.png)
Markets are reacting to DeepSeek’s impressive debut, but does this signal trouble for AI leaders or opportunity for growth? Here’s the breakdown.
Deep sneak. Seek, and ye’ JUST MIGHT find that competition is actually healthy and should be welcomed in fertile markets. In fact, the surest sign that it is time to get out of trade is if there is no competition. Here is something fascinating for you. By last Friday evening I was already planning to write about competition this morning. Interestingly, this weekend’s launch of Chinese upstart AI app DeepSeek fell right into my lap, as markets fell into a bit of an overnight tumult in response. Thank you DeepSeek for the help on this tough conversation.
If you don’t already know, markets got a bit bloodied this early morning, dragged down by none other than the founder of last year's market gains feast. That’s right, tech, but more specifically, everything AI. Why? Because Chinese upstart DeepSeek launched its AI app this weekend and Silicon Valley insiders were kind of impressed with its showing.
Ok, so here is where is starts to get interesting, or weird, depending on how you approach things. Here is the logic. DeepSeek did something really impressive with far fewer employees and funding than OpenAI did, and in less time. Okay, so what does that mean? Well, if you want to make a knee-jerk assessment, you can assume that all the investment made by the AI heavyweights in the US was all for waste. If you assume that DeepSeek found a way to come up with a comparable AI solution to OpenAI (and by the way, it is not the only company in the US in the space) without needing the help of powerful hyperscaler infrastructure, well then, we have ourselves a conundrum.
And here is the challenge. Will the world need all those shiny, powerful, parallel processors by NVIDIA? How about the massive compute power clouds maintained by Microsoft, Amazon, Alphabet, and IBM? Does it mean that those AI-optimized servers made by SuperMicro are now obsolete? And mystery, shadow warrior Palantir; will its competence in using artificial intelligence to solve problems now become obsolete, as well? Will Meta’s Friday announcement that it will invest north of $60+ billion on AI, now be considered a waste?
I am not going to answer those questions. That is your homework assignment. I will, however, spend a second and tell you what we know about DeepSeek. IT DOESN’T COMPETE WITH OPENAI. OpenAI (and Google’s Gemini) is a broad, general-purpose tool, based on a vast corpus of information. It can be used to create more narrowly focused applications as well. OpenAI has vast capabilities in natural language processing, while DeepSeek is created to be task specific. Now, that doesn’t mean that DeepSeek is not good. What is important to understand is that it is not the same thing as OpenAI, so it would logically take less resources.
Let’s take a step back and think about what makes any AI useful. AI takes a vast amount of information, far more than you or I could ever consume in 100 lifetimes, learns it, understands it, and then shares that learning with us in natural language. Now, it is not just memorization. AI must understand and reason with that vast body of knowledge. It is pretty much the same process as we humans go through, but just on a much larger scale. To do this, as you might imagine, it takes lots of computing power. Though, I have not seen any data on this, I am sure that there is a mathematical relationship between computing demand and vastness, or power of the AI engine.
Vast computing power implies… um, lots of computers! The more powerful, the better. Their power comes from the processors within, and NVIDIA has, by plan or happenstance (it doesn’t matter), the best available chips to get the job done. Those computers have to be attached, secure, their data needs to be stored, and those data centers need power, as in electricity. You can’t have AI without that ecosystem. Go ahead and read that again! No matter what AI tool you are using, you need processing power for the “I” in AI.
Now a quick sidetrack on price seeking, or value seeking. Most of us, unless you haven’t gotten the memo yet, can understand how large the opportunity for AI is in the future, far beyond even the next five years. It’s big and it’s obvious how large its impact will be on all industries, not just high tech! Do you know that Deere and Company uses AI to help farmers enhance productivity using its construction and farming machinery? Think autonomous tractors, smart sprayers, and predictive analytics. Yes, those can help increase sustainable crop yields! Friends, AI will permeate every industry over time and enhance them. Similar to the internet, or the use of Word Processors and Spreadsheets. I know, I know, you get the picture. But how big is the opportunity… in dollars? Well, it’s hard to tell exactly, which is why those AI heavyweight stocks bounce around so much. Each bit of information that we get helps the market better understand what that elusive, actual number is. Now, listen to this carefully. IT WORKS IN BOTH DIRECTIONS. We just have to decide whether we overshot the mark, or if we undershot it. This is the market clearing all available information and attempting to realize it in the prices of those companies that we love to talk about. This is completely normal and healthy.
Now, back to this morning’s challenge. Does this information that we got overnight from DeepSeek vastly change what our understanding is of the opportunity for AI in the future? Now, breath. What if, IF, DeepSeek has found a way to crunch the same information as, say OpenAI, with less resources? What if? Would we be satisfied with simply using less computing power to get the same performance? Think about it. If you can get X performance from OpenAI and the same X performance from DeepSeek with less servers, shouldn't you want to get pX performance (where p is some multiplier of greater performance) by using DeepSeek’s algorithm on the more powerful infrastructure? OF COURSE YOU WOULD! DeepSeek’s as yet untested performance increase over today’s state-of-art AI WOULD ONLY MAKE THE AI OPPORTUNITY BIGGER IN THE FUTURE.
So, who or what was challenged by DeepSeek’s outing this weekend? Was it the hyperscalers, data security companies, network equipment makers, chipmakers, IC design software providers, AI users, etc? No, on notice should be LLM (large language model) AI models like OpenAI, Google DeepMind, Anthropic (Claude), and maybe Meta’s LLaMA. Now let’s remember that these are software companies with vast resources which can easily modify their algorithms to reflect the current state of research.
Did DeepSeek seek and find a more efficient processing model for AI? Maybe, but you can count on the incumbents to adopt any new techniques found, no matter who finds them. It is the basis for a competitive and rich market. That all said, it would seem to me that if you still believe that AI is going to be big, this news out of China should only make you feel better… and perhaps present you with some buying opportunities. To be clear, that doesn’t mean that there won’t be bumps or risks (as price seeking continues). But it should be clear that there are some stocks that will be down this morning for the wrong reasons. 😉
FRIDAY’S MARKETS
Stocks declined on Friday, capping off a week of gains propelled higher by a milder-than-expected showing from the neophyte Administration—"it could have been worse” trade. Of course, a really solid start to earnings season gave stocks a boost, rightly so, as earnings are the cornerstone of good performance. 😉Traders were likely eying this week’s star-studded economic and earnings calendar to confirm lofty valuations.
NEXT UP
- New Home Sales (December) probably grew by 2.4% after climbing by 5.9% in November.
- This week: lots of earnings including the first 4 of the Mag-7 in addition to more housing numbers, Consumer Confidence, Q4 GDP, Personal Income, Personal Spending, PCE Price Index. Not to mention an FOMC meeting in the middle of the week that could be a showstopper. Download the attached earnings and economic calendars. Maybe take a peek at my daily chartbook which is chock full of information that may give you a leg up. 😉📈
- Key earnings to watch today: AT&T, Veradigm, SoFi, Nucor, and Western Alliance Bancorp.
DOWNLOAD MY DAILY CHARTBOOK HERE 📈
service@siebert.com
Siebert Financial (800)872-0444