Is DeepSeek 'Cheap AI' Really a Stock Market Black Swan Event?
Introduction of the Chinese AI advancement has sent U.S. stocks tumbling but can it really tip the apple cart?
You're reading 0 of 1 free page.
Register to read more or Unlock Pro — 50% Off Ends Soon
At the start of the year, we are always asked about “Black Swan” events.
Given Academy Securities’ Geopolitical Intelligence Group, which I have the pleasure to work with, the question comes up even more frequently given all the “hot spots” around the globe (for the record, I expect more opportunities than risk from Russia/Ukraine and Iran/Israel).
Less discussed are so-called “Grey Rhino” events. Highly-probable, high-impact, yet neglected threats.
There have been a lot of questions around current valuations, especially in the AI space. I’ve been part of that crowd. Have the valuations gotten ahead of themselves? Are the use cases really so strong to justify the spending and the build out? Does the data center build-out have similarities to the fiber build-out?
What I didn’t foresee was the potential for “cheap AI” upsetting the apple cart.
My thoughts on DeepSeek coming into the week were the following (I probably need to ratchet up my concern):
It seems like this AI model may have been out there for some time, but it exploded in my social media timeline this weekend. On the face of it, we have an AI tool that is cheaper to build and possibly better than existing AI platforms. The “catch” is that it is a Chinese-developed AI engine. Given privacy concerns and cyber threats, I cannot help but wonder who would (or should) use it.
That is assuming it is real. Periodically, we used to get stories about some group achieving “cold fusion” that never turned out to be real, and lately, some similar claims about quantum computing have yet to pan out in reality.
In any case, if extremely good AI can be built with “old school” chips, it would have a lot of ramifications for this market. It does seem unbelievable in some ways, but reminds us of the discussion we’ve been having about China’s efforts to develop chips of their own.
- The smallest chips require state-of-the-art tech manufacturing to be built efficiently. However, small chips can be made inefficiently using old tech. It isn’t an efficient way to make the thinnest chips, but it can be accomplished to a degree, which is presumably how China is making some of its smallest chips (assuming it hasn't figured out how to get access to state-of-the-art tech that the U.S. and others are trying to prevent China from obtaining).
- The “packaging” (vertically stacking multiple chips) may play a bigger role in semiconductors continuing to adhere to Moore’s law (or some variation of it) as opposed to just creating smaller and smaller chips. While packaging is also challenging, it is not necessarily “state of the art” challenging, which would reduce “our” lead over Chinese chipmakers.
Given the importance of AI in our markets (if not yet in our economy), this “story” could be the dominant theme and begs the question: How much good news is priced into this market already?
Add in the usual suspects on the market structure side:
- 0DTE and weekly options on some of the biggest tech names as well as the big indices could amplify risks.
- The leveraged ETFs could amplify risks as well (TQQQ, three-times leveraged QQQ has $26 billion in market cap, representing over $75 billion of risk). SQQQ (which I own, as well as being outright short QQQ) is only $2 billion. The leveraged ETFs on single stocks will also likely come into play
- Bitcoin has dropped below $100,000 and that now has a fairly direct connection into the Nasdaq 100 via MSTR (and MSTX, which is one of those leveraged single stock ETFs).
- The concentrated nature of today’s indices (upwards of 50% of the market value in the indices is concentrated into a small proportion of the total names in the index). It makes “index investing” somewhat the equivalent of “momentum investing” (rather than the traditional concept of “passive”) which could also get nasty.
- The sheer dollar amount dedicated to selling covered calls and open puts is astounding and may also be problematic.
For all the fear, I am covering some shorts this morning: It is a big move and many are sure to question the validity of the claims being made, but make no mistake, if true, this could upset the proverbial apple cart (I do also own some KWEB and FXI, which may do well if the stories are true even though DeepSeek is not a public company, at least not that I have found).
Also, President Trump “suddenly” slapping tariffs on Colombia doesn’t help things either, but that story, which would normally be “top of the fold” will be relegated to the back pages as investors try and figure out what is going on with AI.
At the time of publication, Tchir was long SQQQ, KWEB and FXI, and short QQQ.
