Opinion: Zain Jaffer, co-founder of Vungle
Artificial intelligence is the latest frontier in the battle of centralization versus decentralization. As Bitcoin and Ethereum were built to resist government and corporate management, the Crypto AI project opposes Big Tech’s growing advantage over the AI model.
The question is, can they compete or are they another layer built on the same concentrated infrastructure that they claim to be confused?
AI, cryptography, and decentralization dilemma
One of the central doctrines driving Crypto’s traditional fans is decentralization. It directly opposes the Securities and Exchange Commission’s Howie Test. This defines investment contracts as relying on profits, relying on “common companies” and “the efforts of others.” While most securities are tied to centralized companies, Bitcoin, Ethereum and other well-decentralized networks are designed to function without central authorities.
Why is this important? For control. The Bitcoin White Paper famously describes a “purely peer-to-peer” system that allows transactions without going through financial institutions. This libertarian ideal (two parties that trade freely without interference) promoted the evolution of Crypto.
As AI becomes more and more mainstream, the same decentralisation spirit is expanding to it. Crypto advocates worry if AI goes into the hands of a few tech giants it will become another walled garden controlled by companies like Google, Microsoft, Openai and more.
Crypto’s AI Push
To counter this, blockchain-based AI projects are emerging. Names like Tao, Virtuals (Base), AI16Z (Solana) have launched a decentralized AI model, hoping to disrupt the industry before Big Tech takes over for good. Some have built their own large language models (LLMs) from the ground up and trained independently of the corporate AI Giants.
challenge? data.
Training AI models requires a large amount of high quality data. The Crypto AI team rubs open web, but still has no access to its own enterprise dataset. The deep integration into corporate workflows makes the tech giant a huge advantage here. This means that fully decentralized AI teams are inherently at a disadvantage due to slow progression, weak models and low adoption.
Recently: Deepseek – Wake-up call for responsible innovation and risk management
Conversely, some crypto AI teams have adopted a different approach, rather than building models, leveraging existing centralized AI infrastructure. They use APIs from Openai, Microsoft Copilot, or Google Gemini and act effectively as a distributed frontend for a focused AI backend. This allows them to launch quickly, but that raises the question: are they really decentralized or another layer that relies on big technology?
Cost Factor
Beyond decentralization, there is also the issue of cost. Developers will agree that some degree of AI hallucination has an acceptable threshold as long as it is possible to run and repeat experiments at an affordable price. However, with US AI providers, costs increase quickly. Closed source models such as Openai Force Developers will have a Pages play structure regardless of output quality.
Please enter DeepSeek.
In late January 2025, the China-based AI startup disrupts the landscape by unveiling a smaller, highly efficient LLM, which is reportedly using fewer computing resources, consistent with ChatGPT’s performance. Unlike the US billion-dollar weapons race (where Openai’s $500 billion Stargate initiative controls the headline), Deepseek built the model on a $6 million budget.
What Deepseek means to Crypto AI
Some crypto AI teams have already begun consolidating DeepSeek as an alternative to the US-based AI model. If DeepSeek maintains a truly open source approach, it could reduce costs for AI teams and enable faster innovation. However, decentralists face a dilemma. Deepseek may reduce its dependence on US tech giants, but it will bring about a new dependence on China, a country known for its strict government surveillance of AI development.
This raises concerns beyond costs. Will Deepseek be as resistant to censorship as Crypto AI supports hope? Or will restrictions on content and responses stop potential enterprise users? A distributed AI model that limits what it can say still carries elements of centralized control from different permissions.
What’s ahead?
Deepseek represents a major change in AI accessibility, but it is not a silver bullet. Questions remain about training data, performance consistency, and long-term viability. However, early indications suggest that they could be an essential alternative for emerging AI startups, including crypto startups.
The battle for AI decentralization is not over. Blockchain-based AI teams are driving greater autonomy, but they need to balance trade-offs between ideal and actuality. Fully distributed AI is still in its early stages, and it is still not yet seen whether it can really compete with the big technology.
Opinion: Zain Jaffer, co-founder of Vungle
This article is for general informational purposes and is not intended to be considered legal or investment advice, and should not be done. The views, thoughts and opinions expressed here are the authors alone and do not necessarily reflect or express Cointregraph’s views and opinions.
