Unless you’ve been in airplane mode for the past week, you’ve probably heard about DeepSeek-R1. (If you haven’t, please see my Sunday essay: DeepSeek-R1: The Exception That Could Redefine AI). In short: DeepSeek-R1, an open-source AI model from Chinese startup DeepSeek, has shaken up the industry by delivering performance comparable to leading models like OpenAI’s reasoning engine o1, its LLM GPT-4, and Anthropic’s Claude 3.5 Sonnet while operating at a fraction of the cost. This breakthrough has sparked a debate: is DeepSeek-R1 a preview of a future driven by algorithmic efficiency or an outlier that reinforces the dominance of brute force foundational models?

By today’s opening bell, the U.S. stock reacted (as markets do) to its interpretation of the issue. At the time of this writing, major technology stocks are facing substantial declines. Nvidia’s (NVDA) shares have dropped nearly 12%, reflecting apprehensions about the sustainability of U.S. dominance in AI. Other tech giants, including Microsoft (MSFT), Meta (META), and Alphabet (GOOGL) have also seen their stocks fall by approximately 3% in pre-market trading. The Nasdaq 100 index has decreased by 2.8% and the S&P 500 is down by 1.7%.

Financial pundits are saying that investors are beginning to question the substantial capital expenditures by U.S. firms in AI research, given DeepSeek’s cost-effective approach. Importantly, I am not a financial advisor, and nothing in this article should be considered financial advice. If you are considering any type of investment, you should conduct your own research and, if necessary, seek the advice of a licensed financial advisor.

That said, anyone who believes that they understand the future of AI (or its future computational requirements) is delusional. As with all things, the future will be a combination of every technique. Everything we know about hardware and software tells us that incremental improvement always includes a combination of algorithmic efficiency and increased hardware capabilities. People who sold their shares today believing that we’ve “reached the end” of hardware will get what they signed up for. That’s how markets work.

More By This Author:

DeepSeek-R1: The Exception That Could Redefine AI
Google Invests Additional $1 Billion in Anthropic
What Did You Learn From Wall Street’s DeepSeek Moment?