Opinion by: Daniil and David Liberman, creators of Gonka
Bitcoin taught us something important about building infrastructure: When you reward efficiency, you get incredible innovation.
Fifteen years ago, Bitcoin mining ran on the same graphics cards gamers use. The network was slow, and the hashrate was low. But Bitcoin’s proof-of-work (PoW) system rewards miners who can process blocks most efficiently.
This created a race to build better hardware. Today, Bitcoin runs on machines (called ASICs) that are literally hundreds of thousands of times more efficient than the best Nvidia graphics cards. Not 10% better, 100,000 times better. That’s what happens when you reward the most useful work.
In just 15 years, Bitcoin mining infrastructure exceeded a whopping 16-gigawatt capacity. This represents enough power capacity to run 10 million of the most powerful Blackwell Nvidia GPUs — not just more efficient, but massively larger in sheer scale than OpenAI, Microsoft Azure, Amazon Web Services, Google Cloud and xAI combined.
AI needs the same treatment
Right now, AI runs on expensive, general-purpose chips because that’s what’s available. But imagine if we built AI networks like Bitcoin, where anyone can contribute computing power and get paid for the most efficient execution of useful AI work.
Suddenly, you don’t have to have a team of sales bros with phonebooks who can leverage their C-suite clout to sell chips — anyone can just produce, install and start “issuing money.”
Now it makes so much more sense to build chips designed specifically for AI tasks.
Hardware manufacturers would compete to make the cheapest, most efficient AI processors possible. The same market forces that transformed Bitcoin (BTC) mining would kick in. Only this time with 10x more strength because there are true blockchain believers with billions of dollars who got rich from the Bitcoin revolution.
Don’t get distracted by proof-of-stake
Many early decentralized AI projects use proof-of-stake (PoS) instead — which simply means that rewards go to whoever holds the most tokens, not whoever builds the best infrastructure.
Related: US, UK to collaborate on AI, quantum computing, nuclear energy development
Take, for example, Bittensor. The most powerful compute capacity of the network is concentrated in Subnet 64, which actually runs large AI models. Yet its miners get just 5% of the rewards of the network. The other 95% goes to token stakers or affiliated miners who contribute significantly less or no work at all.
This is backwards. We need networks that reward people building better hardware, not people with the biggest stashes frozen for yield extraction.
The real difference
Proof-of-work isn’t just about cryptocurrency. It’s about using competition to drive innovation instead of accepting whatever hardware currently exists. In just 10 years, the blockchain community could build infrastructure that produces thousands of times more compute than the rest of the centralized market.
For AI, this could mean the difference between expensive, centralized computing and an intelligence that’s as cheap and abundant as electricity. Within a few years, running AI models could cost almost nothing.
What this means for you
Today for AI is like 2009 for Bitcoin. The networks are just starting, and early participants have the biggest opportunities.
Follow proof-of-work AI projects. Contribute computing power, your own or rented from the market. Start mining. The people who build this infrastructure now will benefit the most.
Opinion by: Daniil and David Liberman, creators of Gonka.
This article is for general information purposes and is not intended to be and should not be taken as legal or investment advice. The views, thoughts, and opinions expressed here are the author’s alone and do not necessarily reflect or represent the views and opinions of Cointelegraph.
