Author: Zeke, Researcher at YBB Capital

I. AI is Inevitable, but What about Crypto x AI?

At the beginning of 2025, the "nuclear bomb" dropped by DeepSeek, developed by Fantom Quantitative, shocked the AI world. The AI model trained by China using only 2,048 NVIDIA H800 GPUs, at a cost of $5.58 million (about one-tenth of Meta's cost), outperformed GPT-4, Llama 3.1, and others in benchmarks such as MMLU and GPQA, and even slightly surpassed them in complex reasoning and Chinese semantic understanding. Although the US has imposed chip blockades on China for many years, DeepSeek's computing power has significantly broken the US monopoly, and China has embarked on a technological path that suits its national conditions but is not inferior to the world's top models. This open-source, low-cost, and homogenized approach has broken through the US computing defense at an astonishing speed.

For many years, Chinese tech products have lagged in performance and been labeled as cheap and low-end, which is the intuitive impression of many people about Chinese internet companies. But DeepSeek is different. Regardless of whether it surpasses ChatGPT, the fact that it is frequently mentioned in the US political and tech circles indicates the significance of the issue. It is clear that China has now transformed from a technology follower to a challenger, with far-reaching global chain effects.

Although the first to be affected is my wallet, which can be attributed to the misjudgment of the traditional AI development, I still want to share my views, especially on the impact of DeepSeek on the crypto industry.

1. NVIDIA is the biggest loser in this event.

First, the demand for AI computing power is being questioned; second, NVIDIA's unified hardware and software computing architecture, CUDA, has been bypassed. Those familiar with the AI field must know that CUDA is an important cornerstone in driving the development of modern AI. When using NVIDIA GPUs, AI model developers generally rely on CUDA to complete their work. CUDA has a relatively low technical requirement for developers, as it has pre-packaged functions, and developers don't have to worry about too many details, but this is at the cost of execution efficiency.

Since CUDA is a general-purpose programming framework, it limits the flexibility of model training. DeepSeek's solution is to directly use PTX (NVIDIA's intermediate instruction set framework designed for GPUs), bypassing the hardware's limitations on training speed and shortening the training time. Other models require 10 days of training, while DeepSeek only needs 5 days. This also means that if DeepSeek plans to adapt to China's domestic GPUs in the future, the hardware adaptation process will be more seamless, and it has the potential to shake NVIDIA's dominant position in the AI chip market. (This paragraph is excerpted from a report by Korea Future Asset Securities on DeepSeek's training process.)

In addition to the potential decline in NVIDIA's stock price and the severe impact on the crypto market related to the US stock market, I personally believe that in the long run, this is actually beneficial for decentralized computing projects. First, more individual GPUs will be able to contribute their excess computing power. Second, if DeepSeek's compact open-source model approach succeeds, it will force many AI companies to open-source their models, thereby increasing the demand for local deployment and secondary development computing power. Based on the hardware requirements of DeepSeek R1, from a minimum of 1.5B parameters to a maximum of 70B, from NVIDIA GeForce GTX 1660 Super to the 40, 50 series, and even professional-level A100 and H800 GPUs, there is an opportunity for them to contribute their excess computing power. For the currently underutilized decentralized computing projects, this may be a potential breakthrough, provided the latency is low enough.

2. AI Framework Projects: The Hottest Crypto Sector Before the "Nuclear Bomb" of DeepSeek

Before DeepSeek dropped its "nuclear bomb", AI framework projects were the hottest emerging sector in the crypto industry, which was also the last topic I reported on before the Lunar New Year. However, with DeepSeek's breakthrough, most of these projects are now rapidly trending towards zero. After all, DeepSeek has achieved parity with OpenAI for less than $6 million, while our AI framework projects have FDVs in the billions of dollars, but have not yet produced anything that can be truly considered a practical AI agent.

Since the birth of blockchain, the pursuit of tokenization has been almost obsessive. Currently, the crypto industry's tolerance for tokenization is already very high. For AI framework projects that don't even have on-chain assets, they only need an open-source GitHub repository and a social media account to issue tokens. This "token-based on code repository" approach inevitably carries the risk of being completely eliminated by "second-order attacks" from traditional AI companies.

In the golden age of artificial intelligence development, traditional internet companies are unlikely to stop at DeepSeek as the only weapon. Against the backdrop of China-US competition, the development of artificial intelligence will only accelerate, and the key question is how Crypto can integrate with the upstream and downstream of AI, highlighting the advantages of decentralization, without being unexpectedly hit by AOE.

In general, we can divide the Crypto x AI technology stack into four levels: computing power, data, middleware, and applications. In the current layered structure, I don't see the necessity of Crypto's participation. But from a future perspective, privacy and security may be strengths, as AI agents have become a reality to replace or assist human work. Ensuring the privacy of work and personal data processed by AI may be a challenge that traditional internet companies cannot solve. Furthermore, if AI can act as an agent with payment permissions, ensuring the security of wallets will become an issue. Using blockchain as a compliance and audit layer for AI models may be a key direction for future development.

On the other hand, incentive mechanisms also play a crucial role. In addition to stimulating computing power and model sharing, incentive mechanisms can also teach AI how to interact with the virtual world. Unlike LLMs with decades of global internet data, teaching AI the right behavior requires constant human labeling. For example, teaching a vision model to recognize animals and cars is not something that can be outsourced to a group of college students. To create an AI agent that can interact with the virtual world, a large decentralized personal network is needed to guide the AI. This is a potential direction. In my previous articles, I discussed this in more detail. What else can incentive mechanisms drive? Combining with DePin, teaching AI agents to interact with the physical world, incentivizing AI to gain attention, incentivizing AI's secondary creation (Bittensor's incentive model is a good example), or allowing AI to automatically adjust the token incentive mechanism - these are all fascinating possibilities. This raises another question from my previous article: when a decentralized project grows and enters the mainstream, how should deflation and inflation be managed? Through simple code rules, a few project team members, or those key figures? Of course, we also have governance tokens. But before solving the "sybil problem", governance tokens are meaningless. Democratic voting will never be reflected in governance proposals, as a16z only needs a few wallets to veto the votes of a large community, so what's the point of voting?

Of course, we can't possibly gather a group of top AI talents like traditional internet companies, nor can we buy or rent large-scale GPU clusters for training. Trying to replicate DeepSeek in the blockchain is a daydream. Crypto's role is to bring the irreplaceable decentralized features into another field, just as we once brought financial freedom to the world. AI is an inevitable narrative for humanity, but the key question is: what role can Crypto play in it?

3. Wordcoin: The First Discussion on a Crypto Utopia Project

This is the first time I've mentioned Wordcoin in an article. This crypto utopia project initiated by Sam Altman still seems absurd to me. The decision to record irises is like choosing between state surveillance and corporate surveillance, similar to the choice of the red pill and the blue pill in "The Matrix".

However, the idea of a universal basic income or financial inclusion no longer seems to be a joke at this stage. DeepSeek's AI agents can be deployed locally to rival top models, and they have already started appearing in hospitals and government agencies in China. According to McKinsey's 2024 forecast, up to 50% of jobs may be replaced by AI within the next six years. Future versions of Wordcoin may even be distributed by the government. If this trend intensifies, related universal financial tokens may emerge and be repeatedly hyped. If given another five or six years, this may coincide with Trump's presidential term. Will a crypto president issue such tokens? I think it's quite possible.

4. Elon Musk and the Future of AI Research Funding

Given Elon Musk's recent statements, AI may dominate the Nobel Prize within the next 25 years. I believe that blockchain-based fundraising (even contributing computing power, storage, methods, and other resources) to drive AI research will be more interesting and effective than the current decentralized science (DeSci) movement. Perhaps I can call it decentralized artificial intelligence science, or DeAIS for short.

II. Meme Coins are No Longer Memes

Previously, when we discussed MemeCoins, the focus was on subculture, community consensus, and the viral effect. Now, sitting in front of GMGN, I analyze conspiracy groups, major addresses, and insider trading by developers. When CA starts posting in various "shit coin" groups, that's when it begins. Today's meme coins are more absurd than ever before. In the current stage of Pump.fun, don't expect to find a stable coin to sleep on - the chart may have already crashed by the time you come back from the bathroom.

The simplification of asset issuance thresholds and the high anonymity of blockchains have pushed this casino-like culture to the extreme, allowing an anonymous individual to turn the crypto market into an ATM. The evolution of Meme is becoming more and more arbitrary - not just "code library coins", any event, any person, or even any AI can become a coin. Without a cultural core or community consensus constraints, the so-called top projects are forgotten within a few weeks. The celebrity coin craze sparked by Trump lasted only a month, and billions of dollars flowed out of Sol after a single tweet from President Milei, marking the retreat of memes. Milei's response was simple: delete the tweet and reply "I don't know".

The rapid development of AI has stolen too much of the world's attention, the development of technical factions has stagnated, retail investors have abandoned value investing and can only bet on being the lucky ones whose coin prices are pumped, and the increasingly scarce liquidity is being drained over and over again. The constant loss of liquidity reflects the traditional capital and outsiders' disdain for Altcoins.

III. Clinging to the Boat Cannot Retrieve the Sword

The cycle law has obviously become ineffective, and all "petty tricks" cannot reverse the defeat. A BTC bull market does not necessarily mean that Altcoins will also rise, but if BTC falls, Altcoins will definitely follow. Our understanding of Altcoins must be updated - the Altcoin market can no longer be supported by just a white paper, and large projects listed on top-tier CEXs must be mature enough to support their token prices.

Looking back on the token growth over the past seven years, the number of tokens traded on major exchanges was less than 2,000 in 2017, but by 2024 the number of tokens had approached 25,000 (data from CoinGecko, including delisted tokens). The exponential expansion of token numbers is essentially an irreversible evolution of the blockchain low-entropy value system towards a high-entropy noise system. While each token in 2017 still carried the ideal of "disrupting the world", by 2024 tokens had evolved into liquidity exit chips. The emergence of more tokens has not brought more innovation or real-world applications, but the valuations of star projects have greatly increased the demand for market liquidity.

As mentioned earlier, without external recognition, retail investors cannot support the valuations of these projects. Most Altcoin projects reach their peak market prices at the time of listing, with Binance becoming their final destination. The crypto world needs a revolution. Star projects should prove the rationality of their massive fundraising. Bybit's trial of publishing project financial information may be a potential solution. But in my personal view, the market needs a deep bear market to reshape the valuation system and listing standards of Altcoins.

IV. Chaos

I once saw a glimmer of hope in Ton, thinking that the beginning of crypto's consumer-level applications had arrived. But this fleeting light faded as the Tap to Earn craze subsided. Five years ago, the liquidity mining catalyzed by DeFi pushed the crypto world to an unprecedented peak. Five years have passed, and the only relatively successful area is DeFi.

Now when I communicate with industry insiders, the conversation is very simple: "Did you buy BTC? Did you short? Do you have CA?" Everyone is very confused and can't find a direction. Nowadays, holding anything other than BTC makes people lose sleep. "Diamond hands" is no longer a complimentary term, and not holding BTC is more like being a fool.

Browsing the various blockchain media apps on my phone, it feels like reading the New York Times and tabloid gossip. These phenomena reflect that the hopes of this circle are now mostly pinned on policies and attention. From the VC's perspective, our future may lie in investing only in utility products, while asset issuance platforms can only survive by becoming sellers of shovels and rent collectors.

Conclusion

Clearly, this is not the reality we hope to see. Although the current state of cryptocurrencies seems to be lost in a chaotic fog, the success of DeepSeek proves that innovation and reform are still the most effective ways to break the stagnation. Cryptocurrencies are currently in the best position in history, with favorable policies, attention, capital, and a solid infrastructure. In the near future, many Altcoin ETFs may inject new liquidity into the market.

We are obviously in the mainstream, yet trapped in our own walled city.

The retreat of Meme Coins may be a turning point, and the future of humanity may not just revolve around AI.