Key Takeaways from the Interview with TSMC's Morris Chang
My notes on the recent interview that the guys of Acquired did with TSMC's founder Morris Chang.
The Rise of TSMC
Morris Chang, the founder of TSMC, received a letter in 1997 from Jensen Huang, the CEO of Nvidia, who was seeking a foundry partner. Chang was initially annoyed that the TSMC San Jose office had ignored Nvidia's request, which led him to personally contact Huang. Nvidia was a small company of 50-60 employees at the time and facing bankruptcy, while TSMC was a 10-year-old company with a revenue of over $1 billion. Chang's visit with Huang led to a crucial partnership where TSMC manufactured chips designed by Nvidia.
Nvidia's 2009 Dispute Resolution
Morris Chang resolved the 2009 dispute with Nvidia after returning as CEO of TSMC by taking several key actions. The dispute arose from manufacturing and quality problems with the 40 nanometer node, which was causing delays and financial issues for Nvidia. Here's how Chang addressed the situation:
Immediate Action: Upon retaking the CEO position, Chang identified the ongoing dispute with Nvidia as a priority problem. He personally called major customers, including Jensen Huang of Nvidia, to understand the situation.
Investigation: Chang thoroughly familiarized himself with all aspects of the 40 nanometer problem, including the manufacturing issues, quality concerns, and the financial impact on Nvidia.
Financial Settlement: After weeks of work, Chang decided that the issue was primarily a matter of money. He devised a financial offer, amounting to over $100 million, to compensate Nvidia for the damages they had suffered.
Personal Meeting: Chang emailed Jensen Huang to schedule a meeting at his home in Silicon Valley, starting with a casual family dinner before discussing business.
Clear Offer: At the meeting, Chang presented the offer and made it clear that the offer was effective for only 48 hours. He stated that if Nvidia didn't accept the offer within that time, they would have to go to an arbitrator, as Jensen had previously suggested to the former CEO.
No Negotiation: Chang emphasized that there would be no further negotiations or bargaining, ensuring a quick resolution.
Acceptance: Jensen Huang accepted the offer within the 48-hour timeframe.
TSMC's Pure Play Foundry Strategy
TSMC's initial strategy regarding competition with customers was centered around a "Pure Play Foundry" model, which meant that it would not compete with its customers by designing or selling its own branded chips. This strategy was a deliberate choice by Morris Chang and was key to TSMC's early success and its ability to become a leader in the semiconductor industry.
Here are the key aspects of this strategy:
No in-house chip design: TSMC focused solely on manufacturing chips for other companies and did not have its own design division. This was a significant departure from the traditional model of integrated device manufacturers (IDMs) like Intel, who designed and manufactured their own chips.
Serving all customers: By not competing with its customers, TSMC positioned itself as a neutral and reliable manufacturing partner. This allowed it to attract a wide range of customers, including both established IDMs and emerging fabless companies.
Building trust: The pure-play model fostered trust among customers, who were more willing to work with TSMC knowing that it would not compete with them. This was particularly important for fabless companies that needed a reliable manufacturing partner to bring their chip designs to market.
Focus on manufacturing expertise: TSMC concentrated its resources and expertise on improving its manufacturing processes and technologies. This allowed it to provide high-quality manufacturing services and remain at the forefront of semiconductor manufacturing.
Acknowledging limitations: TSMC recognized it did not have the capability to design chips like Intel. So, rather than attempt to compete directly, it focused on its strength, which was manufacturing. This allowed them to leverage their manufacturing expertise without competing with their customers in the chip design market.
This strategy was not just about avoiding competition; it was also a recognition of the changing dynamics of the semiconductor industry. Morris Chang anticipated the rise of fabless companies that would specialize in chip design and would need a dedicated foundry to manufacture their products. By adopting the pure-play model, TSMC positioned itself to benefit from this trend and became the leading foundry in the industry.
TSMC's pure-play strategy was also about recognizing their own limitations. They did not have the capabilities to compete with established companies like Intel, so they decided not to try. Instead, they focused on manufacturing, which they could do well, and allowed them to forge a different path to success.
This strategy, though initially developed as a way to avoid direct competition, inadvertently created a more collaborative ecosystem where many companies could specialize in different parts of the semiconductor value chain, while relying on TSMC's manufacturing capabilities.
TSMC's 28nm Investment
Several factors contributed to TSMC's decision to heavily invest in the 28-nanometer node, marking a pivotal moment in the company's history and establishing its leadership in the semiconductor industry. These factors include:
Increased R&D Budget: Morris Chang, after returning as CEO, made a significant decision to set the R&D budget at a fixed 8% of revenue, regardless of economic conditions. This was a strategic move to ensure consistent funding for research and development, which was previously negotiated each year. This commitment to R&D provided the necessary resources and stability for TSMC's engineers to focus on developing leading-edge technologies.
R&D Team's Recommendation: The R&D team, empowered by the increased budget, identified the 28-nanometer node as a "sweet spot" in technology development. They presented compelling technical reasons to Chang, who, based on his trust in his team, decided to support their recommendation. This internal confidence and technological insight was a crucial factor in the decision-making process.
Market Forecasting and Business Development: TSMC had a well-established market forecasting group and a business development department that provided strategic insights. The marketing group, in particular, helped identify the market potential and strategic importance of the 28-nanometer node. This combination of technical expertise and market awareness was a driving force behind the decision.
Anticipation of Market Demand: While the specific details of the smartphone market's explosive growth were not fully anticipated by Chang himself, the business development team may have had more detailed visibility. Nonetheless, the market intelligence available to TSMC at the time suggested that the 28-nanometer node would be very significant. This anticipated demand was a key consideration in the decision to invest heavily.
Strategic Timing: Chang quoted Shakespeare, noting that "there is a tide in the affairs of men, which taken at the flood, leads on to fortune". He saw the 28-nanometer node as the opportune moment for TSMC to capitalize on its capabilities and secure a leading position in the industry. This strategic vision and sense of timing was a factor in pushing for the heavy investment.
Building on Previous Successes: TSMC had built up a good infrastructure and organizational capabilities, as well as market forecasting, which allowed it to leverage previous experience and insights for the 28 nm node. This built-in advantage allowed for more confidence in the investment.
Capital Expenditure Increase: The decision to increase capital expenditure to almost $6 billion in 2010 was essential for manufacturing the 28nm chips. The board questioned this increase but Chang was able to convince them, making the large financial bet necessary for success.
TSMC and Apple: A Partnership Forged
TSMC secured Apple as a major customer through a combination of strategic actions, technological capabilities, and a bit of serendipity:
Initial Contact through a Third Party: The initial contact between TSMC and Apple was facilitated by Terry Gou, the founder and CEO of Foxconn. Gou, who is also a second cousin of Morris Chang's wife, Sophie, brought Jeff Williams, Apple's Chief Operating Officer, to dinner at Chang's home in 2010. This unexpected visit was the start of the formal engagement between TSMC and Apple.
Jeff Williams's Pitch: During the dinner, Williams initiated a pitch for TSMC to manufacture chips for Apple. This was a significant opportunity for TSMC, but it also came with specific requirements from Apple.
Initial Pricing Discussion: Williams mentioned a 40% gross margin, which Chang considered low at the time, since TSMC's margin was already 45%. However, Chang did not immediately engage in a detailed pricing discussion, recognizing that there were more important issues to discuss.
Apple's Request for 20nm Node: Apple surprised TSMC by requesting the 20-nanometer node instead of the 28-nanometer node, which TSMC was about to begin production on. This was a diversion, as TSMC had planned to move from 28nm to 16nm, which made the Apple request a "detour."
TSMC's Financial Planning: To meet Apple’s demands, TSMC had to figure out how to fund the additional capital expenditures, after already investing heavily in the 28nm node. After consultations, they decided not to cut dividends or sell new stock, but to borrow money instead.
Negotiating the Order: TSMC decided to only take on half of Apple’s order to mitigate the risk associated with such a large demand [4]. This decision was initially met with skepticism by Apple's purchasing team, but ultimately, Chang convinced Jeff Williams during a meeting that TSMC was prudent in taking half the order.
Temporary Pause Due to Intel: In February 2011, Apple paused discussions with TSMC for two months because Intel had approached Tim Cook, the CEO of Apple, and asked to be considered as a supplier. At this point, Intel was a major supplier for Apple's Mac line, which was built on Intel's x86 architecture.
Tim Cook's Assurance: Morris Chang then reached out to Apple to see if they still wanted to do business. In a follow-up meeting, Tim Cook assured Chang that Intel did not know how to be a foundry, indicating that Apple was still committed to working with TSMC.
Overcoming Technical and Logistical Challenges: TSMC was able to resolve differences with Apple on yield and pricing. This included a personal visit from Jeff Williams, who negotiated pricing with Chang.
Commitment to Technology Leadership: Despite the initial detour, TSMC ultimately delivered on the 20-nanometer node. While this decision delayed the 16-nanometer node development, TSMC's ability to produce high-quality chips and meet Apple's requirements solidified its position as a crucial partner.
Long-Term Partnership: After TSMC began producing the 16-nanometer chips, Apple committed to buying the majority of their chips from TSMC, as they had promised, cementing TSMC’s long term value to Apple.
TSMC's Competitive Strategy
TSMC's competitive strategy is multifaceted, combining a unique business model with a focus on technology, customer relationships, and operational excellence. Here are some of the key aspects of its strategy:
Pure-Play Foundry Model: TSMC's most distinctive competitive advantage is its pure-play foundry business model. Unlike integrated device manufacturers (IDMs) such as Intel, TSMC does not design or market its own chips. This focus allows TSMC to avoid competing directly with its customers, which are often chip design companies. By being a neutral manufacturer, TSMC can serve a wide range of customers without creating conflicts of interest.
Customer Trust: TSMC's commitment to not competing with customers has fostered a high level of trust. Customers know that TSMC will not use their designs to create competing products, which is a significant concern with IDMs that also have a foundry business. This trust is a crucial factor in attracting and retaining major customers like Apple and Nvidia.
Technological Leadership: TSMC has consistently invested in R&D, establishing itself as a leader in advanced semiconductor manufacturing technology. This focus on technology has enabled TSMC to offer leading-edge manufacturing processes that meet the demands of high-performance applications like smartphones and AI. The company's ability to innovate and produce cutting-edge technology is a key differentiator that attracts customers.
Strategic Investments in R&D: TSMC's strategy of allocating a fixed 8% of revenue to R&D, regardless of economic conditions, has empowered its R&D team to pursue ambitious projects and identify key technological opportunities. This consistent investment has enabled TSMC to stay at the forefront of semiconductor manufacturing technology.
Focus on the Learning Curve: TSMC leverages the concept of the learning curve, aiming to achieve the lowest cost per unit by maximizing production volume. This involves starting with lower prices, even at the expense of early profits, in order to secure large orders and move rapidly down the learning curve. This approach allows TSMC to offer competitive pricing and attract more customers.
Ecosystem Engagement: TSMC's location within the Hsinchu Science Park in Taiwan and its engagement with the ecosystem that is located there with its suppliers and customers creates an advantage. The close proximity of equipment vendors like ASML, design software companies like Cadence and Synopsis, and its customers facilitates collaboration and rapid innovation.
Capital Expenditure: TSMC makes massive capital investments in manufacturing capacity to achieve economies of scale. The company has proven to be able to manage these massive expenditures, which are required to compete at the leading edge of semiconductor manufacturing, while also managing to maintain profitability.
Customer-Centric Approach: TSMC has demonstrated a willingness to accommodate customer-specific needs, even when those needs deviate from the company's standard roadmap. For example, TSMC agreed to develop the 20-nanometer process for Apple, despite its original plans to move directly from 28 to 16 nanometers. This flexibility and willingness to work with customers distinguishes it from other foundries.
Talent Management: TSMC has a unique culture that prioritizes its employees. It avoids layoffs, even in times of economic downturn. TSMC’s talent strategy aims to retain their knowledge base and expertise and helps promote a culture of stability that is beneficial to the company’s goals.