Pure-Play Foundry Positioning (Never Compete With Your Customers)
Pick the one link in the value chain where you'll never be your customer's rival.
- Difficulty
- Moderate
- Time to result
- ~ongoing to results
- Steps
- 5
- Confidence
- 85%
TSMC's foundational strategic choice was to occupy a single position in the semiconductor value chain — contract manufacturing — and never expand into chip design or end products that would compete with its own customers. This model, formalized on TSMC's original pitch deck as a 'dedicated pure-play foundry,' meant TSMC declined opportunities (like IBM's 1999 request to co-develop process technology as a 'senior partner') that would compromise its independence or trustworthiness. The payoff compounded over decades: because integrated device manufacturers like Intel design their own competing chips, TSMC could earn structurally higher trust from fabless customers such as Apple, who explicitly cited Intel's conflicted incentives as a reason to switch. The model turned a resource constraint into a durable competitive moat.
Origin
TSMC's founding pitch deck (shown to the Taiwanese government and early investors) explicitly named 'dedicated pure-play foundry' as a core bullet point of the business plan.
Core principles
- 01Choose one value-chain position and never expand into segments that compete with customers
- 02Vertically-integrated rivals carry a structural trust deficit with shared customers
- 03Turning down partnerships that compromise independence protects long-term positioning, even if it costs short-term revenue
- 04Serving many customers with one consistent operating model beats customizing for each
How to run it
- 1
Map the full value chain around your product
Identify every segment (design, manufacturing, packaging, end-product) a company in your industry could occupy.
- 2
Choose the segment where you add the most value without threatening others
TSMC picked contract manufacturing only, leaving design entirely to customers.
- 3
Formally rule out competing segments
Commit explicitly (even in founding documents) to not entering chip design or end-products, so customers never have to guess your intentions.
- 4
Decline deals that compromise independence
Turn down partnerships (like co-development arrangements) that would put you in a subordinate or compromised position relative to a competitor-customer.
Watch out Rivals may react angrily and go to competitors instead, as IBM did when TSMC declined and IBM turned to UMC.
- 5
Let the trust advantage compound
As integrated competitors' conflicts of interest become more visible to shared customers, expect switching in your favor.
In the wild
In 2011, Apple's COO Jeff Williams told Chang the company needed to pause discussions because Intel's CEO had approached Tim Cook directly. Weeks later Cook told Chang in person that Intel 'just does not know how to be a foundry.'
→ Apple committed its leading-edge chip manufacturing to TSMC, becoming one of its largest customers.
IBM, losing Qualcomm's business, asked TSMC to co-develop its next process node as IBM's junior partner. Chang refused instantly, unwilling to cede control of TSMC's own technology roadmap.
→ IBM took the offer to UMC instead, which came to regret accepting it, while TSMC kept full control of its technology.
Common mistakes
Accepting subordinate co-development deals
IBM expected to be the 'senior partner' in any technology partnership; accepting this would have meant TSMC losing the ability to develop its own process independently.
Underestimating a rival's conflict of interest
Integrated competitors like Intel design competing end-products, which customers ultimately distrust, even if the competitor doesn't intend to misuse the relationship.
Is it for you?
Best for
Companies in industries where suppliers and customers can become competitors, and where structural trust is a differentiator.
Not ideal for
Businesses where vertical integration produces significant efficiency gains that outweigh trust concerns.
From the transcript
“he told me there's nothing to worry about because uh Intel uh just does not know how to be a found”
“we declined without having to think about it at all we declined the the IBM”
From the episode
TSMC Founder Morris Chang
Morris Chang