Part 1: Build Everything Yourself
Bangladesh has decided to enter the semiconductor industry, and our July 2026 national roadmap names design as the first step. About 1,200 of our engineers already work in the field. The question for us is no longer whether we enter, but where our scarce engineers and money should go first. Six decades of chip ventures give a clear answer to that question. Value goes to whoever owns the customer and the intellectual property, and this series follows that rule through six eras before we turn to what it means for us.
Nov 9, 2026
Part 1 · 1957–1979 · Build Everything Yourself
Bangladesh has decided to enter the semiconductor industry, and our July 2026 national roadmap names design as the first step. About 1,200 of our engineers already work in the field. The question for us is no longer whether we enter, but where our scarce engineers and money should go first. Six decades of chip ventures give a clear answer to that question. Value goes to whoever owns the customer and the intellectual property, and this series follows that rule through six eras before we turn to what it means for us.
The story starts in 1957, when eight engineers left Shockley Semiconductor and founded Fairchild Semiconductor in California. Their backing came from Fairchild Camera and Instrument, in a deal arranged by a young financier named Arthur Rock. Many historians mark that deal as the start of modern venture capital. Eleven years later, two of the eight, Robert Noyce and Gordon Moore, founded Intel with Rock’s help.
In this first era a chip startup had to own everything. That was the only option. It designed the circuit, built the factory, ran the production line and sold the product. There was no foundry to rent and no tool vendor to call, so the venture model was simple. It was also very expensive. Investors backed a team, and the team built a complete manufacturer.
It worked. The customers were new and the competition was thin. The US defence and space programmes bought early integrated circuits at high prices and paid for the learning curve. Fairchild and Intel owned their designs, their process know-how and their customer relationships, so nothing of value sat with anyone else.
For our country the lesson of this era is not to copy it, because no one starts a chip company this way today. The lesson is quieter. The first winners had a committed buyer before they had scale, and military and space demand carried them through years of low yields. Our new industry needs a patient first customer as much as it needs patient capital.
We will see that pattern again in every era that follows.