Week Ahead - February 3, 2025

Market Commentary – February 3, 2025

Last week began with a surprise when Chinese startup DeepSeek demonstrated that its new artificial intelligence (AI) system could perform as well as expensive American systems – and apparently 100x cheaper and by using domestic Chinese technology.

This generated immediate fear that America’s dominance in AI would rapidly diminish, and that AI investment to-date might be “sunk capital.”

But then cracks started to show in the DeepSeek story. First, it came out that DeepSeek used many thousands of Nvidia chips to achieve its breakthrough. Then allegations began flying that China had engaged in industrial espionage and theft of intellectual property. These accusations were repeated by David Sacks, Trump’s new AI and Crypto Czar, and echoed by leading AI firm OpenAI and its biggest backer, Microsoft.

The shock to the industry (and the market) has been compared to the Soviet launch of Sputnik, the first man-made satellite, in 1957 – and the extreme stress it caused the U.S. and our allies. Fearing Soviet superiority in space technology and missile capabilities, the U.S. barreled full speed into the Space Race. We agree with the DeepSeek::Sputnik analogy, but quickly point out that as a result, NASA was created. With funding that eventually grew to $10.5 billion a year, the U.S. won the Race when Neil Armstrong stepped on the moon in 1969.

Another angle to the DeepSeek story is called the “Jevons Effect” or “Jevons Paradox. “In the mid-1800s, English economist William Stanley Jevons observed that technological advancements of the industrial revolution improved the efficient use of coal but did not lead to a decrease in the usage of coal. Because the cost of using coal was driven down, it led to greater consumption of coal, not less. We’ve seen this paradoxical pattern repeated in history with personal computers in the ’80s and ‘90s , as well as with telephones and the internet. We believe this DeepSeek event will actually increase demand for AI and hasten its distribution and use throughout society. We believe AI development and implementation are in the early stages, with many new technologies expected to emerge in the coming years.

Tech Cycle: Time To Move From Semiconductors To Software

Technology follows a cycle: First, new chips and hardware are developed and marketed, then software is developed to operate on that equipment. We believe the chip/hardware cycle is transitioning to the software cycle, which we highlighted in our 2025 Year Ahead report in December. Investment capital follows this cycle, too. That was evident in the early 1960s when Ross Perot left IBM to found Electronic Data Systems (EDS), a pioneer in outsourced IT services, computer systems management, and data processing – they wrote the software for business computers. As noted above, we’ve seen the cycle in personal computers, telephones and the internet as well. We believe it has now begun in AI, too, presaging wider adoption across society and around the world.