Is DeepSeek a World-Changing Event, or is Everyone Just Hyped up on Cocaine/Adderall?

First, really sorry that I missed yesterday. This article is 1400+ words as payment for my missing a really important day ;). I try to post every three days, and I only miss on that cycle when something personally important happens. Yesterday, that important thing was a math quiz I had to study for, one that I am pretty sure I bombed today >:(, and financial aid stuff for college yesterday blew up.

Speaking of blowing up, bombing, and math, my portfolio, and specifically my ASML position has imploded over the past few days. It has imploded because of the “Chinese Chat-GPT,” “The 21st Century Sputnik,” take your pick of dramatic moniker. DeepSeek AI is a big deal. But what is it actually? In case you have not read the literal thousands of articles poured out over the past two days, here is what DeepSeek is.

DeepSeek is basically Chat-GPT, but it possesses a few key differences. First, DeepSeek was supposedly (and we’ll come back to this) made for only $6 million dollars and with pretty bad Nvidia chips, not the new ones. Second, DeepSeek is open source. That is a big deal. Open source AI is a pretty revolutionary idea. Open source means that the AI’s training data, source code, models, everything, is free and open to the public. I find that to be an extremely interesting concept because it is just so… anti-capitalist.

Imagine taking out a mortgage on your house, quitting your job, and spending 60 hours a week for a decade to make a machine, the first of its kind. This machine can make, I don’t know, cement better than any other machine or company has ever been able to. Then, once you finally get everything to work perfectly, you print out detailed instructions on how to build this machine, how to work it, and how to improve it. That is open source AI. You are giving everyone the keys to the castle. Literally, they can download, for free, your AI model and operate it, manipulate it, and make it better. It is so inherently anti-competitive that it boggles the mind.

Now, back to DeepSeek. The company is wholly owned by a Chinese hedge fund High-Flyer whose CEO, Liang Wenfeng, runs the DeepSeek show in a founder/CEO capacity. DeepSeek was founded 13 months ago, and has already caught all the way back up with American tech giants. I don’t think I mentioned this, but DeepSeek was equal to or better than the latest Chat-GPT models in every capacity.

First, a lot of this smells quite fishy. Before I elaborate on why I find this a tad iffy, just know that I am not/cannot trash the final product. The DeepSeek AI is what it is. The open source code actually helps to confirm that this is a legitimate thing, no hoax. However, the backstory for this company does have a few holes. First, there is simply no way that these guys created this thing this quickly. DeepSeek is a complete Chinese front. The Chinese government is pumping this company like a hot air balloon, filling it to the brim with more tech and financing than they can probably handle. This company is likely the result of a decade, or more, of work from the CCP on AI.

Also, DeepSeek claimed that they did all this with $6 million dollars (total BS, COME ON) and a few thousand of Nvidia’s older, less efficient H100 chips. It is absolutely 100 and a million percent IMPOSSIBLE that they did this so cheaply and in so little time. What is more likely? Microsoft, Meta, Amazon, the US government, literally everybody made an oversight on such a great magnitude that they invested hundreds and hundreds of billions of dollars too many, when they could have just basically used an abacus and a computer from 1990 to make these AI models. This isn’t like The Big Short, or anything like that, where everybody has made a massive oversight out of greed. If Microsoft or OpenAI could make their models without dropping billions, they would have figured that out by now, obviously. I guarantee that DeepSeek used at least 50,000 cutting edge chips from Nvidia, or somebody similar, but that they are lying (basically), so that that guy sitting in the White House doesn’t declare a 300% tariff on China.

It was probably cheaper, but I want to see this improve and keep pace with American AI companies before I will truly be impressed.

Now, what implications does this have for the market? Because the market seemed quite worried. Usually, I espouse my beliefs with immense confidence, and some would say, unearned total certainty in myself and beliefs. This is a highly complicated matter, and the range of possible outcomes is varied, espescially because I am having to make mental leaps based on fundamental information that I am receiving as an input. It’s already so shaky, that anything I assert is being metaphorically built on toothpicks.

THAT SAID, I am still optimistic about these AI-related companies, and about the market in general. I am not, and I don’t think I ever really have been, one of those perma-bulls that believe that Nvidia has been ordained by God himself to break a $5 trillion market cap. I do think though, that this market is super unstable, and that people are supery jittery and looking to be the first one out of this bubble. The thing is though, that because everyone is saying there is a bubble, I become less and less certain each day about the bubble nature of our market. Overvalued, yes, over concentrated, yes, and a little overly optimistic, probably, but I would not call this a bubble. Investors aren’t piling into the market with wild, reckless abandon. I feel like everyone is super nervous about moving money around in this market.

But I digress. I think the companies that are selling the picks and hammers, Nvidia, ASML, based on this AI hype etc… aren’t necessarily hurt by this development, but the field has shifted, certainly. The top-level guys, the hyperscalers, they will still need the most cutting edge chips, and I am still pretty sure that DeepSeek used a few of these higher level chips, and certainly far more of them than they are claiming. If AI becomes more accessible, great! More people using more AI drives demand up for the product.

If I was an OpenAI investor, I would be pretty displeased, because this is real and genuine competition, competition that seems (again, just a guess from me) to be entirely subsidized by the CCP. But I’m not an OpenAI investor, I’m a Microsoft investor. And Microsoft is certainly not hurt by this development, just to use a company I own as an example. Microsoft is making AI to make Word more efficient, CoPilot more efficient, and allow their users to do entire Excel spreadsheets in seconds. LLM (Chat-GPT, DeepSense) does NOT equal AI. Microsoft is trying to make functional AI. These other companies are operating in different fields.

Also, if hundreds of millions, potentially billions of people, have more access to AI than before, with it spreading like a virus due to its open source nature, then these companies are going to need MORE COMPUTE. Microsoft has a ton of compute. In fact, that $80 billy they are actively dropping in the coming year alone is centered primarily around building out compute. DeepSense kept failing because they were missing…guess what….COMPUTE.

In short, this is a big deal, for sure. The playing field has been altered by this development, and I would imagine that the CCP is pretty pleased with the $1 trillion that evaporated from US markets over their Chat-GPT knockoff. HOWEVER, the companies that were set to win from AI before: Apple, Amazon, and Microsoft, are still 100% set.

Companies like ASML and Nvidia, the companies that are the manufacturing process…we’ll see. But I think that they will not have problems finding customers at all. I don’t think Microsoft is going to go to their old chips, but I do think that the field of play has expanded, so smaller and mid-level players can get in on their own; if that makes sense?

Okay, that was over 1400 words. I don’t care enough, and I am too tired to read back over all of that, so if you see mistakes, let a slime know. Also, fuck you David Tepper. This event does not mean that China is a good investment. Good luck to everyone except for David Tepper.