The Bitcoin Pioneer Who Now Warns That AI Is Moving Too Fast
A shift from optimism to dread marks a startling evolution in the thinking of one of the cryptocurrency world’s earliest institutional backers.
When Marc van der Chijs first looked at Bitcoin in 2013, he saw the future. The Dutch entrepreneur, who went on to co-found the mining giant Hut 8, recognized the transformative power of decentralized finance long before it entered the mainstream lexicon. He backed the technology, invested heavily in its infrastructure, and rode the volatile wave of digital assets to become a prominent figure in the sector. But today, the technological landscape that once inspired his entrepreneurial spirit has shifted. The company he helped build has pivoted from Bitcoin mining to the red-hot frontier of artificial intelligence (AI) —and for the first time, van der Chijs is genuinely worried about the direction of the journey.
“I’ve become more of a doomer over the past week, to be honest,” he confessed in a recent interview, his tone a far cry from the bullish confidence of his early crypto days. The admission is striking, coming from an investor who once compared the potential of AI to that of Bitcoin during its formative years. He still believes that AI is poised to fundamentally restructure the global economy, but the optimism has been tempered by a pressing and unsettling question: can humanity actually control the forces it is unleashing? “We have lost control, actually,” he stated bluntly. “And until we get the control back, I’m actually worried that we’re moving too fast.”
This candid shift toward a “doomer” outlook aligns van der Chijs with a growing chorus of technology leaders who are urging caution in the face of unbridled innovation. It represents a significant moment of reflection, highlighting the growing chasm between those who build AI and those who are beginning to fear its consequences.
The Structural Trap of the AI Arms Race
Van der Chijs’s growing concern echoes the warnings of Dario Amodei, the CEO of Anthropic, who has recently pleaded with tech executives to seriously consider slowing down the breakneck speed of frontier model development. Amodei’s caution stems from a fear that we are approaching an inflection point where AI models, capable of recursively improving themselves, will advance beyond our ability to understand, let alone control, their logic. This rapid escalation, he argues, risks not just localized errors but widespread damage to critical infrastructure, financial systems, and the very fabric of societal order.
What is particularly illuminating about the convergence of views between van der Chijs and Amodei is their identification of a “structural trap.” It is not that developers are inherently reckless, but rather that they are locked in an intense, zero-sum competition. The race is not just between corporations like OpenAI, Google, and Anthropic, but increasingly between nation-states vying for geopolitical supremacy. In such an environment, any individual company that unilaterally decides to “slow down” in the name of safety risks being overtaken and rendered irrelevant by a competitor who does not share the same restraint. This prisoner’s dilemma on a global scale makes systemic disruption or catastrophic failure seem almost inevitable before true international guardrails can be established.
The result, according to this school of thought, is a world hurtling toward a precipice with blinders on. The pressure to release the next best model, secure the most funding, and achieve the next breakthrough creates a culture of speed that actively penalizes introspection and safety. Van der Chijs, who is a veteran of the volatile crypto world where speed and “move fast and break things” is the ethos, recognizes that AI operates on a different plane of risk. In crypto, a bug might cost investors money; in AI, a misalignment in a deeply integrated model could cost the grid its power or the military its command.
A Critical Inflection Point for Global Infrastructure
Given this trajectory, Van der Chijs fears that a major disruption—something that actually impacts the daily lives of millions through financial ruin or cascading infrastructure failures—may be the only wake-up call potent enough to force governments into cooperation. While such an event would be painful, he suggests it might be a necessary evil to reset the pace of development. The current regulatory landscape is fragmented, with the European Union leading on rule-based frameworks, the United States relying on voluntary commitments, and China pursuing a state-centric top-down approach. None of these, he implies, can effectively govern a technology that is inherently borderless and operating at clock speed.
The infrastructure concerns are not abstract. Modern AI models require massive data centers that consume enormous amounts of electricity, pushing local grids to their limits. As Hut 8 pivots its mining infrastructure to host high-performance computing for AI, van der Chijs is uniquely positioned to see the physical demands of this digital revolution. The paradox is stark: the technology tasked with solving complex human problems is, in its training phase, straining the physical resources it may one day manage. If a critical system like a national power grid were to be disrupted by an AI logic error, the cascading consequences would dwarf the localized issues seen in past technological crashes.
The entrepreneur’s journey from innovation evangelist to public pessimist is a cautionary tale for the industry. He does not advocate for abandoning AI outright—that ship has sailed. Rather, he calls for a pause, a collective breath, and an honest admission that we are flying without a fully functioning cockpit. The gravity of his statements suggests that even those who have profited enormously from technological disruption sense that we have entered a new, more dangerous phase.
Navigating the Unpredictability of Superhuman Intelligence
The core of the anxiety experienced by van der Chijs and others lies in the inherent unpredictability of advanced AI. Unlike deterministic software, where inputs and outputs are defined by rigid code, large language models and reinforcement learning agents generate novel outputs based on probabilistic weights. At a certain scale, these behaviors become “emergent”—we can see them, but we cannot fully explain how they occurred. When these black boxes are connected to our critical infrastructure, the lack of interpretability becomes a massive security risk.
“Until we get the control back,” van der Chijs repeats, illustrating the urgency of “alignment research”—the scientific effort to ensure AI systems remain aligned with human values. But this is a complex, slow-moving scientific discipline that is losing the race against the rapid, profit-driven advancement of model capabilities. He observes that a significant part of the industry is prioritizing capability over safety, treating the “doomers” as overly cautious Luddites while pushing forward to create systems that could theoretically surpass human cognition within the next decade.
For his part, van der Chijs suggests that a bit more humility in the C-suites of tech giants would go a long way. He points to the “moving too fast” phenomenon not just as a technical observation but as a philosophical one. The industry has seemingly convinced itself that progress is automatically good, without taking the necessary time to consider the long tail of consequences. The debate is no longer whether AI can do things—it clearly can—but whether it should, and at what cost.
The Future Through the Lens of a Crypto Veteran
Looking ahead, van der Chijs draws distinct parallels to the crypto market he knows so well. In crypto, the community has weathered massive crashes, regulatory crackdowns, and technological failures, emerging each time with a more robust, if chastened, ecosystem. He wonders if the AI industry can survive a similar crash—a “crypto winter” of AI—without losing public trust permanently. The stakes are higher now; a catastrophic AI failure would likely trigger a regulatory storm so severe that it could stifle innovation for a generation.
The former Hut 8 co-founder has not given up on the future, though. He remains an investor and a technologist, but he is now an increasingly lonely voice urging the sector to adopt a “safety first” culture that mirrors that of the aviation or nuclear power industries. The transition from the Wild West of Bitcoin to the complex, high-stakes world of AI has taught him that the speed of innovation must be balanced with the rigor of engineering.
His final thought on the subject is a stark warning: if we continue on our current path, the critical infrastructure that underpins our civilization will become increasingly fragile. Without intentional, global cooperation to implement safeguards, the AI revolution could quickly become the AI breakdown. For a man who has built a career on betting on the future, the fact that he is now hedging his bets on humanity’s ability to control its own creation is perhaps the most compelling data point of all. The clock is ticking, and for van der Chijs, the countdown has become deafening.
