Chinese AI app - game changer? The Chinese app had been flying under the radar for a few weeks, but it surfaced in a big way over the weekend of January 25-26. By Monday, it was clear this was going to be a market game-changer. The tech-heavy Nasdaq 100 sank 3% lead by none other than Nvidia – down 17% for the day and resulting in a market cap loss of close to $600 billion, the biggest drop ever for a U.S. company.
To put that in perspective, Nvidia’s decline is more than double the entire market cap of Coca-Cola and Chevron. In a single day, Nvidia sank from number one to the third most-valuable public company, behind Apple and Microsoft.
To put that in perspective, Nvidia’s decline is more than double the entire market cap of Coca-Cola and Chevron. In a single day, Nvidia sank from number one to the third most-valuable public company, behind Apple and Microsoft.
So, what is this disrupter of American mega-cap companies? DeepSeek is a Chinese artificial intelligence company that develops open-source large language models (LLMs). The latest DeepSeek model provides responses comparable to other contemporary large language models, such as OpenAI's GPT. OK, so far it just sounds like competition. The disrupter angle is that it's trained at a significantly lower cost and requires a fraction of the computing power of a comparable LLM.
For example, whereas the world's leading A.I. companies train their chatbots with supercomputers using as many as 16,000 high-end graphics processing units (GPUs or chips), DeepSeek claims to have needed only about 2,000 GPUs. It was trained in 55 days at a cost of $6 million, which is roughly one tenth of what U.S. tech giant Meta spent building its latest A.I. technology.
So how did this Chinese start-up do it? The previous administration had placed sanctions on China for Nvidia chips, which were intended to restrict the ability of the country to develop advanced A.I. systems. But less-advanced models of Nvidia chips were still available. With necessity being the mother of invention, DeepSeek took those less-advanced chips and – to their credit - found a way to make it work.
Within two weeks of launch, DeepSeek’s first free chatbot app for iOS and Android had surpassed ChatGPT as the most-downloaded free app on the Apple App Store in the U.S.
Investor Fear
The fear among investors is that competitive performance at relatively minimal cost and power consumption will challenge the global dominance of American A.I. models, and more specifically, the need for Nvidia’s most expensive chips. Silicon Valley venture capitalist Marc Andreessen described the DeepSeek phenomenon as "AI's Sputnik moment."
Security and Privacy
Some experts are having a ‘hair on fire’ moment over fears the government of China could use the A.I. system for foreign influence operations, spreading disinformation, surveillance and the development of cyberweapons. One needn’t look any further than the company’s own privacy terms to see why: "We store the information we collect in secure servers located in the People's Republic of China... We may collect your text or audio input, prompt, uploaded files, feedback, chat history, or other content that you provide to our model and Services."
Going Forward
And For What It’s Worth…
The average price the Tooth Fairy pays U.S. children for a baby tooth is $4.58. This, per a survey of 4,000 Americans by the dental group Affordable Dentures & Implants. That beats the Franklin half dollar I used to get (yes, I’m dating myself here). While 86% of today’s Tooth Fairies leave cash, 4% give a gift instead, and 3% leave an ‘IOU’ note (really? an IOU?).
And here’s the shocker: 1% pay in bitcoin.
On a side note, I’m assuming that’s fractional shares of bitcoin and not whole numbers. Otherwise, with bitcoin currently trading north of $100,000, the next survey we see will be addressing the surge of American children needing the services of Affordable Dentures & Implants.
As noted before, long term, the strategies will get the trends right. Short term, there may be a miss or two as the market juggles conflicting signals. So keep allocations of strategies reasonable within your portfolio, and remember that protection* remains paramount.
--David
_____
* WHAT DOES PROTECTION LOOK LIKE?
At the extreme, it’s cash. As I mentioned last month, it’s OK to hold some cash. Cash is, in fact, a position. It means you’re prepared to act when circumstances better align with your risk tolerance.
Protection can mean an overweight position in a model built for protection. Lower volatility and lighter drawdowns often indicate that a model is more protective in nature. Bond Bulls, for example, has the lowest volatility and max drawdown of any of the models.
Check out the updated white paper Conservative vs. Aggressive Portfolios for a list of all the strategies ranked from lowest risk to highest in terms of max drawdown.
Protection can mean putting multiple strategies to work in a portfolio, especially when those models tend toward an inverse relationship with each other, or focus on different asset classes or market sectors. Think Bond Bulls and American Muscle. Or Global Trader and The 12% Solution. Or even a bit of the Zen Knuckle combined with a couple of the above.
Because each strategy uses a slightly different mechanism to identify market risks, and because each can employ different funds representing different market sectors (although there is obviously some overlap), there is beneficial diversification at work when using multiple strategies within a portfolio – helping to reduce volatility and max drawdown.
Further down the page in Conservative vs. Aggressive Portfolios you can see examples of various combinations and how they have performed over the years.
Protection can mean keeping an eye on provisional picks during the month. These can provide a heads-up on potential trends -- and breakdowns of existing trends. Look for asset class shifts (a switch from an equity fund to a safe harbor asset like cash or bonds, or the contrary).
See if such a shift holds up for a few days. Not every such move is a trading opportunity or justifies a rebalancing, but information is power.
Finally, employing stop-loss and stop-limit orders. A stop-loss order is an order placed with a broker to buy or sell a specific stock (or ETF) once that asset reaches a specific price. It's designed to limit an investor's loss on a security position. While not perfect, and you'll find my own pro-and-con thoughts on the Q&A tab in the Members Pages, stop-losses have their place in risk management.
Read more on the Investopedia page for Stop-Loss Orders.
For example, whereas the world's leading A.I. companies train their chatbots with supercomputers using as many as 16,000 high-end graphics processing units (GPUs or chips), DeepSeek claims to have needed only about 2,000 GPUs. It was trained in 55 days at a cost of $6 million, which is roughly one tenth of what U.S. tech giant Meta spent building its latest A.I. technology.
So how did this Chinese start-up do it? The previous administration had placed sanctions on China for Nvidia chips, which were intended to restrict the ability of the country to develop advanced A.I. systems. But less-advanced models of Nvidia chips were still available. With necessity being the mother of invention, DeepSeek took those less-advanced chips and – to their credit - found a way to make it work.
Within two weeks of launch, DeepSeek’s first free chatbot app for iOS and Android had surpassed ChatGPT as the most-downloaded free app on the Apple App Store in the U.S.
Investor Fear
The fear among investors is that competitive performance at relatively minimal cost and power consumption will challenge the global dominance of American A.I. models, and more specifically, the need for Nvidia’s most expensive chips. Silicon Valley venture capitalist Marc Andreessen described the DeepSeek phenomenon as "AI's Sputnik moment."
Security and Privacy
Some experts are having a ‘hair on fire’ moment over fears the government of China could use the A.I. system for foreign influence operations, spreading disinformation, surveillance and the development of cyberweapons. One needn’t look any further than the company’s own privacy terms to see why: "We store the information we collect in secure servers located in the People's Republic of China... We may collect your text or audio input, prompt, uploaded files, feedback, chat history, or other content that you provide to our model and Services."
Going Forward
- The DeepSeek impact will likely remain cloudy until the key U.S. players defend or otherwise comment on their heavy investments in A.I. Earnings season has begun, and February will be a telltale month. It’s worth noting that Nvidia CEO Jensen Huang and President Trump are meeting today to discuss A.I. policy, China restrictions, and DeepSeek.
- Speaking of earnings season, of those S&P 500 companies that have already reported for the final quarter of 2024, more than 70% have beaten analysts' expectations.
- The Federal Reserve this past Wednesday paused its interest rate-cutting cycle after reducing rates at its prior three meetings. Fed Chair Jerome Powell said inflation and jobs data would determine when another easing would come.
- Speaking of inflation, this morning’s reading of the "core" Personal Consumption Expenditures (PCE) index showed prices increased in line with expectations in December as inflation remained above the Fed's 2% target – but not worrisome.
- Tariff watch: President Trump says he will impose tariffs on Canada, Mexico, and China Saturday. The market was sinking late Friday as that deadline loomed. We’ll see what Monday brings.
And For What It’s Worth…
The average price the Tooth Fairy pays U.S. children for a baby tooth is $4.58. This, per a survey of 4,000 Americans by the dental group Affordable Dentures & Implants. That beats the Franklin half dollar I used to get (yes, I’m dating myself here). While 86% of today’s Tooth Fairies leave cash, 4% give a gift instead, and 3% leave an ‘IOU’ note (really? an IOU?).
And here’s the shocker: 1% pay in bitcoin.
On a side note, I’m assuming that’s fractional shares of bitcoin and not whole numbers. Otherwise, with bitcoin currently trading north of $100,000, the next survey we see will be addressing the surge of American children needing the services of Affordable Dentures & Implants.
As noted before, long term, the strategies will get the trends right. Short term, there may be a miss or two as the market juggles conflicting signals. So keep allocations of strategies reasonable within your portfolio, and remember that protection* remains paramount.
--David
_____
* WHAT DOES PROTECTION LOOK LIKE?
At the extreme, it’s cash. As I mentioned last month, it’s OK to hold some cash. Cash is, in fact, a position. It means you’re prepared to act when circumstances better align with your risk tolerance.
Protection can mean an overweight position in a model built for protection. Lower volatility and lighter drawdowns often indicate that a model is more protective in nature. Bond Bulls, for example, has the lowest volatility and max drawdown of any of the models.
Check out the updated white paper Conservative vs. Aggressive Portfolios for a list of all the strategies ranked from lowest risk to highest in terms of max drawdown.
Protection can mean putting multiple strategies to work in a portfolio, especially when those models tend toward an inverse relationship with each other, or focus on different asset classes or market sectors. Think Bond Bulls and American Muscle. Or Global Trader and The 12% Solution. Or even a bit of the Zen Knuckle combined with a couple of the above.
Because each strategy uses a slightly different mechanism to identify market risks, and because each can employ different funds representing different market sectors (although there is obviously some overlap), there is beneficial diversification at work when using multiple strategies within a portfolio – helping to reduce volatility and max drawdown.
Further down the page in Conservative vs. Aggressive Portfolios you can see examples of various combinations and how they have performed over the years.
Protection can mean keeping an eye on provisional picks during the month. These can provide a heads-up on potential trends -- and breakdowns of existing trends. Look for asset class shifts (a switch from an equity fund to a safe harbor asset like cash or bonds, or the contrary).
See if such a shift holds up for a few days. Not every such move is a trading opportunity or justifies a rebalancing, but information is power.
Finally, employing stop-loss and stop-limit orders. A stop-loss order is an order placed with a broker to buy or sell a specific stock (or ETF) once that asset reaches a specific price. It's designed to limit an investor's loss on a security position. While not perfect, and you'll find my own pro-and-con thoughts on the Q&A tab in the Members Pages, stop-losses have their place in risk management.
Read more on the Investopedia page for Stop-Loss Orders.
RSS Feed