Key Takeaways
- In 1996, chip manufacturing capacity was so scarce that incumbents like NEC, Toshiba, Fujitsu, and HP demanded startups surrender their proprietary technology licenses just to get fab access.
- NVIDIA bypassed predatory licensing terms by cold-calling TSMC from the Yellow Pages and mailing a letter directly to founder Morris Chang after hitting an automated phone recording.
- Morris Chang called back at 6:30 PM to NVIDIA's noisy office with a single pledge: “Jensen, if we were to work together, I promise you we would jump through hoops for you.”
- Huang chose TSMC over rival UMC, situated right across the street, because Chang offered corporate integrity and alignment instead of attempting to extract startup equity or IP.
The Foundry Squeeze of 1996
In 1996, the personal computer boom triggered a desperate land grab for silicon wafers. Established giants possessed all the fabrication plants, while small design shops had zero chips of their own. Jensen Huang faced a hostile environment. If a young startup wanted access to a production line, the supplier extracted an intolerable tax: handing over proprietary design blueprints.
“At the time, there was so little foundry capacity, if you wanted to use somebody's foundry, you had to license your technology to them,” Huang explained. Big players like NEC, Toshiba, Fujitsu, and Hewlett-Packard treated fab space like a toll booth. They used manufacturing dominance to swallow the designs of their future competitors.
Huang refused to give up his core assets. Instead of surrendering NVIDIA's intellectual property to Japanese conglomerates or Silicon Valley incumbents, he went hunting for a pure-play foundry model where the manufacturer agreed never to compete with its own customers. That search pointed toward a relatively young Taiwanese operation called TSMC.
The 6:30 PM Call in a Noisy Office
Finding Morris Chang in 1996 was not a matter of warm venture capital introductions. Huang searched through phone listings. “I look it up, it's a TSMC. There's a number there, and I call their number, and it's a recording,” Huang recalled.
He did not quit after the dead phone line. Huang sat down, drafted a physical letter outlining NVIDIA's graphic processor architecture, and mailed it across the Pacific directly to Chang. Weeks passed with silence. Then, around 6:30 in the evening, the phone rang inside NVIDIA's crowded, noisy workspace. The voice on the other end identified himself: “This is Morris Chang from TSMC.”
Chang did not demand licensing rights. He did not ask NVIDIA to yield its patents. Rival foundry UMC sat right across the street in Hsinchu Science Park, ready to negotiate terms, but Chang presented a completely different operating philosophy. As Huang remembered: “He said one thing. He said, 'Jensen, if we were to work together, I promise you we would jump through hoops for you.' And for the whole time, 30 years, they've been jumping through hoops for us.”
Why Supplier Alignment Beats Scale
Early-stage founders often believe bargaining power comes from size. Huang proved that picking partners who share your ethical structure matters far more when you have zero chips on the table. UMC had physical plants. Major Japanese conglomerates had massive balance sheets. Yet TSMC won NVIDIA's business because Morris Chang established a pure service charter: TSMC would never build its own branded chips, and it would never steal customer designs.
When you negotiate from a position of weakness, transactional suppliers will squeeze you for equity or IP. Chang recognized that NVIDIA's survival directly fueled TSMC's manufacturing scale. That alignment transformed a cold letter into a thirty-year shared run in computing.
What to Do With This
Audit your top three technical vendors this week. If any supplier requires terms that give them claims on your data, IP, or customer relationships, stop expansion with them immediately. Replace them with pure-service partners whose business model profits strictly from your scale, not your margin.