DeepSeek’s Rise is a Wake-Up Call for the Entire AI Supply Chain — Not Just the Chips

As President and CEO of Cascade Systems Technology (CST) – and as a long-time technologist and participant in the Chips industry through my years at Intel,  I’ve witnessed firsthand how rapidly the tech landscape can evolve.

DeepSeek’s recent and well-publicized emergence as an AI player, and yes, disrupter, have clearly and understandably shaken up the industry, igniting fears, prompting soul-searching, and even rattling stock market valuations of American companies we view as the undisputed leaders of AI (OpenAI, Meta, Google, et al) as well as their field of favored semiconductor chip suppliers (Nvidia, Intel, Qualcomm, Samsung, etc.). DeepSeek’s rise and appearance in the public and technology consciousness is a good reminder of analogous situations during the rise of the Internet and Data center technologies.

Remember virtualization in the Data Center in the late 90s? Virtualization enabled deployment of many virtual machines on a single server, which raised the concern that a smaller number of servers and server chips would be sufficient. If you could run, say, 10 virtual machines on one CPU, why would you buy 10 CPUs? Instead, the torrid growth of the internet and data centers pushed more use of the high-performance chips that could host more virtual machines. Higher efficiency from virtualization technology drove higher — not lower — demand for highest performance chips. Also, remember the notion of “thin client” in the late 90s where it was assumed that most compute capabilities would be in the backend data center with the devices (PCs, laptops, phones….) being “thin” – i.e. not carrying much compute power? Over the last 2+ decades, our devices have become very capable and compute rich with intelligent connectivity and interface to the cloud/data center.

Similarly, DeepSeek’s emphasis on algorithmic efficiency challenges us to rethink the current approach of the AI training engines. More efficient training algorithms will only drive higher demand for the complete AI solution — chips, circuit board assemblies (PCBAs), systems, and software. AI processing (Training, Inferencing….) will need to be properly architected by what the solution use cases need. Vice President JD Vance recently mentioned the need to “ensure that the most powerful AI systems are built in the U.S. with American designed and manufactured chips.” Those “AI systems” that Vice President Vance refers to include the PCBA’s, full systems and software for AI that are proudly designed and manufactured in the USA. AI chips require fully assembled circuit boards: They can’t work by themselves. (Or, to adopt the PCBAA’s mantra: “Chips Don’t Float!”)

In other words, disruptions are a given in any competitive industry. Whether they originate domestically or abroad, they compel us to reassess our strategies and push technological boundaries.

OK, so where to next?

While DeepSeek like disruptions can be unsettling, I see this as an opportunity rather than a setback — an inflection point that will drive innovation and progress across the entire AI Silicon supply chain.

We must take this moment to invest in R&D, explore new architectures, and drive the next wave of innovation across the AI supply chain, including chips, PCBAs, systems, software, supporting services, and more. History proves that industry “shocks” often lead to long-term growth. Higher efficiencies fuel greater demand and accelerate progress.

Next — the recent media, stock market, and electronics industry hullabaloo around DeepSeek steels my resolve to recognize and invest aggressively in the chips, Circuit board and Systems Assembly that enables AI. Manufacturers like CST and other U.S.-based electronics suppliers, who have a role to play in driving innovation will enable us to retain and defend a leading edge in AI. (See other blogs I’ve written about this very topic, prior to the DeepSeek frenzy of two weeks ago)

Simply put, we should see DeepSeek’s rise as a wake-up call, not just a warning bell. Now is the time to double down on our investments and innovation across the complete AI Chips supply chain so it’s “Made in America” for the decades ahead!

The AI race is far from settled – see this good article on Brookings.edu by Sara Kreps that expands on this topic — and with the right approach, the U.S. can continue to define the future.

–  Shantanu R. Gupta,
President & CEO, Cascade Systems Technology

Share on Facebook

Tweet

Follow us