How does DeepSeek rewrite the AI ​​landscape? Is decentralized AI a breakdown?

Source: Grayscale; Translated by Deng Tong, Golden Finance

summary

• The launch of the DeepSeek artificial intelligence model is known as the "Sputnik Moment", highlighting the international competition for artificial intelligence hegemony and the power of open source technology.

• At the same time, the news also highlights the risks associated with centralized AI development, such as data security, bias and censorship. The risks associated with centralized AI companies can be solved through blockchain-based AI platforms such as Bittensor.

• Bittensor helps promote the development of open and global AI systems by using decentralized networks and economic incentives. By leveraging blockchain technology and a global network of participants, platforms like Bittensor can increase transparency, democratize access and allocate ownership of AI systems.

• Grayscale Research believes that the emergence of DeepSeek may reduce the cost and entry barriers for open source decentralized artificial intelligence projects.

what happened?

Recently, China-based startup DeepSeek launched an open source artificial intelligence (AI) model that performs comparable to or exceeds the latter, such as OpenAI’s leading models such as O1. [1] Impressively, DeepSeek achieved this with significantly fewer computing resources, and it reportedly spent only about $5 million to train the model – just hundreds of millions of dollars on OpenAI to train a small part of it. [2] By January 27, DeepSeek surpassed OpenAI's ChatGPT in the Apple App Store rankings. [3]

Tech leaders call it the "Sputnik moment" of artificial intelligence -- we may be witnessing the modern space race between China and the United States in the field of artificial intelligence. [4] The emergence of DeepSeek has led to a historic sell-off in tech stocks, with the market value of companies such as Nvidia and Microsoft evaporating billions of dollars as it forces investors to reconsider previous assumptions about this powerful emerging technology. [5]

However, while DeepSeek 's breakthroughs demonstrate the power of open source AI, they have also attracted more attention to the centralized control and development risks of AI technology. DeepSeek suffered a massive cyberattack after news of the company's performance of new models was announced, prompting the company to temporarily restrict user registration. [6] This incident highlights the inherent vulnerabilities of centralized systems—including the risk of cyber attacks that may disrupt services. Distributed systems can enhance network resilience by distributing responsibilities across multiple entities. The decentralized development of AI models may also help reduce bias and increase transparency of this key technology.

In this article, we will explore these risks associated with AI development and provide detailed introduction to how decentralized AI platforms such as Bittensor address these risks. We will also explore Bittensor’s progress to date and the potential impact of DeepSeek on the wider development of decentralized AI.

Risks of centralized AI

Network effects and intensive capital demands have made it difficult for many AI developers outside of large tech companies, such as small companies or academic researchers, to obtain the resources required for AI development or to monetize their work. This may limit overall AI competition and innovation.

Therefore, the influence on this key technology is mainly concentrated in the hands of a few tech giants, raising serious issues about scrutiny and bias. For example, in February 2024, Google’s AI image generator Gemini exposed racial bias and historical errors, explaining how companies manipulate their models. [7] It is worth noting that these concerns extend to DeepSeek.

This raises broader questions about AI governance. A few control the companies that develop a few models that may increasingly shape and influence society. As AI’s influence and importance continues to grow, many people are worried that a company or government may have the power to make decisions on AI models that have a huge impact on society, may set up guardrails, operate behind closed doors, or manipulate models to seek. Benefits – but at the expense of other members of society.

How do we make sure we can trust the models we use for our data? Lack of true transparency – and so risky – how do we believe that these innovative technologies are built in our best interests rather than at the expense of us?

Decentralized AI and Bittensor

Entering decentralized AI: Potential solutions to these challenges. By leveraging blockchain technology and a global network of participants, platforms like Bittensor can increase transparency, democratize access and allocate ownership of AI systems.

Grayscale Research believes that decentralized AI has the potential to bring important decisions about AI development from closed gardens to public ownership. We believe Bittensor provides a compelling solution as a critical decentralized AI platform that is ready to help address these risks and provide viable alternatives to centralized AI incumbents.

What is Bittensor

Bittensor is a platform that leverages decentralized networks and economic incentives to help promote the development of open and global AI systems. It aims to create an "artificial intelligence internet" where the interconnected ecosystem is called "subnets", each focusing on different specific use cases. Currently, Bittensor has more than 50 subnets covering a wide range of applications and use cases, including video generation, AI proxying, and deep forgery detection. Here is how Bittensor tries to solve problems related to centralized AI:

• Coordinated economic incentives: Centralized AI companies prioritize shareholder value and profits, which often lead to the extraction of value from users. By contrast, by using its TAO tokens, Bittensor coordinates incentives among ecosystem participants, including their users and token holders.

• Build and use AI without permission: Many centralized AI platforms often have high barriers to entry for developers. Additionally, as AI becomes more and more powerful, there may be more and more restrictions on who can build or access these applications. Bittensor provides an alternative to access resources to develop and use AI without permission.

• Open Source Monetization: While open source AI models like DeepSeek's R1 and Meta's Llama provide benefits, open source AI still faces great difficulties in monetization and coordination. Bittensor helped solve this problem through the TAO token issuance, allowing AI developers to monetize and fund their work.

We believe that Bittensor’s tokens (TAO) currently has particularly compelling investment value for the following reasons:

• It is possible to solve the above-mentioned problems related to centralized AI.

• Progress has been made to attract ecosystem investors such as Yuma and subnet builders such as Masa (data crawling and AI agent arena), Dippy (AI role-playing app), and Kaito (decentralized search).

• Dynamic TAO (“dTAO”) upgrade is scheduled for February [9], which will make it possible to invest in a single subnet; we believe this could inject a new wave of liquidity into the Bittensor ecosystem.

A broader decentralized AI landscape

Just recently, some might think that open source AI always lags behind the best closed source models offered by tech giants. DeepSeek shows that this is not necessarily the case in the future; critical AI innovations don’t need to be done in silos, nor do they need to trickle down from top to bottom.

Grayscale Research expects a wide variety of decentralized AI assets to benefit. With the learning and application of efficiency improvements, the development of DeepSeek may stimulate widespread improvements in decentralized AI. Accessing DeepSeek 's high-performance open source model can reduce costs and lower entry barriers to many open source decentralized AI projects, especially at the application layer. [10]

We've seen this happen. For example, the decentralized AI Agent launchpad ai16z already allows the DeepSeek model to be accessed using the agents built by its ELIZA framework. [11] On January 27, Venice.ai launched the token, a decentralized application that allows access to the DeepSeek model on the user's local device while retaining the user's data privacy. The token is valued at more than $1 billion within two hours of its launch. [12]

in conclusion

As developments like DeepSeek continue to shape the AI ​​landscape, international competition for technological hegemony and society, Grayscale Research believes that we must adopt decentralized solutions that address centralized risks. By leveraging such platforms, we can potentially prevent monopoly control and create a safer future for AI.

Comments

[1] “How China’s top artificial intelligence models overcome U.S. sanctions.” MIT Technology Review. January 24, 2025

[2] “How DeepSeek’s artificial intelligence competes with OpenAI’s model.” Wall Street Journal. January 28, 2025

[3] “China’s DeepSeek AI beats ChatGPT in the app store: This is what you should know.” CNBC. January 27.

[4] "China's DeepSeek artificial intelligence shocks the industry and weakens the momentum of the United States." BBC. January 28, 2025.

[5] “China’s DeepSeek AI shocks the industry and weakens the momentum of the United States.” BBC. January 28, 2025.

[6] "After artificial intelligence chatbots top the app store, DeepSeek encountered a 'mass' cyber attack." The Guardian. January 27, 2025.

[7] “Gemini apologized for the 'missing' after he spawned a racially diverse Nazi.” The Verge. February 21, 2024.

[8] The Independent. January 28, 2025.

[9] X.com

[10] “What’s better about DeepSeek than ChatGPT: Cost comparison.” Creole Studios. January 28, 2025.

[11] “Will DeepSeek trigger a major reshuffle in the AI ​​agent field? Is it time to buy or retreat on dips?” AICoin.com. January 27, 2025.

[12] “The total value of Venice AI tokens that allow private access to DeepSeek is $1.6 billion.” TradingView. January 27, 2025.

[13] Shareholding situation is subject to change without notice.