Part 3: Sell to the Giant
The dot-com crash of 2000 ended an easy decade for chip investors. Venture money moved to software and the internet, where companies needed less capital and reached revenue faster. At the same time, the cost of designing a chip rose with every new process generation, and a single tape-out at the leading edge became a major financial event.
Nov 11, 2026
Part 3 · 2000–2009 · Sell to the Giant
The dot-com crash of 2000 ended an easy decade for chip investors. Venture money moved to software and the internet, where companies needed less capital and reached revenue faster. At the same time, the cost of designing a chip rose with every new process generation, and a single tape-out at the leading edge became a major financial event.
Exits changed in this decade as well. Public listings became rare and acquisitions became the normal outcome. When Broadcom bought SiByte in 2000 for about $2 billion, it was not paying for revenue. It was paying for a processor design and a team it needed.
We can learn as much from the failures of this decade. Several startups tried to replace mainstream processors with new computing architectures, and Ambric, MathStar, SiCortex and Stream Processors all closed in 2008 and 2009. Their technology was often impressive. Their problem was that they asked customers to rewrite software and change suppliers at the same time, and few customers were willing.
The pattern is clear to us in hindsight. The startups that did well built something an incumbent needed and could not build quickly enough. The startups that failed tried to replace the incumbent directly, and in a capital-intensive industry a head-on challenge is very hard to finance through a downturn.
This also explains why acquisitions dominated. A large company could absorb a design team in months, while building the same capability internally would take years. Acquirers paid for time, and for engineers who already understood the problem.
For Bangladesh the practical message is that our early companies should aim to be the best supplier in a narrow area rather than a challenger to the global leaders. Verification, analog and mixed-signal blocks, power management and test are all areas where a focused team of our engineers can become essential to larger companies.
There is a caution for us as well. When a local company is acquired, we may keep the engineers but lose the intellectual property. That trade can still be worth making, because an exit trains our founders and returns capital to our investors. But we should make it as a conscious choice, and not by accident.