In a world where trade policies are increasingly weaponized, we find ourselves facing yet another example of the unintended consequences of protectionist measures. The emergence of China’s DeepSeek AI model is a compelling case study in what could become a recurring theme in this blog.
U.S. restrictions on exporting advanced NVIDIA chips to China were intended to curb the nation’s AI advancements. In response – lacking the latest NVIDIA chips – researchers at DeepSeek developed innovative approaches to training large language models (LLMs) that are significantly less resource-intensive and more cost-effective. This strategy allowed them to achieve performance levels comparable to leading Western models, such as OpenAI’s, but at a fraction of the cost. (DeepSeek explained)
The announcement of DeepSeek’s R1 model had a significant impact on NVIDIA’s stock price. Initially, NVIDIA shares experienced a sharp decline, dropping 17% and erasing nearly $600 billion in market value. Although the stock has shown signs of recovery (NVIDIA Stock), this dramatic response raises several critical questions.
Do Trade Barriers Stifle Domestic Innovation?
Michael Porter’s seminal work on competitive advantage (Harvard Business Review) argues that companies perform best when exposed to international competition. Competing globally forces businesses to adopt innovative business models and operational approaches that may not emerge in a protected domestic market.
Over the past two years, investment in new data centers has surged, driven by growing digitalization and the rapid uptake of AI. While this has fueled demand for NVIDIA’s chips, it has also sparked debate over the massive energy requirements of AI data centers. The International Energy Agency (IEA) forecasts a surge in energy demand for data centers in the coming years (IEA Analysis).
With energy demands from AI competing with a projected 600% increase in grid capacity requirements for EVs, companies like Amazon and Google are investing in dedicated power generation using Small Modular Reactors (SMRs) (Data Center Energy Innovations).
This raises a crucial question: Were these investments based on an inefficient AI model that requires nearly ten times the computing energy of DeepSeek’s approach? And if so, why did no one question whether such massive energy consumption was necessary? One could argue that a sheltered approach to AI development has left American data centers and AI models at risk of pricing themselves—and their customers—out of the global market.
Do Trade Barriers Foster Innovation in Targeted Markets?
On the flip side, consider the engineers at DeepSeek. Constrained by sanctions, they were forced to develop more efficient training methods for LLMs. The necessity to innovate led to breakthroughs that Western companies had not yet pursued, demonstrating that trade barriers can drive the emergence of alternative solutions that challenge the industries they are meant to protect.
Lessons from History: Necessity Breeds Innovation
Throughout history, necessity has often led to more sustainable solutions. For instance, during World War II, resource shortages forced engineers and scientists to develop more efficient production techniques, many of which laid the foundation for modern manufacturing processes. This illustrates how constraints can drive the adoption of practical alternatives.
For those interested in how DeepSeek’s innovation caught the world by surprise, The Economist has published a podcast and an article titled “Tech Tycoons Have Got the Economics of AI Wrong”, which delves into the broader implications of DeepSeek’s emergence.
Conclusion: The Unintended Consequences of Trade Sanctions
The DeepSeek case provides yet another example of how trade sanctions can have unintended consequences, spurring innovation in unexpected places and challenging protected industries to adapt. Ironically, these policies may be undermining the very industries they aim to protect by limiting domestic incentives to innovate.
Given the growing number of examples, I can already envision my own very personal “unintended consequence” – which is the potential for a recurring series on this blog about how governments “get it wrong” by imposing trade sanctions that drive innovation abroad while stifling it at home. Stay tuned.
PGTS Inc. 