DeepSeek’s Impact on Crypto - What You Need to Know
AI Mindshare Surge, Market Outlook, & More
The Chinese calendar year is ending with a bang, with DeepSeek making its presence known. We covered DeepSeek’s impact on the AI space in last week’s Ailea Bulletin, but the havoc that this news has wreaked on markets before the week even begins warrants some follow-up coverage. US markets haven’t even opened, yet MAG7 names are down big, with $BTC falling below $100K and the alts seeing red across the board.
This is another one of those cases where an event that shouldn’t necessarily have any immediate impact on crypto, actually does due to crypto’s correlation with risk assets. Specifically, it’s worth underlining that crypto is correlated with US risk assets. Chinese stocks are actually up today, while US Big Tech names are down significantly in the pre-market, something that doesn’t happen very often at all.
In today’s edition, we’ll discuss the recent DeepSeek AI breakthroughs, how markets are reacting, and what it means for crypto in the short to mid-term…
Stay informed, stay alert ⬇
Background on DeepSeek R1
R1, the open-source reasoning model from DeepSeek, was released last week, but has taken some time for markets and the general public to take note of. Now, DeepSeek has outpaced ChatGPT for the #1 spot on the appstore.
Notably, larger Chinese conglomerates involved in AI also reacted to DeepSeek. ByteDance’s Doubao, and Alibaba’s Qwen models have seen upgrades in their own right recently, feeling the brunt of domestic competition. But of course, the strongest reaction from the english-speaking media is sparked by Western AI companies reactions to DeepSeek. OpenAI released their Operator agents demo. If not for DeepSeek, this probably would’ve gotten a good amount of attention, but the launch was indeed spoiled by R1.
Perplexity’s CEO, Aravind Srinivas, gave his own insights in an interview, congratulating the DeepSeek team while also exclaiming that there is now no excuse for smaller AI labs based in the US or even other countries to compete in model training, as most have access to the same level of resources that DeepSeek does.
In the short-term, the advent of R1 calls into question a lot of the massive infrastructure investments that have recently been announced, including the $500B Stargate project, or Meta’s own $60B private investment. Nvidia has fallen a lot in pre-market hours; something like R1 might act as a sort of needle to the balloon that is $NVDA.
However, cost-cutting for model training may just see demand for chips shift to the inference side, which is ultimately where the actual applications that can impact people’s lives take place. How long this takes remains to be seen, and in the meantime, markets might lean toward uncertainty, waiting for long-term demand for infrastructure to be established rather than assuming that this is the case.
Impact on Crypto
The impact that this power shift in AI has had on crypto markets has been immediately negative, dragging down $BTC and other tokens along with US stocks. But in the medium to long-term, these developments should prove to be a good thing, especially for the AI agents sector, which is what has come to dominate crypto x AI as we know it. AI agents in their current state are mostly dependent on the output and models from large LLM labs; significant breakthroughs from these labs only seek to enhance the capabilities of these agents.
Hey Anon, and the AI16Z framework, among various others, have both previously had DeepSeek integrated, prior to R1. If inference is to become cheaper, as it already has, then small teams behind these agents can hopefully put out more engaging products. Revelations in model training can lower costs and perhaps lower infrastructure investment, shifting more focus to the application of AI.
As mentioned above, there is also a growing emphasis on agents as the actual means of applying AI capabilities into everyday life. Crypto mindshare is now flowing back into AI, firmly crossing the 40% threshold, which hadn’t been breached since the launch of $TRUMP. This is a rare instance of mindshare flowing to a sector, while it is still deeply negative price-wise.
Of all the areas of tech it can be argued that AI agents in crypto, or just AI agents independent of any specific LLM, stand to gain the most from this news. Demand for chips and data centers may suffer for some time, as markets adapt and come up with new valuation models that take into account the fact that previous demand may change. This infrastructure is largely on the input side of this equation, being utilized to train models, while Agents can be seen as representing the actual demand for these models.
Become a Premium member to unlock all our research & reports including access to our members-only discord server.
Join thousands of sharp crypto investors & traders by becoming a Premium Member & gain an edge in the markets. For just $129/month, you can access our full suite of offerings:
Gain instant access to Deep Dives, Blueprints, and Perspectives.
Priority access to new features and exclusive content.
Ideal for investors who demand comprehensive insights.
Full access to historical research archive and analytics tools.