DeepSeek is Paving the Way for Web3 AI

From veradiverdict by Paul Veradittakit

DeepSeek started as a side project from the hedge fund High-Flyer, boot-strapped by a few thousand Nvidia GPUs. They shocked the world when DeepSeek R1, which took  over a billion  dollars, 2,000 Nvidia H800  GPUs, and over 55 days, beat benchmarks held by OpenAI’s o1 model which required hundreds of billions of dollars to develop along with over  16,000 advanced GPUs .

DeepSeek R1 has 671 billion parameters  while GPT-4 has  1.76 trillion parameters . OpenAI’s large models likely require thousands of GPUs for training and high-end clusters for inference while DeepSeek 7B & 67B can be run on consumer-grade hardware (a few A100 or H100).

DeepSeek-R1 Versus the Competition

As a response, Nvidia’s stock price dropped 18% (loss of  $600 billion  in market value). The idea that AI models must be closed-source and have loss-leading computational costs to succeed is crumbling.

The Existing Decentralized AI Narrative

Demystifying the Crypto x AI stack

AI x Crypto projects believed that crowdsourced, public, decentralized AI would eventually create better models than their centralized counterparts.

This had thus far not been true, as the highest-performing models had come from closed-source companies like OpenAI and Anthropic. Crypto x AI companies have adapted to this by specializing in infrastructure rather than model-building.

For example, GPU marketplaces like Akash, Render, IoNet, and Exabits have developed sustainable revenues. Companies that allow users to share their network bandwidth like Grass and Gradient have found their niche in supplying services, like distributed web scraping, to web2 clients. Storage networks like Arweave, Filecoin, and Ocean have also done well by being the platforms on which these projects are built. Supply networks have flourished because of their ability to tailor their cheaper and more scalable services to off-chain customers.

Messari Report on DePin and AI x Crypto

Now that GPU and financial resources are no longer limitations to creating quality AI models, web3 AI companies can focus on replicating DeepSeek’s effectiveness while offering new benefits like modality, user ownership, censorship resistance, privacy, and more.

Pantera has funded companies like SaharaAI and Sentient that believe they could match or exceed the performance of traditional AI companies while being competitive by offering other services.

Sahara AI, for example, is building a platform where anyone can monetize AI models, data sets, and applications in a collaborative space. Users can permissionlessly train models manually, provide training data, and create tailored AI models with no-code tools. They are only able to cater to all these stakeholders (AI developers, users, resource providers) because everything is tied to their native Sahara blockchain. I wrote more about why we invested in them in a previous blog post  here .

The Future of AI will be built with Web3 Infrastructure

DeepSeek has shown that performant AI models do not have to be built closed source with loss-leading computational costs. I am excited to see AI models developed with the full web3 AI stack in the coming year.

I believe that supply-side projects will continue to grow, while consumer-facing projects can begin competing with web2 competitors by taking advantage of their ability to build networks that invite community involvement. For example, users willingly sign up to train web3 models and actively experiment with them through hackathons and grant programs. Sahara AI and Sentient have begun setting up systems for users to train models based on the users’ expertise. These platforms will allow users to pick and choose the data and integrations to whatever they are applying the model towards. Sahara already has over  780,000 users  on their waitlist while Sentient has over  1 million interactions .

In the near future, I believe that the most performant AI models will be built on-chain.

- Paul Veradittakit