Did Chinese AI just Ruin America's Economy?

Hello investors,

This week produced shockwaves in the world of artificial intelligence all thanks to Chinese AI firm DeepSeek. 

Here's the Borex TL;DR on DeepSeek: 

• Chinese firm DeepSeek just created an open source AI model that outperforms ChatGPT, Meta's and Google's models
• The total cost to develop this was under $6 million USD
• DeepSeek's tokens are 97% cheaper than OpenAI's
• They did this without Nvidia chips (allegedly) as China has limited access to these cutting edge chips

This is a lot to unwind. To draw a parallel to real life, this is essentially a company releasing a phone on par with the latest and greatest iPhone, but selling it for $30 instead of $1000. It's a dramatic turn of events I was not expecting. 

Is DeepSeek a better AI model?

I'm a heavy ChatGPT user. I subscribe to ChatGPT Plus and use it to run and analyze financial models, diagnose problems, perform research, create powerpoint decks, draft emails, summarize documents, literally most tasks which I used to have to outsource. For the past 3 days I have been exclusively using  DeepSeek and... I'm impressed with it. 

But it's not just me, I'm reading that DeepSeek has rapidly become the preferred tool for researchers at respected institutions like Stanford and MIT, essentially becoming their model of choice virtually overnight. 

Artificial Analysis ran some benchmarking tests against other models, and... it was just as good as OpenAI's godtier, extremely expensive and limited o1 model. It outperformed every other OpenAI, Google, Meta, and Amazon AI model. Impressive. 

What's more, they're releasing it open-source so you even have the option (which OpenAI doesn't offer btw) of not using their API at all and running the model for "free" yourself. Essentially what they have done is reveal the "secret sauce" behind these LLM. 

Why are they doing this? Well apparently DeepSeek is the side project of one of China's top hedge funds. And well, if there's any stereotype on Wall Street that's mostly true, it's probably that Chinese finance guys are really good at math. So I'm sure they ran the numbers and still see opportunities to make money here somehow. Or maybe they can quantify the long term effects of the social impact here for China and see worth there.

Token Cost

Okay, on to the money side. You have to pay to use OpenAI models with tokens if you want to incorporate them into your business to perform tasks.  

A typical enterprise codebase might be 1 million lines of code, or roughly 4 million tokens. That would cost $60 with OpenAI versus just $2.20 with DeepSeek. At OpenAI's prices, doing daily security scans would cost $21,900 per year per codebase; with DeepSeek it's $803.

So if you're an OpenAI customer today you're obviously going to start asking yourself some questions, like "why exactly should I be paying 30x more for a worse product?". This is pretty transformational stuff, it fundamentally challenges the economics of the market. And it's great for China and bad for US tech giants. 

What does this mean for the markets?

DeepSeek reportedly developed their entire platform for less than $6 million (not billion), a stark contrast to Microsoft's $14 billion investment in OpenAI and the recent announcement by President Trump of a $500 billion AI initiative. This disparity raises questions about whether the costs of AI development are being overstated in the West, potentially inflating Nasdaq valuations.

This situation could pose a significant risk to U.S. equity markets, particularly given that DeepSeek's success was achieved without the use of Nvidia chips. Previously deemed essential for AI progress, the necessity of these chips was a key driver behind Nvidia's stock surge last year. The efficiency and cost-effectiveness of DeepSeek's model challenge the justification for the extensive capital expenditures currently flowing into the AI sector. 

Or does it?

The success of DeepSeek could potentially challenge U.S. dominance in technology, positioning China as a frontrunner in the global AI landscape.

However, there's room for skepticism regarding DeepSeek's claims. What if the company isn't being fully transparent about its operations? There's a possibility that DeepSeek might actually be using Nvidia chips, acquired through indirect means from a third-party country like Singapore. This scenario would question the validity of their reported low development costs and the supposed independence from established tech infrastructures, adding a layer of complexity to the narrative surrounding global AI development and its geopolitical implications.

If you're inclined to believe China's newfound innovation capabilities as demonstrated by DeepSeek, it might be wise to reconsider your investment strategy—perhaps by divesting from Nvidia and directing your funds into CQQQ, betting on China's ascendency in AI. However, if you harbor any doubts about the authenticity of these claims, a more prudent approach would be to wait out the initial flurry of reactions. Allow the situation to stabilize before making any significant changes to your QQQ-focused investment strategy.

For those interested in exploring this topic further, below is a CNBC video that delves into the matter, providing additional insights. Enjoy!