Core summary: This article is based on David Liberman’s interview on the Cryptopolitan Podcast. In the interview, David emphasized the potential threat that centralized computing power poses to human civilization, while also taking a deeper look at technology’s social contract, the inevitability of hardware evolution, and how decentralized AI could reshape power structures by reusing Bitcoin’s path to success.

In Silicon Valley and the global AI race, algorithms, data, and computing power are treated as the three engines of progress. But a hidden and brutal reality is coming into view: computing power is becoming more gatekept than ever. The world’s most advanced GPU resources are rapidly concentrating in the hands of a few cloud giants and super-institutions. As Gonka co-founder David Liberman warned in the interview, this highly centralized physical structure not only builds costly “digital rent” at the point of technical access; it could also become a future tool for suppressing human freedom and deepening inequality.

Against this backdrop, the decentralized AI compute network represented by Gonka has moved well beyond a simple technical experiment. At its core, it is a radical restructuring of the production relations around computing power, an attempt to turn computing power back from an “oligopolized resource” into a “distributed digital utility.”

Physical Walls: Why Centralized Computing Power Is Approaching Its Endgame

For years, centralized data centers have been seen as the only technical paradigm for AI development. But with the rise of trillion-parameter models, the physical limits of that paradigm are starting to show.

David argues that the main driver of AI today is scale. Since Microsoft provided OpenAI with billion-dollar-level computing power in 2017, the industry has validated the logic that brute force can push models forward. But now, we are facing a mismatch between linear growth in hardware output and exponential growth in AI demand. Centralized facilities depend on extremely dense energy inputs and extremely expensive cooling systems, a model now squeezed by both the geographic constraints of energy supply and the physical bottlenecks of heat dissipation.

The deeper issue is sovereignty and decision-making risk. When AI becomes the foundational layer for measuring a country’s competitiveness and capacity for scientific and technological innovation, placing the lifeline of computing power in the hands of a small number of decision-makers is, in effect, building a fragile vulnerability that can easily evolve into technological authoritarianism. Decentralization is not simply about pursuing “democracy”; it is about building a “technical antibody” that cannot be shut down at a single point or interfered with by institutions.

Mechanism Distortion: From Hash Competition to the Paradigm Shift of Proof of Compute

Gonka’s key technical breakthrough lies in its reinterpretation of the “proof mechanism.”

In crypto’s first decade, computing power was largely wasted on meaningless hash collisions. David noted that the Bitcoin network consumes more than 23 gigawatts (GW) of electricity, exceeding the combined power consumption of the computing facilities operated by global tech giants. Gonka’s Proof of Compute mechanism tries to answer a central question: how can this enormous and already proven “mining energy” be redirected toward real AI inference and training workloads?

The essence of this mechanism is replacing “institutional trust” with “algorithmic trust.” When you send a prompt to ChatGPT, you are trusting OpenAI as a company; in the Gonka network, you are trusting mathematics. Computing power providers earn rewards by actually completing AI tasks. That means computing power is no longer just a pile of hardware, but a measurable form of liquidity generated in real time. It puts every watt of electricity to work for real intelligence production rather than meaningless numerical computation.

Hardware Evolution: Replaying Bitcoin’s ASIC Path

One of the more thought-provoking points in the interview is that Gonka’s vision does not stop at software and protocols.

David predicts that, just as Bitcoin gave rise to ASIC mining rigs that are tens of thousands of times more efficient than CPUs, decentralized AI networks will also trigger the next phase of hardware evolution. When millions of nodes come together through the Gonka protocol, they will incentivize hardware makers, such as Bitfury, to develop specialized chips designed for distributed AI inference. This autonomous hardware evolution, not dependent on NVIDIA’s single supply system, is the real trump card that could allow decentralized networks to reach “Stargate” scale and challenge the tech giants.

The Technical Deep Waters Behind the Idealism

Yet once the grand vision is stripped away, the path toward decentralized AI is full of unknown deep-water challenges:

The curse of communication and latency: How can large-scale parallel tasks be synchronized across globally distributed nodes with uneven bandwidth?

The game of verification logic: In a permissionless open network, how can the system verify at low cost that nodes have not “fabricated results” or “cut corners”?

The friction of user experience: Developers are used to the simplicity of AWS APIs. Decentralized networks must offer the same smooth experience while preserving protocol transparency.

Conclusion: Computing Power as a Future “Basic Human Right”

If the timeline is extended over the next decade, computing power will become a “basic means of survival” for the digital society, much like electricity, water, and internet access.

Gonka’s real value is not whether it can defeat Google, but that it offers a “Plan B”: a technical path that cannot be unilaterally shut down by a small number of institutions, cannot be completely blocked by geopolitical factors, and can give independent developers around the world fair access.

As David put it, AI can either become a tool that delivers broad-based abundance or a blade used to suppress the majority. Building decentralized AI computing power networks is, in essence, laying a global “intelligence highway.” It points to the arrival of a new era, one in which the production of intelligence no longer depends on whether you stand in the shadow of a super data center, but on how global computing power resources collaborate optimally through protocols.

Born from dispersion, realized through consensus. In the second half of the AI computing power race, whoever can coordinate every chip around the world most effectively will hold the real ticket into the future of intelligent civilization.

Editor’s note: This article is based on David Liberman’s interview on the Cryptopolitan Podcast and the core views of the Gonka protocol. It does not constitute investment advice. As an emerging AI infrastructure project, Gonka still faces risks from technical iteration and market volatility. Investors should remain rational.

About Gonka.ai

Gonka is a decentralized network designed to provide efficient AI computing power. Its goal is to maximize the use of global GPU computing power for meaningful AI workloads. By removing centralized gatekeepers, Gonka gives developers and researchers permissionless access to computing power resources, while rewarding all participants through its native token, GNK.

Gonka was incubated by U.S. AI developer Product Science Inc. The company was founded by Web2 veterans and former Snap Inc. core product directors, the Liberman siblings. It raised $18 million in 2023 and an additional $51 million in 2025, with investors including OpenAI backer Coatue Management, Solana backer Slow Ventures, Bitfury, K5, and partners from Insight and Benchmark. Early contributors to the project include well-known Web2-Web3 leaders such as 6 blocks, Hard Yaka, and Gcore.