’s current priorities, Vadim put demand generation and developer adoption ahead of additional mining investment.

“Yes, it’s important to develop in all directions. But if we had to choose between additional investment in mining and investment in demand generation and the developer ecosystem, we’d choose demand.

First, this means demand from the B2C side. OpenRouter and other model aggregation platforms are good examples. Integrating Gonka into the infrastructure of such services looks like one of the most obvious growth opportunities.

It is also important to make onboarding as simple as possible. For B2C adoption, user interfaces have to be extremely easy to use. We are already seeing more brokers emerge, and the more brokers, interfaces, and access points the network has, the better.

B2B is harder. For businesses, privacy is much more important. That’s why privacy technologies, encrypted requests, and user data protection are among the key items in the roadmap.

One more priority I’d highlight, without taking away from the others, is community building. We need to raise awareness among developers, work with them directly, make integration easier, and promote Gonka at industry conferences.

Developers should know that they can put Gonka under the hood of their applications, and that doing so is easy. We need to achieve what marketers call spontaneous awareness.

When someone is asked, “Which decentralized AI network comes to mind first?” the answer should be Gonka.

That is what really matters. Only real usage creates a strong foundation for long-term growth.”

David Liberman framed the same priority in terms of one concrete benchmark: inference traffic.

“From a developer perspective, we need to achieve more inference traffic than any other decentralized AI network.

I often mention Chutes ( a Bittensor subnet focused on AI inference — editor’s note ) as an example of what’s possible. But if you look closely, the gap between us isn’t actually that large. Surpassing them on inference traffic is entirely achievable.

Once we surpass Chutes on that metric, I believe it will become much easier to make sure that when people think about decentralized AI, they think about us first.”

OpenRouter is only part of the path

Returning to the tactical priorities that could accelerate demand, David pointed to OpenRouter and similar routing platforms as important distribution channels for Gonka’s inference capacity.

“Many counterparties won’t work directly with a protocol. Not because we’d prefer they wouldn’t, but because these organizations have their own requirements and constraints.

We don’t want to be just another small provider on OpenRouter. Whoever represents the network should have the same goal as the community itself: to dominate on OpenRouter, climb the rankings, and keep increasing network usage.

It’s important for Gonka to be highly visible on platforms like these. At the same time, I’d emphasize one more thing.

I strongly encourage all brokers to look beyond OpenRouter. Yes, it is a major and important goal for the community. But there are many other routing platforms too, including open-source projects. In some cases, adding a broker or provider can be as simple as making a contribution on GitHub.

That is why I invite everyone interested in growing the ecosystem to take part: add integrations, connect to new platforms, and create more ways for Gonka to reach users.”

Disclosure: Contributors to this blog may participate in the Gonka ecosystem and may hold project tokens. Nothing in this publication constitutes financial or investment advice.