DeepSeek Is Good News For The AI Gold Rush

•  In July and August of the past year, I explained the world had entered an AI and tech gold rush. Enormous amounts of money were pouring into building computer models that could extract and interpret golden data.

• Since that time, the number of AI miners and camp followers has exploded, making AI or Nvdia names that even your grandmother knows.

• In those articles, I cautioned that I believed the AI miners would find golden data, but how much of it they would find and how much investors would be willing to pay for it were big question marks.

• The other reality that I saw was that, with AI, we are in the mining phase, not the golden-products phase. No single product or service stands out at present as the game-changing prototype of things to come. 

This past week, a more ecnomical Chinese DeepSeek model caused lots of wild gyrations of most tech stocks. Net, net the tech market ended down only 3-4%. On the surface, DeepSeek, which costs a fraction of that of an NVDA, chip would seem to be a category crushing piece of software.

That many big techs rose sharply, such as GOOGL, AMZN, and Meta reminds us of what stage we are traversing in the gold rush. In short, those stocks rising clearly suggest that we are in the mining stage of the gold rush and not in the ingot stage. 

AMZN may be the biggest benefactor of the AI gold rush of any stock in America. They have millions of employees, millions of products, thousands of warehouses, millions of delivery miles worldwide, and billions of customers. If a better, more efficient, and less costly means of doing business is not in AMZN's future, AI will be the biggest bust since Beany Babies. 

Other giant benefactors of AI will be WMT, COST, the banking and insurance industries, and the industrial sector. If industrial sector companies like Raytheon, Caterpillar, Honeywell, 3M, and GE are to 'reshore', jobs from overseas, they must extract enormous costs from domestic manufacturing. My guess is they can do it, but it will take years, maybe decades, to get it done.

I believe DeepSeek will ultimately be viewed as a boon to the AI industry. It reduces the cost of participating in the AI gold rush and broadens the number of institutions and people who can afford to play.

Greg Donaldson is the author of "The Hidden Power of Rising Dividends: How to Produce Security, Income, and Growth." The book is available on Amazon and at your local book store.  

This article is not intended as investment advice. You should seek an investment professional's views before making any investment. 

These views I have shared here are my own and not any company I am associated with.

Caution: We Are Entering A "Prove It" Market

 

• In my recently released book, The HIdden Power of Rising Dividends,(available at Amazon) I make the case that dividend growth is highly correlated with price growth for many stocks and indices.

• In the book, I suggest that dividend growth alone is highly correlated with price growth for 25-35% of S&P 500 stocks. For an additional 50% of stocks, dividend growth is the most important indicator of value, but the correlation scores rise when we add some portion of sales and earnings growth, along with changes in interest rates.

• The consensus view of many stock market prognosticators today is that stocks, now trading at 27 time operating earnings, are extremely overpriced and are due for a big correction.

• My S&P 500 valuation model is telling a much more balanced story:

• 
  
• 
• 
  
• The above chart is what I call a Value Bar chart. The green bars show my model's annual predicted price of the S&P 500 going back to 2005. A quick look at the bar farthest to the right on the chart shows the model's current predicted price of the S&P 500. That figure is approximately 5,400. With the current price of the S&P 500 at around 6,000, the model is saying stocks are overvalued by about 10%. That, however, is before we factor in 2024 year end earnings and 2025 forward earnings.

•  Before we take a deeper look at the current predicted price, let's look back over the years to see how the model has fared.

• Simply speaking, if the red line (actual price) is above its corresponding Value Bar, we would say stocks are overvalued. If the the red line is lower that the Value Bar for the same year, we would say stocks are undervalued. For the last 20 years, the red line has stayed very close to the top of the Value Bars. A significant divergence is evident in only 2007, 2008,and 2022. In almost all other years, the Value Bars and actual prices of the S&P 500 are very close.

• Stocks continued to climb heading into the beginning of the Great Recession in 2007. At some point during the year, the model would have issued an overvalued signal. The model clearly signaled the market was overvalued in 2008.

• After the bear market of 2008 and 2009, the Value Bars stayed in fairly-valued or undervalued territory until the end of 2021, when they gave an overvalued reading. That signal correctly foresaw the selloff in 2022.

• That brings us to the current modest overvaluation. Plugging in Wall Street's current estimates of sales, earnings, dividends, and interest rates give us a figure of 6,500. 

• We are now entering what I call a "Prove It" market. This means, tech stocks, where the majority of the growth is coming from, must "Prove It" that they can continue with mid 20% sales and earnings growth. If they do, we should have another pretty good year. If not . . . .
•          

If any of you would like to discuss this article privately, please email me at info@gregdonaldson.com.