Gonka AI Conversation: The five giants monopolize 80% of the computing power. How can AI belong to everyone?

Interview: The Round Trip

Compiled & Edited by: Yuliya, PANews

As the AI wave sweeps across the globe at an unprecedented speed, a “military race” centered around computing power has already begun. When Nvidia’s market value surpasses one trillion dollars, and giants like AWS and Google Cloud nearly monopolize cloud computing, a profound challenge faces all AI innovators: Will the high concentration of computing power stifle open innovation and lock the future of AI within the “walled gardens” of a few companies?

With a successful track record of selling a company for $60 million to Snapchat and founding Product Science, which provides AI code optimization services for top-tier enterprises, Gonka AI’s co-founders David and Daniel Laborman—brothers with continuous entrepreneurial experience from parallel computing to AR—offer a unique perspective to break this dilemma: building a fully community-driven decentralized AI computing network.

In the new series Founder’s Talk of “The Round Trip,” produced jointly by PANews and Web3.com Ventures, David and Daniel elaborate on why they drew inspiration from the history of Bitcoin’s infrastructure development, aiming to replicate the “ASIC revolution” in the AI field through an open financial incentive framework, to thoroughly break the shackles of computing costs. They share how Gonka AI attracted $50 million in funding from industry giants like Bitfury and offer their insights on the current “AI bubble theory.”

From Gaming, AR to Decentralized AI

PANews: Welcome, David and Daniel! We’re glad to have you here. I know you both have a very strong technical background and have been deeply involved in this field for many years. Could you start by sharing your background story with our audience?

Gonka AI: Hello everyone. First, we are brothers, and our lives and careers have always been closely linked. Our story dates back to 2003, when we first became interested in parallel computing and decentralized networks.

Later, we entered the online gaming industry, which is essentially a form of large-scale parallel computing—thousands of players interacting in real-time over the internet. To improve the efficiency of game animation production and reduce costs, we then delved into the field of computer vision.

Computer vision led us to a new direction: developing AR virtual avatars for Snapchat. This experience was very successful, and ultimately Snapchat acquired our company for $60 million, marking a significant transformation in our careers.

Through various projects and companies, we have always harbored a desire: to create something that can have a truly significant social impact, especially in how society interacts. When AI entered our lives in a new form—the large language models (LLMs)—everything changed. It was no longer just machine learning as we knew it; it became a powerful tool capable of real conversations and genuinely helping us solve problems. We saw that the new generation of AI based on the Transformer architecture is not just language models. Whether it’s image generation, video creation, breakthroughs in biology, chemistry, physics, or even more efficient nuclear power plant design and operation—this wave of AI is affecting everything.

Next, we will also see rapid developments in robot software and autonomous vehicles, and these changes are happening very quickly, right now.

But alongside this comes a concern—not a sci-fi “Terminator” fear, but a worry about the current landscape. Currently, about 65% of global cloud computing power is controlled by three US companies (AWS, Google Cloud, etc.), and if you add China’s Alibaba and Tencent, these five giants control up to 80% of the world’s cloud capacity. The core of AI is computing power, and at present, AI is almost synonymous with cloud computing. These companies are fiercely competing to control 100% of AI compute power. If this continues, we will enter a very strange world:

Only a tiny number of companies will truly own and control all AI, and these AIs will:

• Replace many jobs
• Reshape the entire economic structure
• Change the way society functions

Therefore, we believe that decentralized AI is a crucial and unavoidable issue.

This is why we ultimately decided to join Gonka AI.

PANews: Indeed, you are not newcomers to the AI field. Before founding Gonka AI, you also founded Product Science, a company backed by well-known institutions like Coatue, K5, and Slow Ventures. Can you talk about this experience and how it ultimately led you to Gonka?

Gonka AI: Of course. Our previous focus was on computer vision, which is essentially AI and machine learning. The earliest practical applications of AI largely emerged in image generation and animation production, which helped us establish a reputation in the machine learning industry.

After leaving Snapchat, we founded Product Science. This company used AI to provide code optimization services for top global companies like Walmart, JPMorgan Chase, and Airbnb. Today, AI that helps write code is well known, but equally important is ensuring that this code runs efficiently. Before fully shifting our focus to Gonka and decentralized AI infrastructure, improving code performance was our core business.

Gonka AI’s “Bitcoin”-style Concept

PANews: You mentioned the problem of concentration of computing power, which is indeed concerning. Recently, Cloudflare’s large-scale outage caused half the crypto world to go offline, and AWS frequently experiences failures, impacting many applications. How will Gonka AI address this issue? It seems not to be a general decentralized cloud, but more focused on AI.

Gonka AI: Yes, facing the current dilemma of highly concentrated computing power, we see the only way out is decentralization.

At the model level, independent labs like DeepSeek have already demonstrated that they are fully capable of training high-quality models comparable to tech giants, but the core bottleneck remains compute power. Currently, many cutting-edge labs rely on infrastructure built by large cloud service providers, and in the decentralized space, no comparable large-scale solutions have emerged. Even the largest decentralized AI compute network today, Bittensor, has only about 5,000 data-center-grade GPUs. Meanwhile, companies like OpenAI and xAI are building massive clusters with millions of top-tier GPUs. The scale gap is enormous.

We realize that, to truly make AI belong to the people and avoid single points of failure, the only solution is to build a decentralized compute network of comparable scale. This is where we drew great inspiration from Bitcoin. We don’t just see it as “digital gold,” but as one of the greatest frameworks for building large-scale infrastructure.

Over the past 15 years, the Bitcoin community has built an incredible infrastructure through decentralization. Today, the Bitcoin network has about 26 GW of data center capacity, surpassing the combined total of Google, Amazon, Microsoft, OpenAI, and xAI. It is a massive project built by countless independent participants worldwide, aiming to escape centralized systems.

What’s equally astonishing is the speed of hardware innovation. In 15 years, the energy required to mine 1 TH/s of Bitcoin hash rate has dropped from 5 million joules to just 15 joules—a 300,000-fold efficiency improvement! We believe that, if AI compute power can be transformed in the same way, true “computing abundance” will become possible, and AI will be accessible to everyone on Earth.

Host: I noticed that early Bitcoin infrastructure giant Bitfury recently announced a $50 million investment in you. Does this suggest the market sees a similar pattern? Bitcoin made energy “interchangeable,” because whether energy is in Siberia or Silicon Valley, it can be converted into a homogeneous value of compute power. Are you also making compute power “interchangeable”? Considering AI’s sensitivity to latency, would that be a challenge?

Gonka AI: We believe the same story will unfold in the compute power domain. Currently, Nvidia chips are extremely expensive, and most of the costs in data centers built by companies like OpenAI go to Nvidia. But if we can replicate the ASIC (Application-Specific Integrated Circuit) innovation for AI, the world will be very different.

When the hardware cost of individual compute units drops significantly, energy costs will again become a key variable. Early mining companies and hardware manufacturers like Bitfury investing in this ecosystem send a strong signal: they recognize the same pattern that drove Bitcoin’s early development.

Recall 2012, when GPUs were the dominant mining hardware, but within a few years, ASICs—offering tens of times the efficiency of general-purpose chips—became the only feasible mining path. The companies that made these ASICs weren’t big tech giants but small startups. This was entirely driven by Bitcoin’s financial incentive framework:

• Open competition: Whoever provides the most effective compute power to the network earns the largest share of tokens.
• Positive feedback loop: As token prices rise, rewards become more attractive, encouraging more participants to increase the network’s total compute power.
• Lowering innovation barriers: A small company in Korea or San Francisco, just by designing more efficient chips, can enter the market without large sales teams, without establishing relationships with giants, and even without traditional investors—simply by connecting their chips to the network and proving their effectiveness to start earning.

This framework greatly reduces the barriers and complexity of “producing compute power.” We believe this scenario will repeat in the AI chip domain. Once the protocol is established, people can earn by connecting their devices—whether it’s their own computers, purchased Nvidia GPUs, or rented data center capacity—to the network. We expect that within the next one or two years, this financial-incentive-driven innovation will bring hundreds or thousands of times more compute capacity to the AI network, thoroughly breaking today’s compute bottleneck.

How Will Decentralized Networks Reshape the Compute Market?

PANews: This model is very interesting, reminiscent of early crypto miners using idle GPUs in schools for mining. Now many companies buy expensive H100 GPUs, but most of the time they are idle because they don’t know how to utilize them fully. Does your network also attract such users?

Gonka AI: We’ve indeed encountered many similar and even more exciting cases. Some very successful AI startups, during the early hype, bought hundreds of H200 GPUs with investor money, but to this day, only half are effectively used.

Another common scenario is that many companies rent large data center compute resources to run open-source models. Later, they realize they can do something smarter: instead of running models inefficiently themselves, they can use Gonka’s API to access the same services; meanwhile, they install Gonka nodes on their rented GPUs and contribute to the network. This way, they can both use AI models and earn tokens, achieving much higher efficiency and returns than before.

To efficiently utilize GPUs, you need to handle thousands of requests simultaneously, which is very difficult for a single project. So, companies either tolerate low utilization of their own (or rented) hardware or pay high API fees. Joining the network and becoming part of the ecosystem is a better choice.

Many participants in our network are not just “idle” compute power. For example, data centers like Gcore and Hyperfusion are highly efficient operators with little idle capacity. But in recent months, they found that connecting their GPUs to Gonka can earn higher returns than directly renting to clients, because they gain exposure to the network’s growth. So they are gradually shifting hundreds of GPUs from rental business into our network.

This is key to scaling from thousands to millions of GPUs. Although giants like OpenAI have bought most GPUs on the market, millions remain scattered among independent participants. They can’t compete alone, but together they form a powerful force.

This logic also applies at the national level.

A year ago, when we talked with some governments, their main idea was “building their own clusters to develop sovereign AI.”

A year later, when we meet with ministers from the UAE, Kazakhstan, and others, they all realize that as small independent players with few GPUs, they cannot compete with the giants.

But if they join a large, trusted decentralized network, it’s entirely possible to maintain sovereignty because everyone can trust a decentralized system.

The AI Bubble Debate: Wave of the Era or Collapse of a Specific Bet?

PANews: There’s no denying that AI is experiencing huge enthusiasm and rapid growth. But with high expectations from investors and users, are we heading toward an “AI bubble”? Many compare it to the internet bubble of 2000.

Gonka AI: That’s a very interesting question. Looking back at the 2000 internet bubble, although it experienced a “small crash,” what has the world become 25 years later? The internet is a real technological revolution, and the economic transformation it brought is real. Companies from that era have grown into trillion-dollar giants, fundamentally changing our lives.

Compared to the internet, the revolution AI will bring is even more radical and thorough. Imagine that in the next 30 to 50 years, everyone will have a personal robot capable of working in factories for them—that’s not science fiction, but an imminent reality. So, it’s rational for investors to pour hundreds of billions into this technology.

Of course, there will be failed investments, just like in the past 30 years of venture capital, with many losing money. But overall, the returns are extremely high, and it is genuinely changing the world.

So, whether it’s a bubble depends on your perspective. Some companies will go bankrupt due to wrong assumptions. For example, Gonka’s judgment on the feasibility of decentralized AI might be wrong; conversely, all the current bets on Nvidia could also be a huge bubble.

History has seen similar scenes. In 2012, Nvidia’s stock soared because of the crypto narrative, as the market believed it would dominate mining. But then the ASIC revolution happened, and Nvidia almost completely lost that market. Now, AI is bringing even greater value to Nvidia because the market expects it to be a multi-trillion-dollar giant. That expectation might be correct, but no one can guarantee Nvidia will stay dominant forever. If the ASIC revolution occurs again in AI, what will happen?

Imagine rebuilding the entire Bitcoin network’s compute power, but not with ASIC miners—using Nvidia’s latest Blackwell chips. You’d need to invest $5 trillion! That’s obviously unsustainable.

Therefore, what we’re discussing might not be an “AI bubble,” but rather a “bubble formed by bets on specific companies and technologies.” If the market’s judgment on Nvidia is wrong, 5 to 7 trillion-dollar companies could suffer heavy losses. But that doesn’t mean AI itself is a bubble. The technology will not disappear, nor will the process of changing lives and business. Only the companies carrying these values might change.

PANews: I completely agree. Just as we no longer say “I use the internet,” but rather “I use a certain app,” and that app happens to use the internet, in the future, every application will incorporate AI in some form. It will become ubiquitous, so much so that we won’t even notice it.

Gonka AI: Exactly. If you look at the Nasdaq index’s chart from inception to now, you’ll see that the 2000 “big crisis” was just a tiny wave in a decades-long growth curve. Back then, people thought all goods would be sold online within five years—that didn’t happen, but it did occur within 15 years.

The same applies to AI. The future where robots are everywhere might not happen in five years, but it’s almost inevitable. No force can stop it. From this perspective, the future growth in compute demand will be thousands of times larger. What we need is a long-term economic model, like Bitcoin’s, designed to support this vision for the coming decades.