July 3, 2024 • 20min
The two companies driving the modern economy
Planet Money
Key Takeaways
- Nvidia went from making niche graphics cards for gaming to becoming one of the most valuable companies in the world by dominating the AI chip market
- Nvidia's success is due to its early bet on GPU computing for AI, its CUDA software platform, and CEO Jensen Huang's deep involvement and foresight
- Taiwan Semiconductor Manufacturing Company (TSMC) revolutionized the chip industry by focusing solely on manufacturing chips designed by other companies
- TSMC's founder Morris Chang created this new business model after facing career obstacles as an Asian executive in the US
- TSMC now produces the majority of the world's most advanced chips, making Taiwan a critical part of global tech supply chains
- The US is trying to boost domestic chip manufacturing through the CHIPS Act, but faces challenges competing with TSMC's scale and expertise
Introduction
This episode of Planet Money explores two stories that transformed the semiconductor chip industry over the past 40 years. The first half focuses on Nvidia's rise from a niche graphics card maker to an AI chip powerhouse. The second half examines how Taiwan became the center of advanced chip manufacturing through the story of TSMC founder Morris Chang.
Topics Discussed
Nvidia's Origin and Early Years (3:02)
Nvidia was founded in 1992 by Jensen Huang and two engineer friends, with the goal of improving video games through specialized 3D graphics chips. The company's name comes from the Latin word for "envy."
- Nvidia initially focused on making graphics processing units (GPUs) for gaming
- CEO Jensen Huang is known for his intense management style and deep involvement in technical details
- Huang began moving Nvidia towards AI and machine learning applications about 20 years ago
"Jensen is truly singular in Silicon Valley...He is probably more engaged in every detail at Nvidia than any of those other founders ever were at their companies." - David Rosenthal, tech podcast host
Nvidia's Pivot to AI (5:23)
In the early 2000s, Nvidia made a big bet on selling supercomputing power more broadly. A key part of this strategy was launching CUDA, a software development framework that acts as a middleman between developers and Nvidia's chips.
- The 2012 ImageNet competition was a turning point for AI, with a neural network called AlexNet achieving a breakthrough in image recognition
- AlexNet used Nvidia GPUs to train the neural network much faster than traditional CPUs
- This sparked interest from major tech companies in using GPUs for AI applications
"People at Google and people at Facebook said, holy c**p...They realized these image classification systems to build way better social media feed recommenders." - David Rosenthal
Nvidia's Current Dominance in AI Chips (8:23)
Nvidia has built a strong competitive advantage through its CUDA software platform and network effects. However, this dominance has led to speculation about potential antitrust action.
- CUDA has become the standard development platform for AI, creating high switching costs
- Nvidia's stock price is very highly valued, leaving room for a potential fall
- The Department of Justice has reportedly cleared the way for possible antitrust action against Nvidia
Morris Chang's Early Career (13:51)
Morris Chang came to the US from China as a student, studying engineering at MIT with the goal of making money. He joined Texas Instruments and rose quickly to become a vice president by the early 1970s.
- Chang pioneered outsourcing chip packaging and testing to Taiwan to save costs
- He developed new pricing strategies that helped capture market share
- Despite his achievements, Chang felt his career stalled due to discrimination as an Asian executive
"He wanted to be chief executive...He felt that given the nature of the culture at Texas Instruments, he would never become the CEO." - Chang-Tai Hsieh, University of Chicago economist
Chang's Move to Taiwan and Founding of TSMC (16:18)
In 1983, Chang left Texas Instruments after 25 years. He was recruited by the Taiwanese government to help develop the country's tech industry. This led to the founding of TSMC in 1987.
- Chang initially ran Taiwan's Industrial Technology Research Institute
- He realized Taiwan lacked enough chip designers to compete directly with US companies
- Chang developed a new business model focused solely on manufacturing chips for other companies
"His basic idea was to say, why don't we develop a company that would only serve other customers in which we completely get out of the design business? We only do the manufacturing part of the business." - Chang-Tai Hsieh
TSMC's Impact on the Semiconductor Industry (18:58)
TSMC's "foundry" business model revolutionized the chip industry by separating design and manufacturing. This allowed many new chip design companies to enter the market without massive capital investments.
- Companies like Nvidia could focus on chip design while outsourcing manufacturing to TSMC
- Some established chip companies sold off their manufacturing operations
- TSMC is now the world's largest contract chip manufacturer and 8th largest company overall
"A company like Nvidia would not have existed without TSMC." - Chang-Tai Hsieh
US Efforts to Boost Domestic Chip Manufacturing (20:09)
Concerns about reliance on Taiwan for advanced chips have led to US government efforts to increase domestic manufacturing capacity. In 2022, the CHIPS Act was passed to provide $39 billion in incentives for chip manufacturing in the US.
- Each new advanced chip factory costs around $28 billion to build
- Some experts are skeptical about the effectiveness of the CHIPS Act subsidies
- TSMC is building a factory in Arizona with $6.6 billion in CHIPS Act funding
"I would say that the only plausible case for something like the CHIPS Act is if at some point in the future there's war between Taiwan and China and TSMC is either destroyed or TSMC is taken over by the Chinese Communist Party." - Chang-Tai Hsieh
Conclusion
The stories of Nvidia and TSMC illustrate how innovation and new business models can transform entire industries. Nvidia's early bet on GPU computing for AI and TSMC's focus on contract manufacturing reshaped the semiconductor landscape. Both companies benefited from visionary leadership that saw opportunities others missed.
These developments have made Taiwan central to global technology supply chains, particularly for advanced chips. This has created both economic opportunities and geopolitical risks. The US government's efforts to boost domestic chip manufacturing highlight the strategic importance of semiconductors in today's economy.
Looking ahead, Nvidia's dominance in AI chips and TSMC's manufacturing prowess face potential challenges from competitors and geopolitical tensions. The semiconductor industry will likely continue to evolve rapidly as AI and other emerging technologies drive demand for ever more powerful and efficient chips.