Is DeepSeek really an AI disruptor?
On Monday morning, we were all greeted with an equity market that was in free fall.
The technology stocks that have appreciated the most over the last 24 months were the hardest hit. All of the artificial intelligence innovators—and certainly the largest holdings in many of the Delta Wealth Management portfolios—forming the “picks and shovels” of the AI revolution were down a collective 15% to 20%. With the help of the Wall Street Journal and our own independent research, we will attempt to put this technology disruption in some form of historical context.
So what happened?
DeepSeek, a Chinese AI company, launched its latest model, R1. The company claimed that this program was particularly good at problem solving, performing on par with OpenAI’s model o1 but at a fraction of its operating cost.
What does that mean?
DeepSeek appears to have been able to achieve state-of-the-art performance without employing the most advanced chip technology due to U.S. export control. Early performance testers have claimed that R1 is on par with Open AI’s reasoning-focused model o1—but with a reported development cost of only $5.6 million (this is highly dubious).
What does that mean for investors?
If the performance claims of R1 are true, the large backlog of orders for semiconductors companies would decrease and the margin they have been realizing would become a thing of the past. Decreased earnings would have a negative effect on the stocks’ overall value.
This is particularly important as all Delta Wealth Management portfolios own several of these AI-related holdings.
What is DWM going to do?
Let us begin by saying what we are not going to do. We are not going to overreact and sell all of our AI positions because of the news release on Monday.
Anyone who has ever invested in technology development is aware that it is never a smooth ride. As new technologies change how we behave, so-called disruptors will enter the picture and try to alter the landscape with claims of “faster, better, and cheaper.”
This disruptor revealed itself on Monday of this week and, yes, we should pay attention. But we are skeptical about R1 and its ability for these three reasons:
1. You need horsepower to drive AI
I will use the car analogy. You need computing “horsepower” to develop and run these models. The U.S. has an almost unlimited budget and the best access to the most advanced technology known to man. Yet a group of very smart Chinese engineers supposedly developed a competitive model for under $6 million? Possible but highly unlikely.
2. Developed or, um, “borrowed”?
Microsoft is investigating suspicions that DeepSeek built its new AI app through the back door, Bloomberg News reported Wednesday morning. Microsoft engineers recently observed that large amounts of their internal data were exfiltrated from its OpenAI platform. China’s long-standing policy has been a willingness to go to almost any lengths to co-opt technology that other countries have developed.
3. Competition breeds more spending
“Increased competition rarely reduces aggregate spending,” Pierre Ferragu of New Street Research said. “Advanced frontier models still need to push the technical edge and use the most advanced computing resources, while smaller lagging edge models will push to develop more cost-efficient AI features.”
In the end, the competition to control this new technology should cause more spending rather than less. For DWM clients, this should have a beneficial effect on your equity portfolio.
The landscape of AI is changing on a daily basis as its development could be one of the most powerful and widely adapted technologies of this century. AI is just one of the sectors that DWM has focused its investment efforts on; we also have investments in health care, heavy manufacturing, consumer staples, and our most non-correlated asset class, high yield municipal project bonds.
Our goal is to balance risk in equity investing. If you have any questions about how we accomplish that objective, please feel free to reach out.